<?xml version="1.0" encoding="UTF-8"?>
<?xml-stylesheet href="/stylesheet.css" type="text/css"?>
<?xml-stylesheet href="data:text/css,%3Aroot%7B--art%3Aurl%28%22https%3A%2F%2Fimg.transistorcdn.com%2FLeYk7XO_JtEDdDPEBMKrmYz07uyEIAoBFb8I3LUqk3E%2Frs%3Afill%3A0%3A0%3A1%2Fw%3A400%2Fh%3A400%2Fq%3A60%2Fmb%3A500000%2FaHR0cHM6Ly9pbWct%2FdXBsb2FkLXByb2R1%2FY3Rpb24udHJhbnNp%2Fc3Rvci5mbS9jZTA2%2FMDdjMzA1YjU4MjRm%2FOWI4ODIwNzFlMzQ4%2FYWQ0OS5wbmc.webp%22%29%3B--art-scale%3A1%7D" type="text/css"?>
<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:sy="http://purl.org/rss/1.0/modules/syndication/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd" xmlns:podcast="https://podcastindex.org/namespace/1.0">
  <channel>
    <atom:link rel="self" type="application/rss+xml" href="https://feeds.transistor.fm/azure-ai-fundamentals-certification-prepcast" title="MP3 Audio"/>
    <atom:link rel="hub" href="https://pubsubhubbub.appspot.com/"/>
    <podcast:podping usesPodping="true"/>
    <title>Azure AI Fundamentals Certification PrepCast</title>
    <generator>Transistor (https://transistor.fm)</generator>
    <itunes:new-feed-url>https://feeds.transistor.fm/azure-ai-fundamentals-certification-prepcast</itunes:new-feed-url>
    <description>Azure AI Fundamentals Certification PrepCast by BareMetalCyber.com is a sixty-episode audio course built to help learners prepare for the Microsoft Azure AI Fundamentals certification exam represented in the supplied course scripts as AI-901. The season begins with the current exam blueprint, responsible study methods, and Microsoft Responsible AI principles, then develops the conceptual distinctions needed to identify language, speech, vision, generative AI, agentic, and multimodal information-extraction workloads.

The second half of the course moves from recognition into foundational implementation with Microsoft Foundry. Episodes cover prompting, model selection and deployment, evaluation, lightweight Python clients, Foundry Agent Service, Azure Language, Azure Speech, multimodal vision, image generation, and Azure Content Understanding. Throughout the season, scenarios emphasize a repeatable decision process: define the requirement, match the capability, choose an implementation, and verify the result with observable evidence rather than relying on product names or one successful demonstration.</description>
    <copyright>2026 Baremetalcyber.com</copyright>
    <podcast:guid>73e168d7-c3e7-5135-82ab-e5022c8eaed3</podcast:guid>
    <podcast:podroll>
      <podcast:remoteItem feedGuid="9af25f2f-f465-5c56-8635-fc5e831ff06a" feedUrl="https://feeds.transistor.fm/bare-metal-cyber-a725a484-8216-4f80-9a32-2bfd5efcc240"/>
      <podcast:remoteItem feedGuid="ac645ca7-7469-50bf-9010-f13c165e3e14" feedUrl="https://feeds.transistor.fm/baremetalcyber-dot-one"/>
      <podcast:remoteItem feedGuid="143fc9c4-74e3-506c-8f6a-319fe2cb366d" feedUrl="https://feeds.transistor.fm/certified-the-cissp-prepcast"/>
      <podcast:remoteItem feedGuid="1e81ed4d-b3a7-5035-b12a-5171bdd497b8" feedUrl="https://feeds.transistor.fm/certified-the-crisc-prepcast"/>
      <podcast:remoteItem feedGuid="3620e13c-e0ca-5640-840a-2a3805dddeb7" feedUrl="https://feeds.transistor.fm/bmc-daily-cyber-news"/>
    </podcast:podroll>
    <podcast:locked>yes</podcast:locked>
    <language>en</language>
    <pubDate>Tue, 17 Nov 2026 12:11:03 -0600</pubDate>
    <lastBuildDate>Wed, 18 Nov 2026 23:10:30 -0600</lastBuildDate>
    <image>
      <url>https://img.transistorcdn.com/ippJvJc6kiqpwIYh2u_h14Misvmq9bSAOL5wjg_DVEg/rs:fill:0:0:1/w:1400/h:1400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS9jZTA2/MDdjMzA1YjU4MjRm/OWI4ODIwNzFlMzQ4/YWQ0OS5wbmc.jpg</url>
      <title>Azure AI Fundamentals Certification PrepCast</title>
    </image>
    <itunes:category text="Technology"/>
    <itunes:category text="Education">
      <itunes:category text="Courses"/>
    </itunes:category>
    <itunes:type>serial</itunes:type>
    <itunes:author>Dr Jason Edwards</itunes:author>
    <itunes:image href="https://img.transistorcdn.com/ippJvJc6kiqpwIYh2u_h14Misvmq9bSAOL5wjg_DVEg/rs:fill:0:0:1/w:1400/h:1400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS9jZTA2/MDdjMzA1YjU4MjRm/OWI4ODIwNzFlMzQ4/YWQ0OS5wbmc.jpg"/>
    <itunes:summary>Azure AI Fundamentals Certification PrepCast by BareMetalCyber.com is a sixty-episode audio course built to help learners prepare for the Microsoft Azure AI Fundamentals certification exam represented in the supplied course scripts as AI-901. The season begins with the current exam blueprint, responsible study methods, and Microsoft Responsible AI principles, then develops the conceptual distinctions needed to identify language, speech, vision, generative AI, agentic, and multimodal information-extraction workloads.

The second half of the course moves from recognition into foundational implementation with Microsoft Foundry. Episodes cover prompting, model selection and deployment, evaluation, lightweight Python clients, Foundry Agent Service, Azure Language, Azure Speech, multimodal vision, image generation, and Azure Content Understanding. Throughout the season, scenarios emphasize a repeatable decision process: define the requirement, match the capability, choose an implementation, and verify the result with observable evidence rather than relying on product names or one successful demonstration.</itunes:summary>
    <itunes:subtitle>Azure AI Fundamentals Certification PrepCast by BareMetalCyber.com is a sixty-episode audio course built to help learners prepare for the Microsoft Azure AI Fundamentals certification exam represented in the supplied course scripts as AI-901.</itunes:subtitle>
    <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
    <itunes:owner>
      <itunes:name>Dr Jason Edwards</itunes:name>
      <itunes:email>baremetalcyber@outlook.com</itunes:email>
    </itunes:owner>
    <itunes:complete>No</itunes:complete>
    <itunes:explicit>No</itunes:explicit>
    <item>
      <title>Understanding the AI-901 Exam and Current Skills Blueprint</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>1</itunes:episode>
      <podcast:episode>1</podcast:episode>
      <itunes:title>Understanding the AI-901 Exam and Current Skills Blueprint</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">13046314-e295-45b9-959c-09eea435be5e</guid>
      <link>https://share.transistor.fm/s/74ff8c64</link>
      <description>
        <![CDATA[Azure AI Fundamentals preparation begins by confirming that you are studying the current exam and the correct skills blueprint. This episode explains the transition from the retired AI-900 exam to AI-901, the importance of the dated skills-measured document, and the two broad areas that organize the current course: AI concepts and capabilities, and implementation with Microsoft Foundry. It also explains how exam weighting should influence study time without being treated as an exact question count. The goal is to establish the version, scope, and evidence that should control the rest of your preparation. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Azure AI Fundamentals preparation begins by confirming that you are studying the current exam and the correct skills blueprint. This episode explains the transition from the retired AI-900 exam to AI-901, the importance of the dated skills-measured document, and the two broad areas that organize the current course: AI concepts and capabilities, and implementation with Microsoft Foundry. It also explains how exam weighting should influence study time without being treated as an exact question count. The goal is to establish the version, scope, and evidence that should control the rest of your preparation. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:34:58 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/74ff8c64/ba8e7f22.mp3" length="16740630" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1041</itunes:duration>
      <itunes:summary>Azure AI Fundamentals preparation begins by confirming that you are studying the current exam and the correct skills blueprint. This episode explains the transition from the retired AI-900 exam to AI-901, the importance of the dated skills-measured document, and the two broad ar…</itunes:summary>
      <itunes:subtitle>Azure AI Fundamentals preparation begins by confirming that you are studying the current exam and the correct skills blueprint. This episode explains the transition from the retired AI-900 exam to AI-901, the importance of the dated skills-measured docume</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/74ff8c64/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>How to Use This AI-901 Course Effectively</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>2</itunes:episode>
      <podcast:episode>2</podcast:episode>
      <itunes:title>How to Use This AI-901 Course Effectively</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">8138fde0-fb3f-4dd8-b018-faa3441c5781</guid>
      <link>https://share.transistor.fm/s/e45806c1</link>
      <description>
        <![CDATA[Listening to a certification podcast can create familiarity without proving that you can make a technical decision. This episode explains how the sixty-episode season is organized, why the early foundation lessons matter, and how the course moves from AI concepts into Microsoft Foundry implementation. It introduces a study cycle built around understanding, active recall, scenario reasoning, and practical validation. Learners are encouraged to follow the sequence on the first pass, identify the input and output behind each scenario, and use checkpoints to revisit weak areas instead of treating completed audio as evidence of mastery. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Listening to a certification podcast can create familiarity without proving that you can make a technical decision. This episode explains how the sixty-episode season is organized, why the early foundation lessons matter, and how the course moves from AI concepts into Microsoft Foundry implementation. It introduces a study cycle built around understanding, active recall, scenario reasoning, and practical validation. Learners are encouraged to follow the sequence on the first pass, identify the input and output behind each scenario, and use checkpoints to revisit weak areas instead of treating completed audio as evidence of mastery. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:35:18 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/e45806c1/e1fcda9d.mp3" length="17018556" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1059</itunes:duration>
      <itunes:summary>Listening to a certification podcast can create familiarity without proving that you can make a technical decision. This episode explains how the sixty-episode season is organized, why the early foundation lessons matter, and how the course moves from AI concepts into Microsoft…</itunes:summary>
      <itunes:subtitle>Listening to a certification podcast can create familiarity without proving that you can make a technical decision. This episode explains how the sixty-episode season is organized, why the early foundation lessons matter, and how the course moves from AI </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/e45806c1/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Retrieval Practice, Spaced Review, and Scenario-Based Study</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>3</itunes:episode>
      <podcast:episode>3</podcast:episode>
      <itunes:title>Retrieval Practice, Spaced Review, and Scenario-Based Study</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">d66f7387-cdd4-4dda-b0ec-8cf4a3645ea3</guid>
      <link>https://share.transistor.fm/s/a7387e24</link>
      <description>
        <![CDATA[Familiar terminology is not the same as retrievable knowledge. This episode explains how active recall, delayed review, and scenario-based practice can expose gaps that repeated listening may hide. It shows how to reconstruct a lesson from memory, identify the clue that drives a technical choice, compare a correct option with a tempting alternative, and turn missed distinctions into questions for later review. The emphasis is on building durable decision-making ability so that concepts such as language, speech, vision, and Content Understanding can be distinguished when they appear in unfamiliar exam scenarios. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Familiar terminology is not the same as retrievable knowledge. This episode explains how active recall, delayed review, and scenario-based practice can expose gaps that repeated listening may hide. It shows how to reconstruct a lesson from memory, identify the clue that drives a technical choice, compare a correct option with a tempting alternative, and turn missed distinctions into questions for later review. The emphasis is on building durable decision-making ability so that concepts such as language, speech, vision, and Content Understanding can be distinguished when they appear in unfamiliar exam scenarios. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:35:34 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/a7387e24/b6387de5.mp3" length="18902734" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1176</itunes:duration>
      <itunes:summary>Familiar terminology is not the same as retrievable knowledge. This episode explains how active recall, delayed review, and scenario-based practice can expose gaps that repeated listening may hide. It shows how to reconstruct a lesson from memory, identify the clue that drives a…</itunes:summary>
      <itunes:subtitle>Familiar terminology is not the same as retrievable knowledge. This episode explains how active recall, delayed review, and scenario-based practice can expose gaps that repeated listening may hide. It shows how to reconstruct a lesson from memory, identif</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/a7387e24/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Exam Integrity, Practice Questions, and Test-Day Strategy</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>4</itunes:episode>
      <podcast:episode>4</podcast:episode>
      <itunes:title>Exam Integrity, Practice Questions, and Test-Day Strategy</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">18c1b774-81d8-493e-851f-0b657630881d</guid>
      <link>https://share.transistor.fm/s/34384854</link>
      <description>
        <![CDATA[AI-901 preparation must stay within Microsoft certification security and exam-integrity rules. This episode explains why recalled live questions, brain dumps, unauthorized assistance, and AI help during an exam are not legitimate study methods. It replaces those shortcuts with a practical approach based on published skills, authorized practice resources, careful reading, elimination of mismatched options, and reasoning from the requirement. The discussion also covers the published passing score without inventing an unpublished question count or scoring formula, helping learners prepare for the assessment through understanding rather than exposure to confidential content. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[AI-901 preparation must stay within Microsoft certification security and exam-integrity rules. This episode explains why recalled live questions, brain dumps, unauthorized assistance, and AI help during an exam are not legitimate study methods. It replaces those shortcuts with a practical approach based on published skills, authorized practice resources, careful reading, elimination of mismatched options, and reasoning from the requirement. The discussion also covers the published passing score without inventing an unpublished question count or scoring formula, helping learners prepare for the assessment through understanding rather than exposure to confidential content. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:35:51 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/34384854/393359c5.mp3" length="17669335" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1099</itunes:duration>
      <itunes:summary>AI-901 preparation must stay within Microsoft certification security and exam-integrity rules. This episode explains why recalled live questions, brain dumps, unauthorized assistance, and AI help during an exam are not legitimate study methods. It replaces those shortcuts with a…</itunes:summary>
      <itunes:subtitle>AI-901 preparation must stay within Microsoft certification security and exam-integrity rules. This episode explains why recalled live questions, brain dumps, unauthorized assistance, and AI help during an exam are not legitimate study methods. It replace</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/34384854/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Responsible AI: Principles and Lifecycle Thinking</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>5</itunes:episode>
      <podcast:episode>5</podcast:episode>
      <itunes:title>Responsible AI: Principles and Lifecycle Thinking</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">16819b96-72a9-4168-a2b1-63d69c7b58cc</guid>
      <link>https://share.transistor.fm/s/b73a9119</link>
      <description>
        <![CDATA[Responsible AI should shape an AI solution from the beginning rather than appear as a final review after a model has already been selected and deployed. This episode introduces Microsoft's six Responsible AI principles: fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability. It explains how those principles influence requirement definition, model and service selection, evaluation, deployment, and governance. Using a student-advising assistant as the scenario, the lesson shows why responsible behavior depends on the workload, users, data, limitations, and consequences of the output rather than on a model name or a single safety feature. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Responsible AI should shape an AI solution from the beginning rather than appear as a final review after a model has already been selected and deployed. This episode introduces Microsoft's six Responsible AI principles: fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability. It explains how those principles influence requirement definition, model and service selection, evaluation, deployment, and governance. Using a student-advising assistant as the scenario, the lesson shows why responsible behavior depends on the workload, users, data, limitations, and consequences of the output rather than on a model name or a single safety feature. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:36:08 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/b73a9119/721bb172.mp3" length="17933895" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1116</itunes:duration>
      <itunes:summary>Responsible AI should shape an AI solution from the beginning rather than appear as a final review after a model has already been selected and deployed. This episode introduces Microsoft's six Responsible AI principles: fairness, reliability and safety, privacy and security, inc…</itunes:summary>
      <itunes:subtitle>Responsible AI should shape an AI solution from the beginning rather than appear as a final review after a model has already been selected and deployed. This episode introduces Microsoft's six Responsible AI principles: fairness, reliability and safety, p</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/b73a9119/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Fairness: Evaluating Outcomes Across Groups</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>6</itunes:episode>
      <podcast:episode>6</podcast:episode>
      <itunes:title>Fairness: Evaluating Outcomes Across Groups</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">26a0f369-1083-42e8-ba68-259e4c967abe</guid>
      <link>https://share.transistor.fm/s/0de70421</link>
      <description>
