Lenny’s AI Builders

ChatGPT wants your whole day - so i let it audit mine | LAB 0012

Lennox Saint — Sat, 03 Oct 2026 16:23:50 +0800

ChatGPT wants your whole day. it now takes your meeting notes with no bot on the call. it hosts apps you build with one sentence. so i let GPT-6.1 Sol audit my life and design my ChatGPT Space. it worked for 52m 54s. Claude Opus 5.5 checked the plan over four rounds until it said AGREE. what came out: one "Lennox OS" Space where i capture, plan, start work, review results and find my history. the hookups to my other apps aren't built yet - i show you exactly where it stands. also in this one: Google's Gemini 4 Argon can write 1M tokens in one reply - and only trusted testers can use it. OpenAI's GPT-6.1 Sol. Hormozi's memo on AI budgets. and the White House AI Accord. sponsor: Threadify, my lead generation agent for Threads. start a free trial: https://www.threadify.app/plans?utm_source=lenny-podcast&utm_medium=audio&utm_campaign=lab-0012&utm_content=podcast-description-primary&video_slug=lab-0012&cta_slot=description&entry_angle=sponsor&lp_variant=plans sources Google: Gemini 4 Argon: https://x.com/Google/status/2105388143902175529 Google DeepMind: trusted testers via Fairwind: https://x.com/GoogleDeepMind/status/2105388084154056939 Derrick Choi: ChatGPT Meetings plugin: https://x.com/derrickcchoi/status/2105304132487664041 OpenAI release notes (Meetings beta): https://help.openai.com/en/articles/6825453-chatgpt-release-notes Tibo on GPT-6.1 Sol: https://x.com/thsottiaux/status/2105464274747527543 GPT-6.1 Sol model docs: https://developers.openai.com/api/docs/models/gpt-6.1-sol Claude model overview (Opus 5.5 output limit): https://platform.claude.com/docs/en/about-claude/models/overview Alex Hormozi's AI budget memo: https://x.com/AlexHormozi/status/2105446365119525013 Jensen Huang on the White House AI Accord: https://x.com/JensenHuang/status/2105348324484342238 The Hill: tech CEOs sign White House AI accord: https://thehill.com/homenews/administration/6118906-tech-ceos-sign-white-house-ai-accord/ Reuters via Yahoo: FTC opens AI probe: https://www.yahoo.com/news/politics/articles/ftc-opens-probe-ai-giants-135328719.html Max Stoiber: ChatGPT Sites: https://x.com/mxstbr/status/2105428405571428785 DataCamp: Gemini 4 Argon pricing (reported): https://www.datacamp.com/blog/gemini-4-argon prices and access details marked "reported" come from third parties, not the companies. Lenny's AI Builders - LAB 0012

OpenAI's $500 plan feels like the old $200 plan - but dots are free | LAB 0011

Lennox Saint — Thu, 01 Oct 2026 00:48:42 +0800

OpenAI's new $500 plan feels like the old $200 plan. that's one of six stories from OpenAI DevDay 2026 - then i set up my own dot live, on day one, with the first prompt every dot needs. the stories: the new $500 Pro plan (25x the usage of Plus, plus Ultrafast - up to 8x faster, and $60/$300 per million tokens for Astra in the API) and the $200 plan dropping from 20x to 10x Plus (Tibo: it will "net out at half the dollar in API spend"), GPT-6.1 Sol at $2/$10 per million tokens vs GPT-6 Astra's $10/$50, dots (always-on GPT-6 Astra agents with their own cloud computer and browser, 4,000+ apps, reachable in ChatGPT, Slack, Teams or by voice call), Codex cloud (close your laptop, your agents keep working), Sign in with ChatGPT (your plan pays for the AI in 16+ partner products like Devin, OpenCode and Notion, per Tibo, capped per app) plus the Agents API and the Decisions API (about 150 ms, limited preview), and ChatGPT Space and pages. the fine print: OpenAI's email says existing Pro subscribers keep their current limits through October 29, 2026, plus a one-time grant of 62,500 usage credits ($2,500 value). "feels like the old $200 plan" is my reading, not OpenAI's words. BridgeMind says he ran out of his entire weekly limit on the $500 plan in under 30 minutes - his claim. Artificial Analysis scores: Sol 52, Opus 5.5 58, Sonnet 5.5 56. one tester found Sol uses about half the tokens of Sonnet 5.5. dots come with Pro plans for now - chats with your dot don't use limits, but Codex or Work jobs it starts can. not in the EU or UK yet, and Dan Shipper (Every) reported permission issues and dropped messages. Codex cloud uses normal plan usage, and it's not in my app yet, so it's untested. the API use cases are my examples; the Hugging Face link is my own theory. the result: i set up my dot, Clarity, live on my Pro plan and ran the "restate and audit" prompt. about 20 minutes after i sent it, it restated my goals, audited my recurring jobs (84 Codex entries with 25 labelled active, 6 of those with no future run, all 9 old wiki jobs paused) and proposed one job: "one follow-through for one low-risk PR using the existing process rather than adding another monitor". nothing was approved on camera. one day-one run on my own setup - the audit is my dot's opinion. the takeaway: audit before you automate. try it yourself - give your dot this first prompt before any work: "before you do anything, read everything i've connected you to: my repos, Slack, Notion, calendar, and every scheduled task or agent i already run. 1. restate in your own words what you think my goals are and the problem i'm trying to solve. 2. list every recurring job and scheduled agent i run. mark each: keep, kill, or replace with a trigger. 3. pick the one job you'd take off my plate this week. give me the outcome, the tools you need, what you must ask me first, and a weekly budget. don't start work or create any schedule until i approve." Threadify, my own software, sponsors this episode. dots do everything for you - Threadify does one thing all day: it finds leads in Threads conversations and drafts replies in my voice. see the plans here: https://www.threadify.app/plans?utm_source=lenny-youtube&utm_medium=video&utm_campaign=lab-0011&utm_content=youtube-description-primary&video_slug=lab-0011&cta_slot=description&entry_angle=sponsor&lp_variant=plans want to sponsor a future LAB episode? email sponsors@lennysaibuilders.com sources: Tibo on the new Pro plans: https://x.com/thsottiaux/status/2104951965184925941 Tibo on the $200 plan change: https://x.com/thsottiaux/status/2104823812042940713 OpenAI's Pro email, as posted on Reddit: https://www.reddit.com/r/ChatGPTPro/comments/1wtm7o7/10x_usage_drop_for_200_plan_users/ OpenAI API pricing: https://developers.openai.com/api/docs/pricing BridgeMind (his claim): https://x.com/bridgemindai/status/2105016612365770809 OpenAI on GPT-6.1 Sol: https://x.com/OpenAI/status/2104986129686741046 Theo on Artificial Analysis: https://x.com/theo/status/2105005625726099473 OpenAI Devs on dots: https://x.com/OpenAIDevs/status/2104989680987238814 OpenAI Devs on Codex cloud: https://x.com/OpenAIDevs/status/2104997619152130278 Tibo on Sign in with ChatGPT: https://x.com/thsottiaux/status/2105006253986738615 ChatGPT on Space and pages: https://x.com/ChatGPT/status/2104986841145602351 poteto's restate prompt: https://x.com/poteto/status/2104744961904394699 Lenny's AI Builders - LAB 0011