        <![CDATA[A model can appear accurate overall while producing materially different error rates for different groups. This episode uses that problem to explain fairness as a Responsible AI principle and shows why aggregate performance can hide unequal outcomes. It examines representative evaluation, subgroup results, relevant error types, and the need to connect fairness evidence to the actual decision being supported. Learners also see why fairness is not interchangeable with privacy, reliability, or other Responsible AI concerns, and why a favorable overall metric does not by itself prove that an AI system treats affected groups appropriately. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A model can appear accurate overall while producing materially different error rates for different groups. This episode uses that problem to explain fairness as a Responsible AI principle and shows why aggregate performance can hide unequal outcomes. It examines representative evaluation, subgroup results, relevant error types, and the need to connect fairness evidence to the actual decision being supported. Learners also see why fairness is not interchangeable with privacy, reliability, or other Responsible AI concerns, and why a favorable overall metric does not by itself prove that an AI system treats affected groups appropriately. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:36:25 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/0de70421/83d37dc3.mp3" length="17717386" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1102</itunes:duration>
      <itunes:summary>A model can appear accurate overall while producing materially different error rates for different groups. This episode uses that problem to explain fairness as a Responsible AI principle and shows why aggregate performance can hide unequal outcomes. It examines representative e…</itunes:summary>
      <itunes:subtitle>A model can appear accurate overall while producing materially different error rates for different groups. This episode uses that problem to explain fairness as a Responsible AI principle and shows why aggregate performance can hide unequal outcomes. It e</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/0de70421/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Reliability and Safety: Testing AI Beyond the Happy Path</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>7</itunes:episode>
      <podcast:episode>7</podcast:episode>
      <itunes:title>Reliability and Safety: Testing AI Beyond the Happy Path</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">9da1bea2-8ef6-409b-afea-04dbd32a161b</guid>
      <link>https://share.transistor.fm/s/f5c54b71</link>
      <description>
        <![CDATA[An AI system that performs well on routine inputs can still fail in serious ways when conditions change. This episode explores reliability and safety by examining expected operating conditions, uncommon inputs, failure severity, and the safeguards needed when the system does not behave as intended. It emphasizes representative testing rather than judging a solution from a polished demonstration or average success rate. Learners practice asking what the workload requires, which failures matter most, how those failures will be detected, and what evidence is needed before the system can be considered dependable enough for its intended use. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[An AI system that performs well on routine inputs can still fail in serious ways when conditions change. This episode explores reliability and safety by examining expected operating conditions, uncommon inputs, failure severity, and the safeguards needed when the system does not behave as intended. It emphasizes representative testing rather than judging a solution from a polished demonstration or average success rate. Learners practice asking what the workload requires, which failures matter most, how those failures will be detected, and what evidence is needed before the system can be considered dependable enough for its intended use. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:36:42 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/f5c54b71/7de6a47b.mp3" length="17214594" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1071</itunes:duration>
      <itunes:summary>An AI system that performs well on routine inputs can still fail in serious ways when conditions change. This episode explores reliability and safety by examining expected operating conditions, uncommon inputs, failure severity, and the safeguards needed when the system does not…</itunes:summary>
      <itunes:subtitle>An AI system that performs well on routine inputs can still fail in serious ways when conditions change. This episode explores reliability and safety by examining expected operating conditions, uncommon inputs, failure severity, and the safeguards needed </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/f5c54b71/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Privacy and Security in AI Solutions</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>8</itunes:episode>
      <podcast:episode>8</podcast:episode>
      <itunes:title>Privacy and Security in AI Solutions</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">fb5e052e-4acf-4165-adac-c0ee4d254310</guid>
      <link>https://share.transistor.fm/s/988bb75d</link>
      <description>
        <![CDATA[A technically correct AI response can still represent an unacceptable solution when sensitive information reaches someone who should not receive it. This episode explains privacy and security within Microsoft's Responsible AI framework and shows how data access, authorization, protection, and intended use affect AI design. It distinguishes model performance from the security of the surrounding solution and reinforces that Responsible AI principles apply across the lifecycle. Learners use a policy assistant scenario to identify when correct output is not sufficient evidence because the wrong person, component, or workflow can still gain access to protected information. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A technically correct AI response can still represent an unacceptable solution when sensitive information reaches someone who should not receive it. This episode explains privacy and security within Microsoft's Responsible AI framework and shows how data access, authorization, protection, and intended use affect AI design. It distinguishes model performance from the security of the surrounding solution and reinforces that Responsible AI principles apply across the lifecycle. Learners use a policy assistant scenario to identify when correct output is not sufficient evidence because the wrong person, component, or workflow can still gain access to protected information. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:36:58 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/988bb75d/36e514fe.mp3" length="19066969" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1187</itunes:duration>
      <itunes:summary>A technically correct AI response can still represent an unacceptable solution when sensitive information reaches someone who should not receive it. This episode explains privacy and security within Microsoft's Responsible AI framework and shows how data access, authorization, p…</itunes:summary>
      <itunes:subtitle>A technically correct AI response can still represent an unacceptable solution when sensitive information reaches someone who should not receive it. This episode explains privacy and security within Microsoft's Responsible AI framework and shows how data </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/988bb75d/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Inclusiveness: Designing AI for Diverse Users</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>9</itunes:episode>
      <podcast:episode>9</podcast:episode>
      <itunes:title>Inclusiveness: Designing AI for Diverse Users</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">35830552-95b6-4874-a806-b070cdde7cb9</guid>
      <link>https://share.transistor.fm/s/bf092035</link>
      <description>
        <![CDATA[A voice assistant may work perfectly for a presenter in a quiet room and still exclude real users. This episode examines inclusiveness as a Responsible AI principle, focusing on varied abilities, speech patterns, environments, languages, and alternative ways to interact with a system. It shows why one ideal demonstration cannot establish that an application is broadly usable and why design choices should account for the people and conditions in the intended workload. Learners practice separating inclusiveness from neighboring concerns such as privacy, fairness, and reliability while evaluating whether an AI experience provides practical access for diverse users. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A voice assistant may work perfectly for a presenter in a quiet room and still exclude real users. This episode examines inclusiveness as a Responsible AI principle, focusing on varied abilities, speech patterns, environments, languages, and alternative ways to interact with a system. It shows why one ideal demonstration cannot establish that an application is broadly usable and why design choices should account for the people and conditions in the intended workload. Learners practice separating inclusiveness from neighboring concerns such as privacy, fairness, and reliability while evaluating whether an AI experience provides practical access for diverse users. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:37:15 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/bf092035/509d1ecb.mp3" length="17222106" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1071</itunes:duration>
      <itunes:summary>A voice assistant may work perfectly for a presenter in a quiet room and still exclude real users. This episode examines inclusiveness as a Responsible AI principle, focusing on varied abilities, speech patterns, environments, languages, and alternative ways to interact with a s…</itunes:summary>
      <itunes:subtitle>A voice assistant may work perfectly for a presenter in a quiet room and still exclude real users. This episode examines inclusiveness as a Responsible AI principle, focusing on varied abilities, speech patterns, environments, languages, and alternative w</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/bf092035/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Transparency: Communicating AI Limits and Uncertainty</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>10</itunes:episode>
      <podcast:episode>10</podcast:episode>
      <itunes:title>Transparency: Communicating AI Limits and Uncertainty</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">8470298f-c7e7-4fbd-87fe-21f31035fb8c</guid>
      <link>https://share.transistor.fm/s/c2bdaa6c</link>
      <description>
        <![CDATA[Fluent AI output can look authoritative even when important details are missing or uncertain. This episode explains transparency as the responsibility to provide people with useful information about how an AI system is being used, what its output means, and what limitations or uncertainty remain. It considers what end users, developers, and reviewers may need to know before relying on generated content. The lesson also distinguishes transparency from inclusiveness, privacy, and other Responsible AI principles, showing why clear disclosure, source awareness, and appropriate communication of limitations are part of trustworthy AI use. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Fluent AI output can look authoritative even when important details are missing or uncertain. This episode explains transparency as the responsibility to provide people with useful information about how an AI system is being used, what its output means, and what limitations or uncertainty remain. It considers what end users, developers, and reviewers may need to know before relying on generated content. The lesson also distinguishes transparency from inclusiveness, privacy, and other Responsible AI principles, showing why clear disclosure, source awareness, and appropriate communication of limitations are part of trustworthy AI use. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:37:32 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/c2bdaa6c/7f149649.mp3" length="18421658" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1146</itunes:duration>
      <itunes:summary>Fluent AI output can look authoritative even when important details are missing or uncertain. This episode explains transparency as the responsibility to provide people with useful information about how an AI system is being used, what its output means, and what limitations or u…</itunes:summary>
      <itunes:subtitle>Fluent AI output can look authoritative even when important details are missing or uncertain. This episode explains transparency as the responsibility to provide people with useful information about how an AI system is being used, what its output means, a</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/c2bdaa6c/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Accountability: Human Ownership and Oversight</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>11</itunes:episode>
      <podcast:episode>11</podcast:episode>
      <itunes:title>Accountability: Human Ownership and Oversight</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">5cf9811a-77e2-4b70-a3b9-1c17e71d98ad</guid>
      <link>https://share.transistor.fm/s/8d8450b5</link>
      <description>
        <![CDATA[When an AI recommendation contributes to a harmful outcome, the organization cannot transfer responsibility to the model. This episode focuses on accountability, including ownership of decisions, monitoring, escalation, review, and corrective action. It explains why people and organizations remain responsible for designing, deploying, and using AI systems even when a model produces the immediate output. Learners examine how clear roles, authority, and response processes support responsible operation and why statements such as “the model decided” do not resolve who must investigate, challenge, correct, or govern the system. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[When an AI recommendation contributes to a harmful outcome, the organization cannot transfer responsibility to the model. This episode focuses on accountability, including ownership of decisions, monitoring, escalation, review, and corrective action. It explains why people and organizations remain responsible for designing, deploying, and using AI systems even when a model produces the immediate output. Learners examine how clear roles, authority, and response processes support responsible operation and why statements such as “the model decided” do not resolve who must investigate, challenge, correct, or govern the system. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:37:49 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/8d8450b5/85a7425b.mp3" length="19307306" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1202</itunes:duration>
      <itunes:summary>When an AI recommendation contributes to a harmful outcome, the organization cannot transfer responsibility to the model. This episode focuses on accountability, including ownership of decisions, monitoring, escalation, review, and corrective action. It explains why people and o…</itunes:summary>
      <itunes:subtitle>When an AI recommendation contributes to a harmful outcome, the organization cannot transfer responsibility to the model. This episode focuses on accountability, including ownership of decisions, monitoring, escalation, review, and corrective action. It e</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/8d8450b5/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Generative AI Fundamentals: Tokens, Inference, and Output</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>12</itunes:episode>
      <podcast:episode>12</podcast:episode>
      <itunes:title>Generative AI Fundamentals: Tokens, Inference, and Output</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">e602c309-6519-4f55-9562-76b097a5d546</guid>
      <link>https://share.transistor.fm/s/ac0485e4</link>
      <description>
        <![CDATA[Generative AI does not retrieve a complete stored answer for every prompt. This episode introduces tokens and inference as the foundations for understanding how generative models process input and produce output. It explains why two similar requests can produce different but plausible responses, why fluent language does not prove factual accuracy, and how generated text emerges from learned patterns and current context. Learners build a practical mental model for recognizing generative AI workloads and evaluating their results without assuming that natural-sounding language is deterministic, human reasoning, or guaranteed truth. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Generative AI does not retrieve a complete stored answer for every prompt. This episode introduces tokens and inference as the foundations for understanding how generative models process input and produce output. It explains why two similar requests can produce different but plausible responses, why fluent language does not prove factual accuracy, and how generated text emerges from learned patterns and current context. Learners build a practical mental model for recognizing generative AI workloads and evaluating their results without assuming that natural-sounding language is deterministic, human reasoning, or guaranteed truth. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:38:06 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/ac0485e4/b71a85f1.mp3" length="17382616" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1081</itunes:duration>
      <itunes:summary>Generative AI does not retrieve a complete stored answer for every prompt. This episode introduces tokens and inference as the foundations for understanding how generative models process input and produce output. It explains why two similar requests can produce different but pla…</itunes:summary>
      <itunes:subtitle>Generative AI does not retrieve a complete stored answer for every prompt. This episode introduces tokens and inference as the foundations for understanding how generative models process input and produce output. It explains why two similar requests can p</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/ac0485e4/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Context Windows, Token Limits, and Long-Prompt Design</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>13</itunes:episode>
      <podcast:episode>13</podcast:episode>
      <itunes:title>Context Windows, Token Limits, and Long-Prompt Design</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">05c85ecd-339c-43b9-b893-df0cd83d7871</guid>
      <link>https://share.transistor.fm/s/6463d681</link>
      <description>
        <![CDATA[Long prompts can fail in ways that look like instruction-following problems even when the deeper issue is the amount of information being processed. This episode explains tokenization, context limits, accumulated conversation text, source material, and response capacity. Using a long meeting transcript as the scenario, it shows why important instructions and content can become difficult to handle when a request approaches practical model limits. Learners focus on the workload requirement, the information that must fit into context, and observable validation rather than assuming that a larger prompt automatically produces a better answer. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Long prompts can fail in ways that look like instruction-following problems even when the deeper issue is the amount of information being processed. This episode explains tokenization, context limits, accumulated conversation text, source material, and response capacity. Using a long meeting transcript as the scenario, it shows why important instructions and content can become difficult to handle when a request approaches practical model limits. Learners focus on the workload requirement, the information that must fit into context, and observable validation rather than assuming that a larger prompt automatically produces a better answer. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:38:22 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/6463d681/373236af.mp3" length="19485782" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1213</itunes:duration>
      <itunes:summary>Long prompts can fail in ways that look like instruction-following problems even when the deeper issue is the amount of information being processed. This episode explains tokenization, context limits, accumulated conversation text, source material, and response capacity. Using a…</itunes:summary>