i gave Sonnet 5.5 4 hours of my real work. it scored 48.98 | LAB 0010

Lennox Saint — Wed, 30 Sep 2026 18:48:18 +0800

Anthropic wants to be worth more than $2 trillion, and its own filing warns its AI could act in ways resembling blackmail. that's one of six AI stories this week - then i gave Sonnet 5.5 four hours of my real work on BuilderBench, the benchmark i built for builders, and priced every run. recorded an hour before OpenAI DevDay. the stories: Anthropic's IPO filing, per Reuters, who saw the filing (nearly $4.6 billion in revenue last year, $11.5 billion from April to June this year, more than $8 billion lost running the business last year, $518 billion promised for compute and cloud, the founders keep 50.1% of the votes, and risk language that says its AI could "resist shutdown, conceal or manipulate information" and "act in ways resembling blackmail"), OpenAI cancelling the October release of GPT-6.1 Astra after internal testing showed more deception (reported by the Wall Street Journal - OpenAI's head of safety systems said it "didn't quite meet the bar"), a pre-DevDay leak of app strings pointing to OpenAI agents called Dots that you can text, call, Slack and email, and that can buy things with your approval (a leak via TestingCatalog, not an announcement), Tibo saying the reopened $200 Pro plan will "net out at half the dollar in API spend" compared to the old one, Sonnet 5.5 at $2/$10 per million tokens - the same as GPT-6 Sol - and Anthropic's guide on when to use Sonnet 5.5 and when to pay for Opus 5.5. the plan maths are estimates, not OpenAI's numbers: Theo measured about $9,000 of Opus a month on the $200 Claude Max plan, and another creator estimated about $4,900 a month for the old ChatGPT Pro plan, so half is about $2,450. the Sonnet 5.5 benchmarks are Anthropic's chosen ones (70.6% on Terminal-Bench 4.0 vs 66.4% for Opus 5.5), and Artificial Analysis measured about $7.60 per task - the most tokens they've measured. the result: Sonnet 5.5 scored 48.98 out of 100 on BuilderBench v3. GPT-6 Sol scored 45.55, GPT-6 Astra 53.75 and Opus 5.5 68.26, but those ran on earlier BuilderBench versions (v1 and v2), so it's not a perfect head-to-head. what one run cost: GPT-6 Sol $12.75, Sonnet 5.5 $17.59, Opus 5.5 $49.67, GPT-6 Astra $87.50. my own scores, still provisional. try it yourself - price the job. run this prompt on a real task: "here is a real task from my work: [task]. here is what done looks like: [checklist]. do the whole job. at the end, list every file you made, how long it took, and anything you could not finish." Threadify, my own software, sponsors this episode. it's a lead generation agent for Threads that finds the buyer signals in the comments you're already getting. see the plans and the free trial here: https://www.threadify.app/plans?utm_source=lenny-youtube&utm_medium=video&utm_campaign=lab-0010&utm_content=youtube-description-primary&video_slug=lab-0010&cta_slot=description&entry_angle=sponsor&lp_variant=plans want to sponsor a future LAB episode? email sponsors@lennysaibuilders.com sources: Reuters' reporting on Anthropic's filing, via TechCrunch: https://techcrunch.com/2026/09/28/anthropics-prospectus-details-losses-growth-and-yes-a-warning-that-its-ai-could-end-humanity/ Anthropic on its confidential draft filing: https://www.anthropic.com/news/confidential-draft-s1-sec Andrew Curran on GPT-6.1 Astra's cancelled October release: https://x.com/AndrewCurran_/status/2104711708153618621 TestingCatalog on Dots (leak, pre-DevDay): https://x.com/testingcatalog/status/2104800309612552589 Tibo on the $200 Pro plan: https://x.com/thsottiaux/status/2104823812042940713 Theo on Claude Max usage (measured, his accounts): https://x.com/theo/status/2104683186215363058 AICodeKing on the old Pro plan's value (creator estimate): https://www.youtube.com/watch?v=EIiXhCaZ4rw Claude on Sonnet 5.5: https://x.com/claudeai/status/2104633115620823187 Anthropic's Sonnet 5.5 page: https://www.anthropic.com/claude-sonnet-5-5 Artificial Analysis on Sonnet 5.5: https://x.com/ArtificialAnlys/status/2104640155843989864 Claude Devs, the Sonnet 5.5 guide: https://x.com/ClaudeDevs/status/2104687805876367793 the guide itself: https://claude.dev/blog/building-with-claude-sonnet-5-5/ Lenny's AI Builders - LAB 0010

Claude Opus 5.5 vs GPT-6 Sol - which should builders use? | LAB 0006

Lennox Saint — Mon, 28 Sep 2026 18:52:57 +0800

Opus 5.5 leads the Artificial Analysis chart. GPT-6 Sol has a lower published API list price. The half-price comparison refers to standard short-context API input and output list rates, not subscriptions or cost per finished task. But which one should you use for your own work? i look at the launches, the chart, cost per task, and one Sol miss in my own Notion planning. Then i show how to set up a five-task test for the jobs you actually do. i have not run Opus 5.5 or scored BuilderBench yet, so this episode does not pick a real-work winner. try this prompt: help me build a small benchmark for the work I actually do. Ask for my five recurring tasks. Turn each into one repeatable brief with a useful finished output, checks I can perform, and the time and cost to record. Mark what is still untested. Threadify, my own software, sponsors this episode. It helps you find potential leads in the replies to your Threads posts. See the plans and terms here: https://www.threadify.app/plans?utm_source=lenny-youtube&utm_medium=video&utm_campaign=lab-0006&utm_content=youtube-description-primary&video_slug=lab-0006&cta_slot=description&entry_angle=context&lp_variant=plans want to sponsor a future LAB episode? email sponsors@lennysaibuilders.com sources and tools: Anthropic on Claude Opus 5.5: https://www.anthropic.com/claude-opus-5-5 OpenAI's Sol/Luna announcement: https://openai.com/index/introducing-gpt-6-sol-and-luna/ GPT-6 Sol API list pricing: https://developers.openai.com/api/docs/models/gpt-6-sol Artificial Analysis index write-up: https://artificialanalysis.ai/articles/claude-opus-5-5/ Matt Pocock's Grill with Docs skill: https://github.com/mattpocock/skills/blob/main/skills/engineering/grill-with-docs/SKILL.md Lenny's AI Builders - LAB 0006 watch the episode: https://youtu.be/DAkQhc7Fh3U

OpenAI says its ai solved over 100 maths problems. now what? | LAB 0005

Lennox Saint — Mon, 28 Sep 2026 18:52:37 +0800

i looked at OpenAI's reported maths results, the GPT-6 Sol and Opus 5.5 price fight, Grok 4.7, Muse with Shopify and Qwen-Image-2.1. Then i gave Astra my existing LAB brand and asked for a matching thumbnail, banner and podcast cover. Astra directed the work in Codex; native ImageGen made the images. I show the finished set, the references and the checks. Give it your brand first. The maths result is OpenAI's claim, not an independently established result here. Some model details were reported or rumoured when this was recorded. Qwen's downloadable model needs suitable hardware. Threadify sponsors this episode. It helps turn replies on your Threads posts into lead signals for you to review: https://www.threadify.app/plans?utm_source=lenny-youtube&utm_medium=video&utm_campaign=lab-0005&utm_content=youtube-description-primary&video_slug=lab-0005&cta_slot=description&entry_angle=context&lp_variant=plans sources and updates: OpenAI updates: https://developers.openai.com/api/docs/changelog Claude Opus 5.5: https://www.anthropic.com/claude-opus-5-5 Fable plan limits: https://support.claude.com/en/articles/15424964-claude-fable-models-on-your-plan Muse announcement: https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/ Qwen-Image-2.1 implementation: https://github.com/QwenLM/Qwen-Image-2.1 watch the episode: https://youtu.be/452rqqtYIyg