      <itunes:subtitle>Long prompts can fail in ways that look like instruction-following problems even when the deeper issue is the amount of information being processed. This episode explains tokenization, context limits, accumulated conversation text, source material, and re</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/6463d681/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Selecting Models by Required Capability</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>14</itunes:episode>
      <podcast:episode>14</podcast:episode>
      <itunes:title>Selecting Models by Required Capability</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">f4278f82-fd9e-43af-8d8c-ac1f8e2c30e4</guid>
      <link>https://share.transistor.fm/s/471b7acf</link>
      <description>
        <![CDATA[A high benchmark score is useless if a model cannot accept the input or produce the output the application requires. This episode teaches model selection from the workload outward, beginning with required modalities, reasoning tasks, structured output, vision, code, and other needed capabilities. It uses an image-analysis scenario to show why a text-only model cannot satisfy a visual requirement regardless of its general quality score. Learners also distinguish capability from preference and practice evaluating whether a model can actually support the required input, processing behavior, and output contract. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A high benchmark score is useless if a model cannot accept the input or produce the output the application requires. This episode teaches model selection from the workload outward, beginning with required modalities, reasoning tasks, structured output, vision, code, and other needed capabilities. It uses an image-analysis scenario to show why a text-only model cannot satisfy a visual requirement regardless of its general quality score. Learners also distinguish capability from preference and practice evaluating whether a model can actually support the required input, processing behavior, and output contract. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:38:41 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/471b7acf/676475d6.mp3" length="18872204" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1174</itunes:duration>
      <itunes:summary>A high benchmark score is useless if a model cannot accept the input or produce the output the application requires. This episode teaches model selection from the workload outward, beginning with required modalities, reasoning tasks, structured output, vision, code, and other ne…</itunes:summary>
      <itunes:subtitle>A high benchmark score is useless if a model cannot accept the input or produce the output the application requires. This episode teaches model selection from the workload outward, beginning with required modalities, reasoning tasks, structured output, vi</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/471b7acf/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Comparing Models: Quality, Safety, Cost, and Throughput</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>15</itunes:episode>
      <podcast:episode>15</podcast:episode>
      <itunes:title>Comparing Models: Quality, Safety, Cost, and Throughput</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">5ba9459f-cc26-4eee-b46c-c87a5d1b2346</guid>
      <link>https://share.transistor.fm/s/ca45d2d6</link>
      <description>
        <![CDATA[There is no universally best model for every workload. This episode explains how model comparison can involve quality, safety, estimated cost, throughput, feature support, endpoints, and scenario-specific benchmarks. A customer-support example shows how a model with stronger quality results may still be a poor fit when the workload places greater weight on cost, volume, or another operational constraint. Learners practice defining which dimensions matter first, comparing evidence from representative tasks, and avoiding the mistake of treating one public benchmark or familiar provider name as a complete deployment decision. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[There is no universally best model for every workload. This episode explains how model comparison can involve quality, safety, estimated cost, throughput, feature support, endpoints, and scenario-specific benchmarks. A customer-support example shows how a model with stronger quality results may still be a poor fit when the workload places greater weight on cost, volume, or another operational constraint. Learners practice defining which dimensions matter first, comparing evidence from representative tasks, and avoiding the mistake of treating one public benchmark or familiar provider name as a complete deployment decision. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:38:58 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/ca45d2d6/a20fb7fe.mp3" length="19260086" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1199</itunes:duration>
      <itunes:summary>There is no universally best model for every workload. This episode explains how model comparison can involve quality, safety, estimated cost, throughput, feature support, endpoints, and scenario-specific benchmarks. A customer-support example shows how a model with stronger qua…</itunes:summary>
      <itunes:subtitle>There is no universally best model for every workload. This episode explains how model comparison can involve quality, safety, estimated cost, throughput, feature support, endpoints, and scenario-specific benchmarks. A customer-support example shows how a</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/ca45d2d6/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Deploying Models in Microsoft Foundry</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>16</itunes:episode>
      <podcast:episode>16</podcast:episode>
      <itunes:title>Deploying Models in Microsoft Foundry</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">e9363d8c-fcef-487d-ab21-a23aca9765f7</guid>
      <link>https://share.transistor.fm/s/8a45dc9d</link>
      <description>
        <![CDATA[Selecting a model and making it available to an application are separate decisions. This episode examines deployment choices in Microsoft Foundry, including the documented serverless API path, preview managed-compute options for selected models, and preview instant-access scenarios. It explains how infrastructure, networking, billing, model support, and operational requirements affect deployment decisions. Learners practice choosing a supported path from the workload requirements and verifying that the resulting deployment is actually usable rather than assuming that a model visible in the catalog is already ready for application inference. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Selecting a model and making it available to an application are separate decisions. This episode examines deployment choices in Microsoft Foundry, including the documented serverless API path, preview managed-compute options for selected models, and preview instant-access scenarios. It explains how infrastructure, networking, billing, model support, and operational requirements affect deployment decisions. Learners practice choosing a supported path from the workload requirements and verifying that the resulting deployment is actually usable rather than assuming that a model visible in the catalog is already ready for application inference. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:39:14 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/8a45dc9d/4c6b6f9a.mp3" length="19856078" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1236</itunes:duration>
      <itunes:summary>Selecting a model and making it available to an application are separate decisions. This episode examines deployment choices in Microsoft Foundry, including the documented serverless API path, preview managed-compute options for selected models, and preview instant-access scenar…</itunes:summary>
      <itunes:subtitle>Selecting a model and making it available to an application are separate decisions. This episode examines deployment choices in Microsoft Foundry, including the documented serverless API path, preview managed-compute options for selected models, and previ</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/8a45dc9d/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Inference Configuration, Prompting, and Model Behavior</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>17</itunes:episode>
      <podcast:episode>17</podcast:episode>
      <itunes:title>Inference Configuration, Prompting, and Model Behavior</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">788b6ab9-85a7-45f1-83f8-348cd229210a</guid>
      <link>https://share.transistor.fm/s/f569157c</link>
      <description>
        <![CDATA[When a deployed chat model produces inconsistent responses, the cause may lie in the prompt, the selected model, inference configuration, deployment path, or application behavior. This episode separates those layers so troubleshooting can follow evidence instead of guesswork. It explains how model capability, deployment, instructions, configuration, and observed output contribute different parts of the solution. Learners define acceptance criteria, test representative requests, and revise the part of the system that the evidence actually implicates rather than treating one excellent answer or one poor answer as proof about the entire design. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[When a deployed chat model produces inconsistent responses, the cause may lie in the prompt, the selected model, inference configuration, deployment path, or application behavior. This episode separates those layers so troubleshooting can follow evidence instead of guesswork. It explains how model capability, deployment, instructions, configuration, and observed output contribute different parts of the solution. Learners define acceptance criteria, test representative requests, and revise the part of the system that the evidence actually implicates rather than treating one excellent answer or one poor answer as proof about the entire design. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:39:31 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/f569157c/014a864e.mp3" length="16977192" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1056</itunes:duration>
      <itunes:summary>When a deployed chat model produces inconsistent responses, the cause may lie in the prompt, the selected model, inference configuration, deployment path, or application behavior. This episode separates those layers so troubleshooting can follow evidence instead of guesswork. It…</itunes:summary>
      <itunes:subtitle>When a deployed chat model produces inconsistent responses, the cause may lie in the prompt, the selected model, inference configuration, deployment path, or application behavior. This episode separates those layers so troubleshooting can follow evidence </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/f569157c/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Identifying AI Workloads by Input and Output</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>18</itunes:episode>
      <podcast:episode>18</podcast:episode>
      <itunes:title>Identifying AI Workloads by Input and Output</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">fcce76b7-c567-4455-bfaa-158328b51400</guid>
      <link>https://share.transistor.fm/s/21b7d5db</link>
      <description>
        <![CDATA[A single AI project may include sentiment classification, speech transcription, invoice extraction, and generated responses, but those requests do not describe one workload. This episode teaches a repeatable method for identifying AI workloads by examining what information enters the system, what transformation is required, and what output another person or application needs. It uses language, speech, structured extraction, and generative AI examples to show why product-first design creates mismatches. Learners practice classifying the required behavior before choosing a model, service, or implementation path. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A single AI project may include sentiment classification, speech transcription, invoice extraction, and generated responses, but those requests do not describe one workload. This episode teaches a repeatable method for identifying AI workloads by examining what information enters the system, what transformation is required, and what output another person or application needs. It uses language, speech, structured extraction, and generative AI examples to show why product-first design creates mismatches. Learners practice classifying the required behavior before choosing a model, service, or implementation path. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:39:47 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/21b7d5db/c99599d1.mp3" length="20718753" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1290</itunes:duration>
      <itunes:summary>A single AI project may include sentiment classification, speech transcription, invoice extraction, and generated responses, but those requests do not describe one workload. This episode teaches a repeatable method for identifying AI workloads by examining what information enter…</itunes:summary>
      <itunes:subtitle>A single AI project may include sentiment classification, speech transcription, invoice extraction, and generated responses, but those requests do not describe one workload. This episode teaches a repeatable method for identifying AI workloads by examinin</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/21b7d5db/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Generative AI Applications vs. AI Agents</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>19</itunes:episode>
      <podcast:episode>19</podcast:episode>
      <itunes:title>Generative AI Applications vs. AI Agents</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">0e566626-51fe-49d9-9195-22f851bfcd86</guid>
      <link>https://share.transistor.fm/s/14df80b5</link>
      <description>
        <![CDATA[Two applications may both use a large language model and present a chat interface while having very different capabilities. This episode distinguishes a generative application that produces a response from an input from an agent that can use instructions, tools, state, and authorized actions to pursue a defined task. It explains why the visible conversation alone does not determine whether a solution is agentic. Learners focus on what the system must actually do, whether it needs external tools or action paths, and how those requirements change the architecture. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Two applications may both use a large language model and present a chat interface while having very different capabilities. This episode distinguishes a generative application that produces a response from an input from an agent that can use instructions, tools, state, and authorized actions to pursue a defined task. It explains why the visible conversation alone does not determine whether a solution is agentic. Learners focus on what the system must actually do, whether it needs external tools or action paths, and how those requirements change the architecture. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:40:04 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/14df80b5/a5a848af.mp3" length="17603699" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1095</itunes:duration>
      <itunes:summary>Two applications may both use a large language model and present a chat interface while having very different capabilities. This episode distinguishes a generative application that produces a response from an input from an agent that can use instructions, tools, state, and autho…</itunes:summary>
      <itunes:subtitle>Two applications may both use a large language model and present a chat interface while having very different capabilities. This episode distinguishes a generative application that produces a response from an input from an agent that can use instructions,</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/14df80b5/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Natural Language Processing and Azure Language Capabilities</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>20</itunes:episode>
      <podcast:episode>20</podcast:episode>
      <itunes:title>Natural Language Processing and Azure Language Capabilities</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">982fe874-f98f-448b-8d56-2c2845e2dedb</guid>
      <link>https://share.transistor.fm/s/a386f2ab</link>
      <description>
        <![CDATA[Text workloads can require classifications, extracted phrases, recognized entities, summaries, or other structured results. This episode introduces natural language processing through Azure Language capabilities and shows why sending every text problem to a general-purpose generative model is not always the closest fit. Using customer comments as the scenario, it distinguishes repeatable text-analysis outputs from open-ended prose. Learners practice identifying the requested result first and then mapping that requirement to capabilities such as sentiment analysis, key phrase extraction, named entity recognition, personal information detection, and summarization. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Text workloads can require classifications, extracted phrases, recognized entities, summaries, or other structured results. This episode introduces natural language processing through Azure Language capabilities and shows why sending every text problem to a general-purpose generative model is not always the closest fit. Using customer comments as the scenario, it distinguishes repeatable text-analysis outputs from open-ended prose. Learners practice identifying the requested result first and then mapping that requirement to capabilities such as sentiment analysis, key phrase extraction, named entity recognition, personal information detection, and summarization. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:40:21 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/a386f2ab/5c7de604.mp3" length="18253227" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1136</itunes:duration>
      <itunes:summary>Text workloads can require classifications, extracted phrases, recognized entities, summaries, or other structured results. This episode introduces natural language processing through Azure Language capabilities and shows why sending every text problem to a general-purpose gener…</itunes:summary>
      <itunes:subtitle>Text workloads can require classifications, extracted phrases, recognized entities, summaries, or other structured results. This episode introduces natural language processing through Azure Language capabilities and shows why sending every text problem to</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/a386f2ab/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Key Phrase Extraction vs. Named Entity Recognition</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>21</itunes:episode>
      <podcast:episode>21</podcast:episode>
      <itunes:title>Key Phrase Extraction vs. Named Entity Recognition</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">cdb7f16b-4b17-4b7a-864e-1f0e163057c3</guid>
      <link>https://share.transistor.fm/s/6694a7ab</link>
      <description>
        <![CDATA[Key phrases and named entities can come from the same document, but they answer different questions. This episode compares extracting central topics with identifying recognized people, places, organizations, and other named things. A news-publishing scenario shows why one result cannot reliably substitute for the other even when some words overlap. Learners examine the output contract for each capability, distinguish these tasks from sentiment analysis and summarization, and practice selecting the language capability that matches what the downstream application actually needs. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Key phrases and named entities can come from the same document, but they answer different questions. This episode compares extracting central topics with identifying recognized people, places, organizations, and other named things. A news-publishing scenario shows why one result cannot reliably substitute for the other even when some words overlap. Learners examine the output contract for each capability, distinguish these tasks from sentiment analysis and summarization, and practice selecting the language capability that matches what the downstream application actually needs. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:40:38 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/6694a7ab/cb556747.mp3" length="17299435" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1076</itunes:duration>
      <itunes:summary>Key phrases and named entities can come from the same document, but they answer different questions. This episode compares extracting central topics with identifying recognized people, places, organizations, and other named things. A news-publishing scenario shows why one result…</itunes:summary>
      <itunes:subtitle>Key phrases and named entities can come from the same document, but they answer different questions. This episode compares extracting central topics with identifying recognized people, places, organizations, and other named things. A news-publishing scena</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/6694a7ab/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Sentiment Analysis vs. Summarization</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>22</itunes:episode>
      <podcast:episode>22</podcast:episode>
      <itunes:title>Sentiment Analysis vs. Summarization</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">0c5fffd2-db80-4f47-85e6-1be00f8140d5</guid>
      <link>https://share.transistor.fm/s/bfe89b80</link>
      <description>
        <![CDATA[A long customer-support message may need both an attitude signal and a concise account of what happened. This episode explains why sentiment analysis and summarization are different language tasks even when they operate on the same text. Sentiment produces a classification about expressed opinion, while summarization produces a shorter representation of the source material. Learners also compare these results with key phrases, recognized entities, and other text-analysis outputs so that a familiar language feature is not mistaken for the one the scenario actually requires. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A long customer-support message may need both an attitude signal and a concise account of what happened. This episode explains why sentiment analysis and summarization are different language tasks even when they operate on the same text. Sentiment produces a classification about expressed opinion, while summarization produces a shorter representation of the source material. Learners also compare these results with key phrases, recognized entities, and other text-analysis outputs so that a familiar language feature is not mistaken for the one the scenario actually requires. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:40:54 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/bfe89b80/9f636149.mp3" length="19562252" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1218</itunes:duration>
      <itunes:summary>A long customer-support message may need both an attitude signal and a concise account of what happened. This episode explains why sentiment analysis and summarization are different language tasks even when they operate on the same text. Sentiment produces a classification about…</itunes:summary>
      <itunes:subtitle>A long customer-support message may need both an attitude signal and a concise account of what happened. This episode explains why sentiment analysis and summarization are different language tasks even when they operate on the same text. Sentiment produce</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/bfe89b80/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Speech-to-Text and Text-to-Speech Fundamentals</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>23</itunes:episode>
      <podcast:episode>23</podcast:episode>
      <itunes:title>Speech-to-Text and Text-to-Speech Fundamentals</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">84ce539f-37d6-4005-adb0-ee8cb86eb588</guid>
      <link>https://share.transistor.fm/s/295d97cb</link>
      <description>
        <![CDATA[A voice interaction often uses two opposite speech operations. Incoming audio must be recognized and converted into text before an application can work with the request, while response text may later be synthesized into spoken audio. This episode distinguishes speech-to-text from text-to-speech and explains why a natural conversation can hide multiple technical stages. Learners practice following the direction of the data, identifying the input and expected output at each stage, and avoiding the mistake of treating all speech features as interchangeable. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A voice interaction often uses two opposite speech operations. Incoming audio must be recognized and converted into text before an application can work with the request, while response text may later be synthesized into spoken audio. This episode distinguishes speech-to-text from text-to-speech and explains why a natural conversation can hide multiple technical stages. Learners practice following the direction of the data, identifying the input and expected output at each stage, and avoiding the mistake of treating all speech features as interchangeable. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:41:11 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/295d97cb/c5d95bf2.mp3" length="18342239" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1141</itunes:duration>
      <itunes:summary>A voice interaction often uses two opposite speech operations. Incoming audio must be recognized and converted into text before an application can work with the request, while response text may later be synthesized into spoken audio. This episode distinguishes speech-to-text fro…</itunes:summary>
      <itunes:subtitle>A voice interaction often uses two opposite speech operations. Incoming audio must be recognized and converted into text before an application can work with the request, while response text may later be synthesized into spoken audio. This episode distingu</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/295d97cb/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Computer Vision: Interpreting Existing Images</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>24</itunes:episode>
      <podcast:episode>24</podcast:episode>
      <itunes:title>Computer Vision: Interpreting Existing Images</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">982f3495-1d90-447f-be6d-16d1f1b7da30</guid>
      <link>https://share.transistor.fm/s/e7b6cec4</link>
      <description>
        <![CDATA[When an application receives an existing photograph and needs information about what the image contains, the workload is different from creating a new image or extracting predefined fields from mixed content. This episode introduces visual interpretation and computer-vision reasoning through an industrial-equipment scenario. It emphasizes the direction of the task: an image is the input and information about that image is the output. Learners compare visual analysis with image generation and structured information extraction so that image-related product names do not obscure the actual requirement. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[When an application receives an existing photograph and needs information about what the image contains, the workload is different from creating a new image or extracting predefined fields from mixed content. This episode introduces visual interpretation and computer-vision reasoning through an industrial-equipment scenario. It emphasizes the direction of the task: an image is the input and information about that image is the output. Learners compare visual analysis with image generation and structured information extraction so that image-related product names do not obscure the actual requirement. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:41:27 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/e7b6cec4/d9a101ce.mp3" length="17679773" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1100</itunes:duration>
      <itunes:summary>When an application receives an existing photograph and needs information about what the image contains, the workload is different from creating a new image or extracting predefined fields from mixed content. This episode introduces visual interpretation and computer-vision reas…</itunes:summary>
      <itunes:subtitle>When an application receives an existing photograph and needs information about what the image contains, the workload is different from creating a new image or extracting predefined fields from mixed content. This episode introduces visual interpretation </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/e7b6cec4/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Image Generation: Creating New Visual Content</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>25</itunes:episode>
      <podcast:episode>25</podcast:episode>
      <itunes:title>Image Generation: Creating New Visual Content</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">3aac4b29-6413-4951-a48d-c0f5b7f52ea5</guid>
      <link>https://share.transistor.fm/s/6210cc78</link>
      <description>
        <![CDATA[A written creative brief that asks for an image that does not already exist describes a generative visual workload. This episode explains how text-to-image generation differs from analyzing an existing photograph or extracting structured information from visual content. It uses a design scenario to focus on the requested output and the evidence needed to determine whether the generated image satisfies the brief. Learners practice recognizing when new visual content, rather than visual interpretation, is the central requirement. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A written creative brief that asks for an image that does not already exist describes a generative visual workload. This episode explains how text-to-image generation differs from analyzing an existing photograph or extracting structured information from visual content. It uses a design scenario to focus on the requested output and the evidence needed to determine whether the generated image satisfies the brief. Learners practice recognizing when new visual content, rather than visual interpretation, is the central requirement. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:41:44 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/6210cc78/8f9a697f.mp3" length="19041066" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1185</itunes:duration>
      <itunes:summary>A written creative brief that asks for an image that does not already exist describes a generative visual workload. This episode explains how text-to-image generation differs from analyzing an existing photograph or extracting structured information from visual content. It uses…</itunes:summary>
      <itunes:subtitle>A written creative brief that asks for an image that does not already exist describes a generative visual workload. This episode explains how text-to-image generation differs from analyzing an existing photograph or extracting structured information from </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/6210cc78/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Content Understanding and Multimodal Information Extraction</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>26</itunes:episode>
      <podcast:episode>26</podcast:episode>
      <itunes:title>Content Understanding and Multimodal Information Extraction</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">dd665287-f238-4cc3-9645-c9227d0b007d</guid>
      <link>https://share.transistor.fm/s/bd177a80</link>
      <description>
        <![CDATA[Documents, photographs, audio recordings, and video can all contain information that a business needs in consistent fields. This episode introduces Azure Content Understanding as a multimodal extraction capability for turning unstructured content into structured outputs that applications can consume. Using an insurance-claim package, it explains why storing several files together is not the same as extracting useful case information from them. Learners focus on analyzers, expected fields, modality-specific processing, and the distinction between free-form description and predictable structured results. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Documents, photographs, audio recordings, and video can all contain information that a business needs in consistent fields. This episode introduces Azure Content Understanding as a multimodal extraction capability for turning unstructured content into structured outputs that applications can consume. Using an insurance-claim package, it explains why storing several files together is not the same as extracting useful case information from them. Learners focus on analyzers, expected fields, modality-specific processing, and the distinction between free-form description and predictable structured results. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:42:00 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/bd177a80/571a21a1.mp3" length="19079532" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1187</itunes:duration>
      <itunes:summary>Documents, photographs, audio recordings, and video can all contain information that a business needs in consistent fields. This episode introduces Azure Content Understanding as a multimodal extraction capability for turning unstructured content into structured outputs that app…</itunes:summary>
      <itunes:subtitle>Documents, photographs, audio recordings, and video can all contain information that a business needs in consistent fields. This episode introduces Azure Content Understanding as a multimodal extraction capability for turning unstructured content into str</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/bd177a80/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Choosing the Right AI Capability Across Workloads</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>27</itunes:episode>
      <podcast:episode>27</podcast:episode>
      <itunes:title>Choosing the Right AI Capability Across Workloads</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">1a729c8a-24cc-47f4-a2ff-a0a304dc7b19</guid>
      <link>https://share.transistor.fm/s/72394922</link>
      <description>
        <![CDATA[A project can require text summarization, speech transcription, visual analysis, image generation, and structured document extraction at the same time. This episode brings those workload categories together and teaches a comparison method based on input, processing behavior, output, and evidence. Learners practice distinguishing closely related capabilities and rejecting plausible alternatives that solve a different problem. The objective is to make workload selection a reasoned decision rather than a memorization exercise built around product names. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A project can require text summarization, speech transcription, visual analysis, image generation, and structured document extraction at the same time. This episode brings those workload categories together and teaches a comparison method based on input, processing behavior, output, and evidence. Learners practice distinguishing closely related capabilities and rejecting plausible alternatives that solve a different problem. The objective is to make workload selection a reasoned decision rather than a memorization exercise built around product names. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:42:17 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/72394922/26625666.mp3" length="20040828" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1247</itunes:duration>
      <itunes:summary>A project can require text summarization, speech transcription, visual analysis, image generation, and structured document extraction at the same time. This episode brings those workload categories together and teaches a comparison method based on input, processing behavior, out…</itunes:summary>
      <itunes:subtitle>A project can require text summarization, speech transcription, visual analysis, image generation, and structured document extraction at the same time. This episode brings those workload categories together and teaches a comparison method based on input, </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/72394922/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Integrated AI Solution Design and Responsible AI Decisions</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>28</itunes:episode>
      <podcast:episode>28</podcast:episode>
      <itunes:title>Integrated AI Solution Design and Responsible AI Decisions</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">e1823f77-d533-4221-8200-dfe34514a5bb</guid>
      <link>https://share.transistor.fm/s/b8f937f8</link>
      <description>
        <![CDATA[Real solutions often combine several AI workloads and several governance concerns. This episode uses a sensitive multimodal healthcare assistant to connect speech recognition, vision, information extraction, language analysis, generative AI, model selection, deployment, and Responsible AI. It shows how separate stages can be defined and validated so that failures can be traced to the transcript, extraction, generation, access, or another part of the workflow. Learners apply the requirement-capability-implementation-evidence decision chain to a complete solution while considering fairness, safety, privacy, inclusiveness, transparency, and accountability. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Real solutions often combine several AI workloads and several governance concerns. This episode uses a sensitive multimodal healthcare assistant to connect speech recognition, vision, information extraction, language analysis, generative AI, model selection, deployment, and Responsible AI. It shows how separate stages can be defined and validated so that failures can be traced to the transcript, extraction, generation, access, or another part of the workflow. Learners apply the requirement-capability-implementation-evidence decision chain to a complete solution while considering fairness, safety, privacy, inclusiveness, transparency, and accountability. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:42:34 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/b8f937f8/53596a3e.mp3" length="19097085" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1188</itunes:duration>
      <itunes:summary>Real solutions often combine several AI workloads and several governance concerns. This episode uses a sensitive multimodal healthcare assistant to connect speech recognition, vision, information extraction, language analysis, generative AI, model selection, deployment, and Resp…</itunes:summary>
      <itunes:subtitle>Real solutions often combine several AI workloads and several governance concerns. This episode uses a sensitive multimodal healthcare assistant to connect speech recognition, vision, information extraction, language analysis, generative AI, model selecti</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/b8f937f8/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>System Prompts vs. User Prompts</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>29</itunes:episode>
      <podcast:episode>29</podcast:episode>
      <itunes:title>System Prompts vs. User Prompts</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">a5c0292a-db88-4bac-9ac0-810cf85566e6</guid>
      <link>https://share.transistor.fm/s/445fd6b3</link>
      <description>
        <![CDATA[Stable application behavior and changing user requests belong in different parts of a prompt design. This episode explains how system prompts provide durable instructions about role, tone, boundaries, format, and general expectations, while user prompts carry the task-specific request for the current interaction. It also connects prompt structure to tokens and context limits without claiming that careful prompting guarantees correctness. Learners use a customer-support assistant to decide where instructions belong and how to define an input and output contract that can later be tested. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Stable application behavior and changing user requests belong in different parts of a prompt design. This episode explains how system prompts provide durable instructions about role, tone, boundaries, format, and general expectations, while user prompts carry the task-specific request for the current interaction. It also connects prompt structure to tokens and context limits without claiming that careful prompting guarantees correctness. Learners use a customer-support assistant to decide where instructions belong and how to define an input and output contract that can later be tested. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:42:50 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/445fd6b3/63a85206.mp3" length="17346227" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1079</itunes:duration>