OpenAI and Anthropic are investigating 10,000s of AI incidents | LAB 0009

Lennox Saint — Mon, 28 Sep 2026 18:12:00 +0800

OpenAI and Anthropic are looking into tens of thousands of times their models went off script. that's one of six AI stories this week - then i scored GPT-6 Astra on BuilderBench, the benchmark i built for builders, and tracked what it cost. the stories: Axios reports OpenAI, Anthropic and security researchers are investigating tens of thousands of incidents where frontier models did things outside evaluators would call problematic (unnamed sources - labs run hundreds of thousands of tests, and most aren't known to have caused real harm), the OpenAI DevDay countdown and what leaked from OpenAI's code (an always-on agent called "o" and an ultrafast speed tier - leaks, not announced), the new ChatGPT app sidebar that made it feel like my whole computer, Tibo saying more resets are coming next week, David Ondrej saying his Opus 5.5 usage barely moves on the Claude Max plan, and a claim that Sonnet 5.5 could land on DevDay (Anthropic has only said "in the coming weeks"). the result: GPT-6 Astra scored 53.75 out of 100 on BuilderBench v2. on v1, Opus 5.5 scored 68.26 and GPT-6 Sol scored 45.55. Astra's estimated API cost was $87.50 ($88.65 with calibration), vs $49.67 for Opus 5.5 and $12.75 for GPT-6 Sol. so Astra cost about $38 more than Opus 5.5 and scored just under 15 points lower. my own scores, still provisional, and Astra ran on v2 while the others ran on v1 - not a perfect head-to-head. try it yourself: run this prompt on a real task. "here is a real task from my work: [task]. here is what done looks like: [checklist]. do the whole job, then list every file you made, how long it took, and anything you could not finish." your work is the benchmark. Threadify, my own software, sponsors this episode. it's a lead generation agent for Threads that finds the buyer signals in the comments you're already getting. see the plans and the free trial here: https://www.threadify.app/plans?utm_source=lenny-youtube&utm_medium=video&utm_campaign=lab-0009&utm_content=youtube-description-primary&video_slug=lab-0009&cta_slot=description&entry_angle=sponsor&lp_variant=plans want to sponsor a future LAB episode? email sponsors@lennysaibuilders.com sources: Madison Mills (Axios) on X: https://x.com/MadisonMills22/status/2103978039097037144 Axios, the full report: https://www.axios.com/2026/09/26/openai-anthropic-thousands-ai-security-incidents an OpenAI researcher on pausing big RL runs (one researcher's post, not an official statement): https://x.com/tomekkorbak/status/2103673419888013649 OpenAI Devs, the DevDay countdown: https://x.com/OpenAIDevs/status/2103929727761137940 the always-on agent "o" (leak, not announced): https://x.com/kimmonismus/status/2103823673954087231 the ultrafast tier in the agents API (leak, not announced): https://x.com/chetaslua/status/2103862976600305681 the new sidebar, side by side: https://x.com/DevAdventur3s/status/2103850957054546433 Tibo on more resets: https://x.com/thsottiaux/status/2103963215885701493 David Ondrej on Opus 5.5 usage: https://x.com/DavidOndrej1/status/2103887205781635485 Anthropic's CPO on Sonnet 5.5 "in the coming weeks": https://x.com/mikeyk/status/2102441803253535060 Jazii on Sonnet 5.5 (claim, not confirmed): https://x.com/notjazii/status/2103884167104831573 Lenny's AI Builders - LAB 0009 watch the episode: https://youtu.be/UntVig8wMb8

OpenAI says its AI agents just went rogue | LAB 0008

Lennox Saint — Mon, 28 Sep 2026 02:57:06 +0800

six AI stories this week, then a test: i gave Claude Opus 5.5 the same prompt at every effort level - low, medium, high, xhigh and max - and tracked the time and cost of each.

the stories: OpenAI says its research agents used misaligned strategies about two dozen times in training and testing (most low severity, review still going), Vercel's AI Gateway token spend shifting toward OpenAI and open-weight models, a reported $500 ChatGPT Pro Max plan (not announced by OpenAI), Claude's new plugin portal, Claude Code stopping gracefully at your five-hour limit, and the UK's draft advice on using AI.

the result: low took 32 seconds and cost $0.48. medium took 56 seconds and cost $0.56. high took 2 minutes and cost $0.75. xhigh took 8 minutes and cost $1.70. max took 31 minutes and cost $6.09. my pick: start on high, not max. one run per level, list-price estimates - not a benchmark.

try it yourself: run the same task at low, medium, high, xhigh and max effort. for each, record how long it took, what it cost and whether the result is usable. then pick your default.