      <itunes:summary>Stable application behavior and changing user requests belong in different parts of a prompt design. This episode explains how system prompts provide durable instructions about role, tone, boundaries, format, and general expectations, while user prompts carry the task-specific r…</itunes:summary>
      <itunes:subtitle>Stable application behavior and changing user requests belong in different parts of a prompt design. This episode explains how system prompts provide durable instructions about role, tone, boundaries, format, and general expectations, while user prompts c</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/445fd6b3/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Writing Effective Prompts and Resolving Conflicting Instructions</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>30</itunes:episode>
      <podcast:episode>30</podcast:episode>
      <itunes:title>Writing Effective Prompts and Resolving Conflicting Instructions</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">41489a43-77a1-4498-8043-0bd7d3cdd3b9</guid>
      <link>https://share.transistor.fm/s/aeb7cdcd</link>
      <description>
        <![CDATA[A model cannot reliably satisfy two instructions that directly contradict each other. This episode uses a support assistant with conflicting system rules to explain prompt engineering, scope, constraints, expected output, and fallback behavior. It shows why longer wording is not automatically better and why contradictions should be removed rather than hidden inside additional instructions. Learners practice writing prompts that make acceptable and unacceptable behavior distinguishable, then validating whether the model follows the intended boundary across representative requests. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A model cannot reliably satisfy two instructions that directly contradict each other. This episode uses a support assistant with conflicting system rules to explain prompt engineering, scope, constraints, expected output, and fallback behavior. It shows why longer wording is not automatically better and why contradictions should be removed rather than hidden inside additional instructions. Learners practice writing prompts that make acceptable and unacceptable behavior distinguishable, then validating whether the model follows the intended boundary across representative requests. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:43:08 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/aeb7cdcd/a9335904.mp3" length="17912595" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1114</itunes:duration>
      <itunes:summary>A model cannot reliably satisfy two instructions that directly contradict each other. This episode uses a support assistant with conflicting system rules to explain prompt engineering, scope, constraints, expected output, and fallback behavior. It shows why longer wording is not…</itunes:summary>
      <itunes:subtitle>A model cannot reliably satisfy two instructions that directly contradict each other. This episode uses a support assistant with conflicting system rules to explain prompt engineering, scope, constraints, expected output, and fallback behavior. It shows w</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/aeb7cdcd/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Deploying a Model from the Foundry Catalog</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>31</itunes:episode>
      <podcast:episode>31</podcast:episode>
      <itunes:title>Deploying a Model from the Foundry Catalog</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">930629c8-ec56-4c89-b3dd-2e0db862dd9d</guid>
      <link>https://share.transistor.fm/s/58510d47</link>
      <description>
        <![CDATA[Finding a suitable model in the Microsoft Foundry catalog does not make that model callable by an application. This episode explains the difference between model selection and deployment and shows how deployment, identity, networking, role-based access, monitoring, tracing, and evaluation contribute different parts of an operational solution. Learners start with the required input, output, and constraints, choose a supported deployment path, and define observable evidence that proves an authorized client can invoke the intended deployment. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Finding a suitable model in the Microsoft Foundry catalog does not make that model callable by an application. This episode explains the difference between model selection and deployment and shows how deployment, identity, networking, role-based access, monitoring, tracing, and evaluation contribute different parts of an operational solution. Learners start with the required input, output, and constraints, choose a supported deployment path, and define observable evidence that proves an authorized client can invoke the intended deployment. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:43:25 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/58510d47/fc68b9b9.mp3" length="17723655" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1103</itunes:duration>
      <itunes:summary>Finding a suitable model in the Microsoft Foundry catalog does not make that model callable by an application. This episode explains the difference between model selection and deployment and shows how deployment, identity, networking, role-based access, monitoring, tracing, and…</itunes:summary>
      <itunes:subtitle>Finding a suitable model in the Microsoft Foundry catalog does not make that model callable by an application. This episode explains the difference between model selection and deployment and shows how deployment, identity, networking, role-based access, m</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/58510d47/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Testing and Evaluating a Deployed Model</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>32</itunes:episode>
      <podcast:episode>32</podcast:episode>
      <itunes:title>Testing and Evaluating a Deployed Model</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">ab8a641b-abc4-4f99-ab3e-5c630def41a5</guid>
      <link>https://share.transistor.fm/s/0e002c0f</link>
      <description>
        <![CDATA[A deployment can be technically available and still fail the workload it was created to support. This episode explains how to test a deployed model in Microsoft Foundry by supplying representative inputs, observing outputs, and comparing the results with explicit acceptance criteria. It separates deployment success from application suitability and shows how omitted details, excessive length, unsupported claims, or other failures should drive targeted revisions. Learners use evaluation as an evidence loop rather than judging a model from one polished response. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A deployment can be technically available and still fail the workload it was created to support. This episode explains how to test a deployed model in Microsoft Foundry by supplying representative inputs, observing outputs, and comparing the results with explicit acceptance criteria. It separates deployment success from application suitability and shows how omitted details, excessive length, unsupported claims, or other failures should drive targeted revisions. Learners use evaluation as an evidence loop rather than judging a model from one polished response. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:43:42 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/0e002c0f/cfcd6cd4.mp3" length="17417288" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1083</itunes:duration>
      <itunes:summary>A deployment can be technically available and still fail the workload it was created to support. This episode explains how to test a deployed model in Microsoft Foundry by supplying representative inputs, observing outputs, and comparing the results with explicit acceptance crit…</itunes:summary>
      <itunes:subtitle>A deployment can be technically available and still fail the workload it was created to support. This episode explains how to test a deployed model in Microsoft Foundry by supplying representative inputs, observing outputs, and comparing the results with </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/0e002c0f/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Connecting Python Applications to Microsoft Foundry</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>33</itunes:episode>
      <podcast:episode>33</podcast:episode>
      <itunes:title>Connecting Python Applications to Microsoft Foundry</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">d34947a6-8dd2-4769-9844-7921eb183f97</guid>
      <link>https://share.transistor.fm/s/fb807a9f</link>
      <description>
        <![CDATA[A model that works in the Foundry portal can still fail from Python when the application uses the wrong endpoint, client, project path, or authentication. This episode explains the connection model for a lightweight Foundry application and separates destination from identity. It also distinguishes the project client from the OpenAI-compatible client used for supported model interactions. Learners diagnose application connectivity by checking the intended project, endpoint, client role, and authorization instead of assuming that a healthy deployment guarantees that every external application configuration will work. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A model that works in the Foundry portal can still fail from Python when the application uses the wrong endpoint, client, project path, or authentication. This episode explains the connection model for a lightweight Foundry application and separates destination from identity. It also distinguishes the project client from the OpenAI-compatible client used for supported model interactions. Learners diagnose application connectivity by checking the intended project, endpoint, client role, and authorization instead of assuming that a healthy deployment guarantees that every external application configuration will work. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:43:58 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/fb807a9f/3a113980.mp3" length="18300030" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1139</itunes:duration>
      <itunes:summary>A model that works in the Foundry portal can still fail from Python when the application uses the wrong endpoint, client, project path, or authentication. This episode explains the connection model for a lightweight Foundry application and separates destination from identity. It…</itunes:summary>
      <itunes:subtitle>A model that works in the Foundry portal can still fail from Python when the application uses the wrong endpoint, client, project path, or authentication. This episode explains the connection model for a lightweight Foundry application and separates desti</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/fb807a9f/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Building a Lightweight Python Model Client</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>34</itunes:episode>
      <podcast:episode>34</podcast:episode>
      <itunes:title>Building a Lightweight Python Model Client</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">441850c3-454e-4ac6-82a9-961bdd615b28</guid>
      <link>https://share.transistor.fm/s/bdd0b356</link>
      <description>
        <![CDATA[A simple application that sends one question to a deployed model does not require an agent or persistent conversation state. This episode walks through the conceptual path for a lightweight Python client: connect to the Foundry project, authenticate appropriately, obtain the client used for model interaction, submit input to the intended deployment, receive a response, and extract the output the application needs. The focus is on understanding each component and validating the result rather than memorizing one code sample or adding unnecessary architecture. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A simple application that sends one question to a deployed model does not require an agent or persistent conversation state. This episode walks through the conceptual path for a lightweight Python client: connect to the Foundry project, authenticate appropriately, obtain the client used for model interaction, submit input to the intended deployment, receive a response, and extract the output the application needs. The focus is on understanding each component and validating the result rather than memorizing one code sample or adding unnecessary architecture. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:44:15 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/bdd0b356/cfc6c19c.mp3" length="18200547" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1132</itunes:duration>
      <itunes:summary>A simple application that sends one question to a deployed model does not require an agent or persistent conversation state. This episode walks through the conceptual path for a lightweight Python client: connect to the Foundry project, authenticate appropriately, obtain the cli…</itunes:summary>
      <itunes:subtitle>A simple application that sends one question to a deployed model does not require an agent or persistent conversation state. This episode walks through the conceptual path for a lightweight Python client: connect to the Foundry project, authenticate appro</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/bdd0b356/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Managing Multi-Turn Chat and Conversation State</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>35</itunes:episode>
      <podcast:episode>35</podcast:episode>
      <itunes:title>Managing Multi-Turn Chat and Conversation State</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">9f76df01-3b46-4f8c-8a65-de7abe5fd442</guid>
      <link>https://share.transistor.fm/s/9ada60b4</link>
      <description>
        <![CDATA[A chat interface can look continuous while the underlying requests remain unrelated. This episode explains how multi-turn conversation state preserves the context needed for follow-up questions such as “What would be the simplest version of that?” It distinguishes a single model call from an explicitly managed conversation and shows why on-screen message history does not prove that earlier turns were sent to the model. Learners practice defining what context must persist, reusing the appropriate conversation state, and validating continuity through dependent follow-up requests. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A chat interface can look continuous while the underlying requests remain unrelated. This episode explains how multi-turn conversation state preserves the context needed for follow-up questions such as “What would be the simplest version of that?” It distinguishes a single model call from an explicitly managed conversation and shows why on-screen message history does not prove that earlier turns were sent to the model. Learners practice defining what context must persist, reusing the appropriate conversation state, and validating continuity through dependent follow-up requests. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:44:32 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/9ada60b4/db303809.mp3" length="18785695" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1169</itunes:duration>
      <itunes:summary>A chat interface can look continuous while the underlying requests remain unrelated. This episode explains how multi-turn conversation state preserves the context needed for follow-up questions such as “What would be the simplest version of that?” It distinguishes a single model…</itunes:summary>
      <itunes:subtitle>A chat interface can look continuous while the underlying requests remain unrelated. This episode explains how multi-turn conversation state preserves the context needed for follow-up questions such as “What would be the simplest version of that?” It dist</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/9ada60b4/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Designing a Single-Agent Solution with Foundry Agent Service</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>36</itunes:episode>
      <podcast:episode>36</podcast:episode>
      <itunes:title>Designing a Single-Agent Solution with Foundry Agent Service</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">ad55ebb9-9817-4585-b0ed-ad87b429842f</guid>
      <link>https://share.transistor.fm/s/e37b3e42</link>
      <description>
        <![CDATA[An AI agent combines a model with instructions and, when needed, tools around a defined job. This episode introduces Microsoft Foundry Agent Service as a managed environment for building, deploying, and scaling agents while preserving the need for design decisions about role, behavior, tools, safety, and validation. It distinguishes prompt agents from hosted agents and explains when custom runtime control may justify the latter. Learners focus on one single-agent scenario and select the least complex implementation that satisfies the actual requirement. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[An AI agent combines a model with instructions and, when needed, tools around a defined job. This episode introduces Microsoft Foundry Agent Service as a managed environment for building, deploying, and scaling agents while preserving the need for design decisions about role, behavior, tools, safety, and validation. It distinguishes prompt agents from hosted agents and explains when custom runtime control may justify the latter. Learners focus on one single-agent scenario and select the least complex implementation that satisfies the actual requirement. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:44:48 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/e37b3e42/6c6bf573.mp3" length="17075419" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1062</itunes:duration>
      <itunes:summary>An AI agent combines a model with instructions and, when needed, tools around a defined job. This episode introduces Microsoft Foundry Agent Service as a managed environment for building, deploying, and scaling agents while preserving the need for design decisions about role, be…</itunes:summary>
      <itunes:subtitle>An AI agent combines a model with instructions and, when needed, tools around a defined job. This episode introduces Microsoft Foundry Agent Service as a managed environment for building, deploying, and scaling agents while preserving the need for design </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/e37b3e42/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Configuring and Testing a Prompt Agent</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>37</itunes:episode>
      <podcast:episode>37</podcast:episode>
      <itunes:title>Configuring and Testing a Prompt Agent</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">abb094f8-58f9-4dd5-a601-50fba86158af</guid>
      <link>https://share.transistor.fm/s/aa48169b</link>
      <description>
        <![CDATA[Saving an agent configuration proves that the platform accepted the settings, not that the agent behaves correctly. This episode examines the three central parts of a prompt agent: instructions, model, and tools. It shows how vague instructions can create inconsistent answers and how representative testing can reveal missing boundaries, unsupported detail, and formatting problems. Learners define the agent's role, expected output, constraints, and fallback behavior, then compare observed responses with those expectations before treating the agent as ready. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Saving an agent configuration proves that the platform accepted the settings, not that the agent behaves correctly. This episode examines the three central parts of a prompt agent: instructions, model, and tools. It shows how vague instructions can create inconsistent answers and how representative testing can reveal missing boundaries, unsupported detail, and formatting problems. Learners define the agent's role, expected output, constraints, and fallback behavior, then compare observed responses with those expectations before treating the agent as ready. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:45:05 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/aa48169b/0790c71b.mp3" length="17736190" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1103</itunes:duration>
      <itunes:summary>Saving an agent configuration proves that the platform accepted the settings, not that the agent behaves correctly. This episode examines the three central parts of a prompt agent: instructions, model, and tools. It shows how vague instructions can create inconsistent answers an…</itunes:summary>
      <itunes:subtitle>Saving an agent configuration proves that the platform accepted the settings, not that the agent behaves correctly. This episode examines the three central parts of a prompt agent: instructions, model, and tools. It shows how vague instructions can create</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/aa48169b/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Selecting and Restricting Agent Tools</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>38</itunes:episode>
      <podcast:episode>38</podcast:episode>
      <itunes:title>Selecting and Restricting Agent Tools</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">a40e4587-d1e7-4ede-b6cb-5cc01b96bcbd</guid>
      <link>https://share.transistor.fm/s/fc6e292c</link>
      <description>
        <![CDATA[Giving an agent every available tool can create unnecessary access paths and make behavior harder to control. This episode explains how tools extend an agent beyond the model itself and why tool selection should follow the job the agent is assigned. Using an internal-policy assistant, it compares controlled retrieval with unnecessary public web search or code execution. Learners practice least-capability design by choosing only the tools required for the task, defining instructions and boundaries around those tools, and testing whether the agent remains grounded in the approved source. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Giving an agent every available tool can create unnecessary access paths and make behavior harder to control. This episode explains how tools extend an agent beyond the model itself and why tool selection should follow the job the agent is assigned. Using an internal-policy assistant, it compares controlled retrieval with unnecessary public web search or code execution. Learners practice least-capability design by choosing only the tools required for the task, defining instructions and boundaries around those tools, and testing whether the agent remains grounded in the approved source. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:45:22 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/fc6e292c/e5819ee0.mp3" length="17354592" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1080</itunes:duration>
      <itunes:summary>Giving an agent every available tool can create unnecessary access paths and make behavior harder to control. This episode explains how tools extend an agent beyond the model itself and why tool selection should follow the job the agent is assigned. Using an internal-policy assi…</itunes:summary>
      <itunes:subtitle>Giving an agent every available tool can create unnecessary access paths and make behavior harder to control. This episode explains how tools extend an agent beyond the model itself and why tool selection should follow the job the agent is assigned. Using</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/fc6e292c/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Building a Python Client for an Existing Agent</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>39</itunes:episode>
      <podcast:episode>39</podcast:episode>
      <itunes:title>Building a Python Client for an Existing Agent</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">f99bff93-e927-438b-bc5c-d0a3bcd3618e</guid>
      <link>https://share.transistor.fm/s/42283293</link>
      <description>
        <![CDATA[A Python application that invokes an existing agent has a different job from the code or configuration that defines the agent itself. This episode explains how a client identifies the intended Foundry project and agent, submits a user request, receives the response, and verifies that the interaction came through the expected agent rather than a direct model call. It distinguishes client responsibilities from agent design and emphasizes authentication, agent references, version awareness, and observable evidence that the correct capability handled the request. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A Python application that invokes an existing agent has a different job from the code or configuration that defines the agent itself. This episode explains how a client identifies the intended Foundry project and agent, submits a user request, receives the response, and verifies that the interaction came through the expected agent rather than a direct model call. It distinguishes client responsibilities from agent design and emphasizes authentication, agent references, version awareness, and observable evidence that the correct capability handled the request. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:45:39 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/42283293/4974f233.mp3" length="17425654" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1084</itunes:duration>
      <itunes:summary>A Python application that invokes an existing agent has a different job from the code or configuration that defines the agent itself. This episode explains how a client identifies the intended Foundry project and agent, submits a user request, receives the response, and verifies…</itunes:summary>
      <itunes:subtitle>A Python application that invokes an existing agent has a different job from the code or configuration that defines the agent itself. This episode explains how a client identifies the intended Foundry project and agent, submits a user request, receives th</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/42283293/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Managing Agent Conversation State Across Turns</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>40</itunes:episode>
      <podcast:episode>40</podcast:episode>
      <itunes:title>Managing Agent Conversation State Across Turns</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">fba4ab5f-d545-4a30-8f38-aecbe754436e</guid>
      <link>https://share.transistor.fm/s/9a83e5e8</link>
      <description>
        <![CDATA[Multi-turn agent interactions require more than sending several messages to the same model. This episode explains how conversation state allows a later request to depend on information established earlier, such as a selected product. It distinguishes per-conversation context from general agent instructions and shows why customer-specific details should not be embedded as permanent behavior rules. Learners compare reused and new conversations, validate dependent follow-up questions, and troubleshoot continuity by checking the actual conversation reference rather than relying on variable names or a chat-like interface. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Multi-turn agent interactions require more than sending several messages to the same model. This episode explains how conversation state allows a later request to depend on information established earlier, such as a selected product. It distinguishes per-conversation context from general agent instructions and shows why customer-specific details should not be embedded as permanent behavior rules. Learners compare reused and new conversations, validate dependent follow-up questions, and troubleshoot continuity by checking the actual conversation reference rather than relying on variable names or a chat-like interface. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:45:56 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/9a83e5e8/ae771306.mp3" length="17450314" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1086</itunes:duration>
      <itunes:summary>Multi-turn agent interactions require more than sending several messages to the same model. This episode explains how conversation state allows a later request to depend on information established earlier, such as a selected product. It distinguishes per-conversation context fro…</itunes:summary>
      <itunes:subtitle>Multi-turn agent interactions require more than sending several messages to the same model. This episode explains how conversation state allows a later request to depend on information established earlier, such as a selected product. It distinguishes per-</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/9a83e5e8/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Building a Lightweight Text-Analysis Application</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>41</itunes:episode>
      <podcast:episode>41</podcast:episode>
      <itunes:title>Building a Lightweight Text-Analysis Application</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">a3967291-2981-46e7-990e-e3bb28a8b4dc</guid>
      <link>https://share.transistor.fm/s/f5f7bfea</link>
      <description>
        <![CDATA[A text-analysis application should return the structured result the business needs rather than another paragraph that someone must interpret. This episode introduces a lightweight implementation pattern for Azure Language capabilities such as sentiment analysis, key phrase extraction, named entity recognition, personal information detection, and summarization. Learners begin with the requested field or classification, choose the matching capability, submit the text, and validate the returned result. The focus is on matching output type to business use and proving that the application received the expected structured evidence. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A text-analysis application should return the structured result the business needs rather than another paragraph that someone must interpret. This episode introduces a lightweight implementation pattern for Azure Language capabilities such as sentiment analysis, key phrase extraction, named entity recognition, personal information detection, and summarization. Learners begin with the requested field or classification, choose the matching capability, submit the text, and validate the returned result. The focus is on matching output type to business use and proving that the application received the expected structured evidence. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:46:13 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/f5f7bfea/f118d25d.mp3" length="18581314" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1156</itunes:duration>
      <itunes:summary>A text-analysis application should return the structured result the business needs rather than another paragraph that someone must interpret. This episode introduces a lightweight implementation pattern for Azure Language capabilities such as sentiment analysis, key phrase extra…</itunes:summary>
      <itunes:subtitle>A text-analysis application should return the structured result the business needs rather than another paragraph that someone must interpret. This episode introduces a lightweight implementation pattern for Azure Language capabilities such as sentiment an</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/f5f7bfea/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Implementing Key Phrase Extraction and Named Entity Recognition</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>42</itunes:episode>
      <podcast:episode>42</podcast:episode>
      <itunes:title>Implementing Key Phrase Extraction and Named Entity Recognition</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">5f634351-8a6c-4914-a0a0-bc158e5d838f</guid>
      <link>https://share.transistor.fm/s/128be8bd</link>
      <description>
        <![CDATA[The same press release can contain both important topic phrases and recognized organization or location names, but those are different outputs. This episode shows how a lightweight language application can use key phrase extraction and named entity recognition without confusing the two. Learners define what the downstream system expects, submit the source text to the appropriate capability, inspect the returned results, and compare them with the known content. The lesson reinforces that language features may share the same input while serving different decisions. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[The same press release can contain both important topic phrases and recognized organization or location names, but those are different outputs. This episode shows how a lightweight language application can use key phrase extraction and named entity recognition without confusing the two. Learners define what the downstream system expects, submit the source text to the appropriate capability, inspect the returned results, and compare them with the known content. The lesson reinforces that language features may share the same input while serving different decisions. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:46:29 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/128be8bd/386b2d0b.mp3" length="18471823" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1149</itunes:duration>
      <itunes:summary>The same press release can contain both important topic phrases and recognized organization or location names, but those are different outputs. This episode shows how a lightweight language application can use key phrase extraction and named entity recognition without confusing…</itunes:summary>
      <itunes:subtitle>The same press release can contain both important topic phrases and recognized organization or location names, but those are different outputs. This episode shows how a lightweight language application can use key phrase extraction and named entity recogn</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/128be8bd/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Implementing Sentiment Analysis</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>43</itunes:episode>
      <podcast:episode>43</podcast:episode>
      <itunes:title>Implementing Sentiment Analysis</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">1bc4ec46-8443-4233-b0cd-7c96a34384c4</guid>
      <link>https://share.transistor.fm/s/fe35ecab</link>
      <description>
        <![CDATA[A support dashboard may need a sentiment indicator while still preserving the customer's original words for human review. This episode explains how sentiment analysis classifies expressed opinion and why that classification should not be treated as a complete interpretation of the message. Learners build the reasoning for a lightweight implementation, submit text to the appropriate Azure Language capability, interpret the returned sentiment result, and validate it against the source. The episode also reinforces that sentiment is distinct from key phrases, entities, summarization, and the original text itself. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A support dashboard may need a sentiment indicator while still preserving the customer's original words for human review. This episode explains how sentiment analysis classifies expressed opinion and why that classification should not be treated as a complete interpretation of the message. Learners build the reasoning for a lightweight implementation, submit text to the appropriate Azure Language capability, interpret the returned sentiment result, and validate it against the source. The episode also reinforces that sentiment is distinct from key phrases, entities, summarization, and the original text itself. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:46:46 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/fe35ecab/ea7b44ce.mp3" length="17053656" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1061</itunes:duration>
      <itunes:summary>A support dashboard may need a sentiment indicator while still preserving the customer's original words for human review. This episode explains how sentiment analysis classifies expressed opinion and why that classification should not be treated as a complete interpretation of t…</itunes:summary>
      <itunes:subtitle>A support dashboard may need a sentiment indicator while still preserving the customer's original words for human review. This episode explains how sentiment analysis classifies expressed opinion and why that classification should not be treated as a comp</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/fe35ecab/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Implementing Text and Conversation Summarization</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>44</itunes:episode>
      <podcast:episode>44</podcast:episode>
      <itunes:title>Implementing Text and Conversation Summarization</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">b3d92158-de3c-411e-8c8c-68f706efbff4</guid>
      <link>https://share.transistor.fm/s/166b8cbc</link>
      <description>
        <![CDATA[Summarization can help reviewers understand long text or conversations more quickly, but a summary must not be mistaken for the complete source. This episode explains how Azure Language summarization fits a case-management workload and how an application can send substantial source content, receive a concise result, and preserve the original material for verification. Learners define what the summary must retain, compare the result with the source, and recognize that sentiment analysis or other language classifications do not satisfy the same requirement. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Summarization can help reviewers understand long text or conversations more quickly, but a summary must not be mistaken for the complete source. This episode explains how Azure Language summarization fits a case-management workload and how an application can send substantial source content, receive a concise result, and preserve the original material for verification. Learners define what the summary must retain, compare the result with the source, and recognize that sentiment analysis or other language classifications do not satisfy the same requirement. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:47:03 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/166b8cbc/fb625beb.mp3" length="17658878" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1099</itunes:duration>
      <itunes:summary>Summarization can help reviewers understand long text or conversations more quickly, but a summary must not be mistaken for the complete source. This episode explains how Azure Language summarization fits a case-management workload and how an application can send substantial sou…</itunes:summary>
      <itunes:subtitle>Summarization can help reviewers understand long text or conversations more quickly, but a summary must not be mistaken for the complete source. This episode explains how Azure Language summarization fits a case-management workload and how an application </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/166b8cbc/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Designing a Voice-Enabled AI Application</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>45</itunes:episode>
      <podcast:episode>45</podcast:episode>
      <itunes:title>Designing a Voice-Enabled AI Application</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">1e3ef9ae-a117-41a2-8dcd-5aada31a034c</guid>
      <link>https://share.transistor.fm/s/ade6cbe8</link>
      <description>
        <![CDATA[A voice assistant is a pipeline, not a single magic capability. This episode traces spoken input from the microphone through speech handling, model processing, and application response, showing where speech-to-text and text-to-speech may fit around a generative model. Learners identify what each stage receives and produces, determine whether raw audio can be processed directly or must first become text, and define the evidence required to validate the complete interaction. The goal is to reason about the end-to-end voice experience rather than name one Azure service. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A voice assistant is a pipeline, not a single magic capability. This episode traces spoken input from the microphone through speech handling, model processing, and application response, showing where speech-to-text and text-to-speech may fit around a generative model. Learners identify what each stage receives and produces, determine whether raw audio can be processed directly or must first become text, and define the evidence required to validate the complete interaction. The goal is to reason about the end-to-end voice experience rather than name one Azure service. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:47:20 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/ade6cbe8/b3e53498.mp3" length="17117195" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1065</itunes:duration>
      <itunes:summary>A voice assistant is a pipeline, not a single magic capability. This episode traces spoken input from the microphone through speech handling, model processing, and application response, showing where speech-to-text and text-to-speech may fit around a generative model. Learners i…</itunes:summary>
      <itunes:subtitle>A voice assistant is a pipeline, not a single magic capability. This episode traces spoken input from the microphone through speech handling, model processing, and application response, showing where speech-to-text and text-to-speech may fit around a gene</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/ade6cbe8/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Building Real-Time Spoken Interactions</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>46</itunes:episode>
      <podcast:episode>46</podcast:episode>
      <itunes:title>Building Real-Time Spoken Interactions</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">aa7dfbb1-a106-41fa-bb8f-7134189f474d</guid>
      <link>https://share.transistor.fm/s/671268b0</link>
      <description>
        <![CDATA[A live voice assistant fails the user experience if it treats every spoken request as an offline batch job. This episode explains the timing requirements of spoken interaction and why real-time or near-real-time processing differs from asynchronous transcription of prerecorded files. It also connects spoken prompts to multimodal model scenarios in the AI-901 scope. Learners examine latency, input flow, speech recognition, model processing, and response behavior to determine whether the architecture supports an actual conversation rather than a sequence of delayed recordings. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A live voice assistant fails the user experience if it treats every spoken request as an offline batch job. This episode explains the timing requirements of spoken interaction and why real-time or near-real-time processing differs from asynchronous transcription of prerecorded files. It also connects spoken prompts to multimodal model scenarios in the AI-901 scope. Learners examine latency, input flow, speech recognition, model processing, and response behavior to determine whether the architecture supports an actual conversation rather than a sequence of delayed recordings. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:47:37 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/671268b0/0c9b860f.mp3" length="19185673" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1194</itunes:duration>
      <itunes:summary>A live voice assistant fails the user experience if it treats every spoken request as an offline batch job. This episode explains the timing requirements of spoken interaction and why real-time or near-real-time processing differs from asynchronous transcription of prerecorded f…</itunes:summary>
      <itunes:subtitle>A live voice assistant fails the user experience if it treats every spoken request as an offline batch job. This episode explains the timing requirements of spoken interaction and why real-time or near-real-time processing differs from asynchronous transc</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/671268b0/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Connecting Applications to Azure Speech</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>47</itunes:episode>
      <podcast:episode>47</podcast:episode>
      <itunes:title>Connecting Applications to Azure Speech</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">b8a27ee5-0c2b-45ca-b716-fd1b5498b935</guid>
      <link>https://share.transistor.fm/s/51671543</link>
      <description>
        <![CDATA[A desktop or Python application can use Azure Speech capabilities without requiring a person to operate the portal for every interaction. This episode explains how speech-to-text and text-to-speech fit into an application and how a developer reasons from the required input and output to the appropriate speech capability and client integration. Learners distinguish recognition from synthesis, identify the connection and authorization requirements, and define tests that prove the application can receive spoken audio, obtain text, and later synthesize an expected spoken response. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A desktop or Python application can use Azure Speech capabilities without requiring a person to operate the portal for every interaction. This episode explains how speech-to-text and text-to-speech fit into an application and how a developer reasons from the required input and output to the appropriate speech capability and client integration. Learners distinguish recognition from synthesis, identify the connection and authorization requirements, and define tests that prove the application can receive spoken audio, obtain text, and later synthesize an expected spoken response. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:47:54 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/51671543/d7473946.mp3" length="18704185" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1164</itunes:duration>
      <itunes:summary>A desktop or Python application can use Azure Speech capabilities without requiring a person to operate the portal for every interaction. This episode explains how speech-to-text and text-to-speech fit into an application and how a developer reasons from the required input and o…</itunes:summary>
      <itunes:subtitle>A desktop or Python application can use Azure Speech capabilities without requiring a person to operate the portal for every interaction. This episode explains how speech-to-text and text-to-speech fit into an application and how a developer reasons from </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/51671543/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Real-Time vs. Batch Speech-to-Text</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>48</itunes:episode>
      <podcast:episode>48</podcast:episode>
      <itunes:title>Real-Time vs. Batch Speech-to-Text</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">ef6ad11b-4f66-4c01-9f05-a7eac9bd4ebd</guid>
      <link>https://share.transistor.fm/s/f1cc0d98</link>
      <description>
        <![CDATA[Live calls and large archives both require speech recognition, but the timing and volume of the work are different. This episode compares real-time transcription for interactive audio with batch transcription for prerecorded content that can be processed asynchronously. Learners identify when the application needs text quickly enough to support an ongoing interaction and when throughput over many stored recordings matters more. The decision begins with how audio arrives, when the result is needed, and how the application will consume the transcript rather than with the shared label speech-to-text. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Live calls and large archives both require speech recognition, but the timing and volume of the work are different. This episode compares real-time transcription for interactive audio with batch transcription for prerecorded content that can be processed asynchronously. Learners identify when the application needs text quickly enough to support an ongoing interaction and when throughput over many stored recordings matters more. The decision begins with how audio arrives, when the result is needed, and how the application will consume the transcript rather than with the shared label speech-to-text. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:48:11 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/f1cc0d98/d0f42516.mp3" length="18837509" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1172</itunes:duration>
      <itunes:summary>Live calls and large archives both require speech recognition, but the timing and volume of the work are different. This episode compares real-time transcription for interactive audio with batch transcription for prerecorded content that can be processed asynchronously. Learners…</itunes:summary>
      <itunes:subtitle>Live calls and large archives both require speech recognition, but the timing and volume of the work are different. This episode compares real-time transcription for interactive audio with batch transcription for prerecorded content that can be processed </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/f1cc0d98/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Real-Time vs. Batch Text-to-Speech</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>49</itunes:episode>
      <podcast:episode>49</podcast:episode>
      <itunes:title>Real-Time vs. Batch Text-to-Speech</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">de538c50-beb0-49ca-b289-414fb201e06d</guid>
      <link>https://share.transistor.fm/s/174d2c5a</link>
      <description>
        <![CDATA[Short interactive prompts and full narrated lessons both convert text into speech, but they require different processing patterns. This episode compares real-time synthesis for immediate application responses with asynchronous generation for longer audio content that can finish later. Learners evaluate timing, volume, output handling, and user expectations before choosing the implementation mode. The lesson reinforces that the direction of the transformation is the same while the operational requirement determines whether real-time or batch processing is the closer fit. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Short interactive prompts and full narrated lessons both convert text into speech, but they require different processing patterns. This episode compares real-time synthesis for immediate application responses with asynchronous generation for longer audio content that can finish later. Learners evaluate timing, volume, output handling, and user expectations before choosing the implementation mode. The lesson reinforces that the direction of the transformation is the same while the operational requirement determines whether real-time or batch processing is the closer fit. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:48:29 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/174d2c5a/864a6271.mp3" length="17970243" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1118</itunes:duration>
      <itunes:summary>Short interactive prompts and full narrated lessons both convert text into speech, but they require different processing patterns. This episode compares real-time synthesis for immediate application responses with asynchronous generation for longer audio content that can finish…</itunes:summary>
      <itunes:subtitle>Short interactive prompts and full narrated lessons both convert text into speech, but they require different processing patterns. This episode compares real-time synthesis for immediate application responses with asynchronous generation for longer audio </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/174d2c5a/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Troubleshooting Speech Recognition and Synthesis</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>50</itunes:episode>
      <podcast:episode>50</podcast:episode>
      <itunes:title>Troubleshooting Speech Recognition and Synthesis</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">91edb554-b6e4-448f-9c8a-07266f1c527f</guid>
      <link>https://share.transistor.fm/s/34de6ae3</link>
      <description>
        <![CDATA[Late transcripts and misrecognized technical terms are different symptoms and may have different causes. This episode teaches a structured approach to troubleshooting speech workloads by separating processing mode, audio input, recognition accuracy, vocabulary, configuration, and output behavior. It also reinforces the direction of speech-to-text and text-to-speech so that a familiar product name does not replace diagnosis. Learners form a prediction, change the lowest-complexity relevant factor, and compare the observed result with the expected improvement instead of assuming that hardware is responsible for every speech problem. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Late transcripts and misrecognized technical terms are different symptoms and may have different causes. This episode teaches a structured approach to troubleshooting speech workloads by separating processing mode, audio input, recognition accuracy, vocabulary, configuration, and output behavior. It also reinforces the direction of speech-to-text and text-to-speech so that a familiar product name does not replace diagnosis. Learners form a prediction, change the lowest-complexity relevant factor, and compare the observed result with the expected improvement instead of assuming that hardware is responsible for every speech problem. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:48:45 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/34de6ae3/08828363.mp3" length="16929539" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1053</itunes:duration>
      <itunes:summary>Late transcripts and misrecognized technical terms are different symptoms and may have different causes. This episode teaches a structured approach to troubleshooting speech workloads by separating processing mode, audio input, recognition accuracy, vocabulary, configuration, an…</itunes:summary>
      <itunes:subtitle>Late transcripts and misrecognized technical terms are different symptoms and may have different causes. This episode teaches a structured approach to troubleshooting speech workloads by separating processing mode, audio input, recognition accuracy, vocab</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/34de6ae3/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Interpreting Images with Multimodal Models</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>51</itunes:episode>
      <podcast:episode>51</podcast:episode>
      <itunes:title>Interpreting Images with Multimodal Models</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">83dde15d-6175-435c-a7ca-69befb6258ae</guid>
      <link>https://share.transistor.fm/s/f4686b25</link>
      <description>
        <![CDATA[A vision-enabled multimodal model can receive an image together with language that tells it what to look for. This episode uses a manufacturing-quality scenario to explain how visual input and a focused prompt work together through a supported Foundry model interface. Learners define the image-analysis requirement, select a model that can accept the modality, ask a question tied to the task, and validate the response against known visual evidence. The lesson also emphasizes that a plausible visual answer is not automatically a verified fact. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A vision-enabled multimodal model can receive an image together with language that tells it what to look for. This episode uses a manufacturing-quality scenario to explain how visual input and a focused prompt work together through a supported Foundry model interface. Learners define the image-analysis requirement, select a model that can accept the modality, ask a question tied to the task, and validate the response against known visual evidence. The lesson also emphasizes that a plausible visual answer is not automatically a verified fact. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:49:02 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/f4686b25/c61ca91c.mp3" length="18268674" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1137</itunes:duration>
      <itunes:summary>A vision-enabled multimodal model can receive an image together with language that tells it what to look for. This episode uses a manufacturing-quality scenario to explain how visual input and a focused prompt work together through a supported Foundry model interface. Learners d…</itunes:summary>
      <itunes:subtitle>A vision-enabled multimodal model can receive an image together with language that tells it what to look for. This episode uses a manufacturing-quality scenario to explain how visual input and a focused prompt work together through a supported Foundry mod</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/f4686b25/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Vision Detail, Limitations, and Validation</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>52</itunes:episode>
      <podcast:episode>52</podcast:episode>
      <itunes:title>Vision Detail, Limitations, and Validation</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">e65a6a17-7b2f-4505-a1f1-d6adf6f896d3</guid>
      <link>https://share.transistor.fm/s/1bbeb8ff</link>
      <description>
        <![CDATA[A model may correctly describe an image while still missing a small serial number or another detail the application considers critical. This episode explores the limits of vision-enabled models and explains why changing an image-detail setting is not a substitute for defining the exact requirement. Learners distinguish broad visual interpretation from precise text or attribute recognition, test representative images, and validate the returned result against known values. The focus is on the evidence needed to determine whether the vision implementation is accurate enough for the intended task. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A model may correctly describe an image while still missing a small serial number or another detail the application considers critical. This episode explores the limits of vision-enabled models and explains why changing an image-detail setting is not a substitute for defining the exact requirement. Learners distinguish broad visual interpretation from precise text or attribute recognition, test representative images, and validate the returned result against known values. The focus is on the evidence needed to determine whether the vision implementation is accurate enough for the intended task. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:49:19 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/1bbeb8ff/100beb9a.mp3" length="19171466" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1193</itunes:duration>
      <itunes:summary>A model may correctly describe an image while still missing a small serial number or another detail the application considers critical. This episode explores the limits of vision-enabled models and explains why changing an image-detail setting is not a substitute for defining th…</itunes:summary>
      <itunes:subtitle>A model may correctly describe an image while still missing a small serial number or another detail the application considers critical. This episode explores the limits of vision-enabled models and explains why changing an image-detail setting is not a su</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/1bbeb8ff/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Generating Images with Foundry Agent Service</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>53</itunes:episode>
      <podcast:episode>53</podcast:episode>
      <itunes:title>Generating Images with Foundry Agent Service</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">36a2b8ed-44ca-4252-bd31-4892e76ad63f</guid>
      <link>https://share.transistor.fm/s/2dc47fee</link>
      <description>
        <![CDATA[When the required output is a new image created from a written brief, the solution needs image generation rather than visual interpretation. This episode explains how Foundry Agent Service can use an image-generation tool when the required model and project configuration are available. Learners define the creative requirement, provide the text instruction, generate visual output, and compare the result with the intended brief. The lesson keeps the distinction between creating an image, analyzing an existing image, and extracting structured information from visual content. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[When the required output is a new image created from a written brief, the solution needs image generation rather than visual interpretation. This episode explains how Foundry Agent Service can use an image-generation tool when the required model and project configuration are available. Learners define the creative requirement, provide the text instruction, generate visual output, and compare the result with the intended brief. The lesson keeps the distinction between creating an image, analyzing an existing image, and extracting structured information from visual content. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:49:36 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/2dc47fee/b41b4444.mp3" length="18245270" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1135</itunes:duration>
      <itunes:summary>When the required output is a new image created from a written brief, the solution needs image generation rather than visual interpretation. This episode explains how Foundry Agent Service can use an image-generation tool when the required model and project configuration are ava…</itunes:summary>
      <itunes:subtitle>When the required output is a new image created from a written brief, the solution needs image generation rather than visual interpretation. This episode explains how Foundry Agent Service can use an image-generation tool when the required model and proje</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/2dc47fee/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Evaluating Generated Images for Responsible Use</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>54</itunes:episode>
      <podcast:episode>54</podcast:episode>
      <itunes:title>Evaluating Generated Images for Responsible Use</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">54739f18-11c0-4541-8345-a58e6bd40d0f</guid>
      <link>https://share.transistor.fm/s/181d0915</link>
      <description>
        <![CDATA[A valid image file is not automatically suitable for publication. This episode examines generated-image evaluation through a marketing scenario in which the visual output conflicts with the intended audience, activity, or placement even though generation technically succeeded. Learners consider the brief, safety expectations, representation, intended use, and relevant Responsible AI concerns before accepting the result. The lesson reinforces that technical completion and responsible approval are separate decisions and that generated visual content should be reviewed against explicit acceptance criteria. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A valid image file is not automatically suitable for publication. This episode examines generated-image evaluation through a marketing scenario in which the visual output conflicts with the intended audience, activity, or placement even though generation technically succeeded. Learners consider the brief, safety expectations, representation, intended use, and relevant Responsible AI concerns before accepting the result. The lesson reinforces that technical completion and responsible approval are separate decisions and that generated visual content should be reviewed against explicit acceptance criteria. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:49:52 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/181d0915/ebff707c.mp3" length="17059523" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1061</itunes:duration>
      <itunes:summary>A valid image file is not automatically suitable for publication. This episode examines generated-image evaluation through a marketing scenario in which the visual output conflicts with the intended audience, activity, or placement even though generation technically succeeded. L…</itunes:summary>
      <itunes:subtitle>A valid image file is not automatically suitable for publication. This episode examines generated-image evaluation through a marketing scenario in which the visual output conflicts with the intended audience, activity, or placement even though generation </itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/181d0915/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Building a Lightweight Vision Application</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>55</itunes:episode>
      <podcast:episode>55</podcast:episode>
      <itunes:title>Building a Lightweight Vision Application</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">8302488b-69e7-4a5f-b8cf-07dc4bcbaa61</guid>
      <link>https://share.transistor.fm/s/c1d8947e</link>
      <description>
        <![CDATA[A lightweight vision application must do more than upload a photograph. This episode follows a small client that sends an image and focused question to a vision-enabled model, receives the response, and marks the result for appropriate verification. Learners identify the model capability, prepare the input, submit the request through the intended interface, handle the output, and compare it with the original requirement. The emphasis is on building and validating the complete input-to-response path while preserving the limits of what the model has actually established. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A lightweight vision application must do more than upload a photograph. This episode follows a small client that sends an image and focused question to a vision-enabled model, receives the response, and marks the result for appropriate verification. Learners identify the model capability, prepare the input, submit the request through the intended interface, handle the output, and compare it with the original requirement. The emphasis is on building and validating the complete input-to-response path while preserving the limits of what the model has actually established. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:50:10 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/c1d8947e/0024ddc6.mp3" length="18715889" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1165</itunes:duration>
      <itunes:summary>A lightweight vision application must do more than upload a photograph. This episode follows a small client that sends an image and focused question to a vision-enabled model, receives the response, and marks the result for appropriate verification. Learners identify the model c…</itunes:summary>
      <itunes:subtitle>A lightweight vision application must do more than upload a photograph. This episode follows a small client that sends an image and focused question to a vision-enabled model, receives the response, and marks the result for appropriate verification. Learn</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/c1d8947e/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Designing Content Understanding Analyzers</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>56</itunes:episode>
      <podcast:episode>56</podcast:episode>
      <itunes:title>Designing Content Understanding Analyzers</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">6a7ee347-e9ac-4aa5-941f-447b08bbe8c9</guid>
      <link>https://share.transistor.fm/s/063e42c9</link>
      <description>
        <![CDATA[Azure Content Understanding can process documents, images, audio, and video into predictable structured outputs, but the system still needs a definition of what information should be produced. This episode introduces analyzers as the processing definition that connects unstructured input to an expected schema or output. Learners begin with a mixed-content business workflow, identify the fields the downstream system needs, choose processing that fits each modality, and validate whether the analyzer returns the required structure rather than an unspecified AI response. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Azure Content Understanding can process documents, images, audio, and video into predictable structured outputs, but the system still needs a definition of what information should be produced. This episode introduces analyzers as the processing definition that connects unstructured input to an expected schema or output. Learners begin with a mixed-content business workflow, identify the fields the downstream system needs, choose processing that fits each modality, and validate whether the analyzer returns the required structure rather than an unspecified AI response. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:50:26 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/063e42c9/d367e424.mp3" length="17904631" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1114</itunes:duration>
      <itunes:summary>Azure Content Understanding can process documents, images, audio, and video into predictable structured outputs, but the system still needs a definition of what information should be produced. This episode introduces analyzers as the processing definition that connects unstructu…</itunes:summary>
      <itunes:subtitle>Azure Content Understanding can process documents, images, audio, and video into predictable structured outputs, but the system still needs a definition of what information should be produced. This episode introduces analyzers as the processing definition</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/063e42c9/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Extracting Structured Data from Documents and Forms</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>57</itunes:episode>
      <podcast:episode>57</podcast:episode>
      <itunes:title>Extracting Structured Data from Documents and Forms</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">3cd763a4-dbf9-4861-9636-0bc2bb696957</guid>
      <link>https://share.transistor.fm/s/dcd4abbe</link>
      <description>
        <![CDATA[Reading the text on an invoice is not the same as extracting the vendor, total, table rows, selection marks, and other fields a business process needs. This episode explains document and form extraction with Azure Content Understanding, including text, layout, tables, fields, and structured output. Learners define the required schema, select or configure an analyzer, process representative documents, and verify that the result preserves the relationships the downstream application requires. The lesson also distinguishes structured extraction from OCR-only thinking. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Reading the text on an invoice is not the same as extracting the vendor, total, table rows, selection marks, and other fields a business process needs. This episode explains document and form extraction with Azure Content Understanding, including text, layout, tables, fields, and structured output. Learners define the required schema, select or configure an analyzer, process representative documents, and verify that the result preserves the relationships the downstream application requires. The lesson also distinguishes structured extraction from OCR-only thinking. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:50:43 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/dcd4abbe/4c635a16.mp3" length="18793222" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1170</itunes:duration>
      <itunes:summary>Reading the text on an invoice is not the same as extracting the vendor, total, table rows, selection marks, and other fields a business process needs. This episode explains document and form extraction with Azure Content Understanding, including text, layout, tables, fields, an…</itunes:summary>
      <itunes:subtitle>Reading the text on an invoice is not the same as extracting the vendor, total, table rows, selection marks, and other fields a business process needs. This episode explains document and form extraction with Azure Content Understanding, including text, la</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/dcd4abbe/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Extracting Structured Information from Images</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>58</itunes:episode>
      <podcast:episode>58</podcast:episode>
      <itunes:title>Extracting Structured Information from Images</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">18b3a8c6-b735-4300-bf6d-682f73f9b6ad</guid>
      <link>https://share.transistor.fm/s/c2610799</link>
      <description>
        <![CDATA[A claims application may need defined attributes from a damage photograph rather than a free-form caption. This episode explains image information extraction with Azure Content Understanding and shows how an analyzer and field schema give the output a predictable structure. Learners distinguish broad image description from extracting named fields that software can store, compare, route, or review. Validation focuses on whether the expected attributes are present and correct for representative images, not merely on whether the generated language sounds convincing. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[A claims application may need defined attributes from a damage photograph rather than a free-form caption. This episode explains image information extraction with Azure Content Understanding and shows how an analyzer and field schema give the output a predictable structure. Learners distinguish broad image description from extracting named fields that software can store, compare, route, or review. Validation focuses on whether the expected attributes are present and correct for representative images, not merely on whether the generated language sounds convincing. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:50:59 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/c2610799/e1512216.mp3" length="17360870" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1080</itunes:duration>
      <itunes:summary>A claims application may need defined attributes from a damage photograph rather than a free-form caption. This episode explains image information extraction with Azure Content Understanding and shows how an analyzer and field schema give the output a predictable structure. Lear…</itunes:summary>
      <itunes:subtitle>A claims application may need defined attributes from a damage photograph rather than a free-form caption. This episode explains image information extraction with Azure Content Understanding and shows how an analyzer and field schema give the output a pre</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/c2610799/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Extracting Information from Audio and Video</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>59</itunes:episode>
      <podcast:episode>59</podcast:episode>
      <itunes:title>Extracting Information from Audio and Video</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">10122d8f-125b-424b-a3a5-23874b6ba0f1</guid>
      <link>https://share.transistor.fm/s/56bd4759</link>
      <description>
        <![CDATA[Recorded meetings, training sessions, and videos can require more than a plain transcript. This episode explains how Azure Content Understanding can extract structured information from audio and video so that downstream applications receive organized results rather than one long block of text. Learners define the fields and evidence the workflow needs, configure the analyzer around that expected output, and compare the result with the source content. The lesson also separates structured extraction from speech transcription, which may preserve words without producing the business fields required by the application. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[Recorded meetings, training sessions, and videos can require more than a plain transcript. This episode explains how Azure Content Understanding can extract structured information from audio and video so that downstream applications receive organized results rather than one long block of text. Learners define the fields and evidence the workflow needs, configure the analyzer around that expected output, and compare the result with the source content. The lesson also separates structured extraction from speech transcription, which may preserve words without producing the business fields required by the application. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:51:18 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/56bd4759/32d23a15.mp3" length="18888091" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1175</itunes:duration>
      <itunes:summary>Recorded meetings, training sessions, and videos can require more than a plain transcript. This episode explains how Azure Content Understanding can extract structured information from audio and video so that downstream applications receive organized results rather than one long…</itunes:summary>
      <itunes:subtitle>Recorded meetings, training sessions, and videos can require more than a plain transcript. This episode explains how Azure Content Understanding can extract structured information from audio and video so that downstream applications receive organized resu</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/56bd4759/transcript.txt" type="text/plain"/>
    </item>
    <item>
      <title>Building a Multimodal Content Understanding Workflow</title>
      <itunes:season>1</itunes:season>
      <podcast:season>1</podcast:season>
      <itunes:episode>60</itunes:episode>
      <podcast:episode>60</podcast:episode>
      <itunes:title>Building a Multimodal Content Understanding Workflow</itunes:title>
      <itunes:episodeType>full</itunes:episodeType>
      <guid isPermaLink="false">ae670503-c670-4bfa-9fdb-34c68c2ee667</guid>
      <link>https://share.transistor.fm/s/1becdc64</link>
      <description>
        <![CDATA[The final episode brings documents, photographs, audio, and video together in one claims-intake workflow. It shows how Azure Content Understanding can convert several kinds of unstructured content into a consistent case record containing the fields another application needs. Learners apply the complete decision chain: identify each input, define the structured output, select the appropriate analyzer behavior, implement the workflow, and verify the result across modalities. The capstone reinforces that multimodal automation succeeds when the output contract is clear and evidence shows that each source contributes the required information accurately. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </description>
      <content:encoded>
        <![CDATA[The final episode brings documents, photographs, audio, and video together in one claims-intake workflow. It shows how Azure Content Understanding can convert several kinds of unstructured content into a consistent case record containing the fields another application needs. Learners apply the complete decision chain: identify each input, define the structured output, select the appropriate analyzer behavior, implement the workflow, and verify the result across modalities. The capstone reinforces that multimodal automation succeeds when the output contract is clear and evidence shows that each source contributes the required information accurately. Visit BareMetalCyber.com for more certification preparation, technology education, and practical AI guidance.]]>
      </content:encoded>
      <pubDate>Tue, 17 Nov 2026 11:51:38 -0600</pubDate>
      <author>Dr Jason Edwards</author>
      <enclosure url="https://media.transistor.fm/1becdc64/67fdfd91.mp3" length="19385889" type="audio/mpeg"/>
      <itunes:author>Dr Jason Edwards</itunes:author>
      <itunes:duration>1207</itunes:duration>
      <itunes:summary>The final episode brings documents, photographs, audio, and video together in one claims-intake workflow. It shows how Azure Content Understanding can convert several kinds of unstructured content into a consistent case record containing the fields another application needs. Lea…</itunes:summary>
      <itunes:subtitle>The final episode brings documents, photographs, audio, and video together in one claims-intake workflow. It shows how Azure Content Understanding can convert several kinds of unstructured content into a consistent case record containing the fields anothe</itunes:subtitle>
      <itunes:keywords>Azure AI Fundamentals, AI-901, Microsoft Azure, Microsoft Foundry, Azure AI, generative AI, Responsible AI, AI agents, Azure Language, Azure Speech, computer vision, multimodal AI, Content Understanding, certification training, exam preparation</itunes:keywords>
      <itunes:explicit>No</itunes:explicit>
      <podcast:transcript url="https://share.transistor.fm/s/1becdc64/transcript.txt" type="text/plain"/>
    </item>
  </channel>
</rss>