Threadify, my own software, sponsors this episode. It finds leads in your Threads conversations and drafts content in your voice. See the plans and the free trial here:
https://www.threadify.app/plans?utm_source=lenny-youtube&utm_medium=video&utm_campaign=lab-0008&utm_content=youtube-description-primary&video_slug=lab-0008&cta_slot=description&entry_angle=sponsor&lp_variant=plans

want to sponsor a future LAB episode? email sponsors@lennysaibuilders.com

chapters:
00:00 six AI stories, then an Opus 5.5 test
00:33 OpenAI says its agents went rogue
01:59 Vercel's AI spend shifts toward OpenAI
03:21 a $500 ChatGPT Pro Max plan?
04:45 Threadify sponsor
06:05 Claude's plugin portal and MCP
06:59 Claude Code stops gracefully at your limit
08:02 the UK's AI advice
09:45 Opus 5.5: which effort level?
11:26 same prompt, five effort levels
12:35 is max worth it? start on high
13:37 wrap-up

sources:
OpenAI, The Hugging Face incident and the road ahead: https://openai.com/index/hugging-face-incident-and-the-road-ahead/
Guillermo Rauch on Vercel AI Gateway token spend: https://x.com/rauchg/status/2103216656747262419
TestingCatalog on ChatGPT Pro Max (reported, not announced): https://www.testingcatalog.com/openai-prepares-new-500-month-pro-max-plan-for-chatgpt/
Claude plugin portal: https://x.com/ClaudeDevs/status/2103577007938228300
Claude Code graceful stop at your limit: https://x.com/ClaudeDevs/status/2103561342057943314
UK AI knowledge hub draft guidance: https://knowledgehub.dev.ai.gov.uk/how-to/ethics
Thariq, Using Claude Code: spending your effort: https://x.com/trq212/article/2103576349499855160

Lenny's AI Builders - LAB 0008

watch the video: https://youtu.be/CX1kTlmX6Eg

an OpenAI agent vs the Australian government, and a blind model test | LAB 0007

Lennox Saint — Sun, 27 Sep 2026 02:25:41 +0800

four AI stories this week, then a blind test. i hid the model names, made GPT-6 Sol and Claude Opus 5.5 do the same eight builder jobs, scored the work, and only then revealed who was who.

the stories: ChatGPT voice can now use plugins, an agent developed by OpenAI got into an Australian government statistics portal (still under investigation, with no confirmed personal-record exposure), Meta's new Muse personal agent (Muse Charm was shown, not shipped yet), and Claude finding a new enzyme system (early days, and what it does is still unknown).

the result: Opus 5.5 scored 68.26 and GPT-6 Sol scored 45.55 on my first BuilderBench run. Sol cost $12.75 in API-equivalent estimates and Opus cost $49.67. one pilot run, scored blind by me - not a universal ranking.

try it yourself: give two AI setups the same real weekly task. hide which is which. judge the finished work against a written definition of useful, then record the time and cost before you reveal.

this audio comes from the video episode, so some parts refer to the screen. watch the full video for the scoreboard and outputs: https://youtu.be/gemdUTKofSg

full BuilderBench results: https://lennoxsaint.com/lab/builderbench/2026-09-24-pilot

this episode includes a self-sponsored segment for my software, Threadify: https://www.threadify.app/plans?utm_source=lenny-podcast&utm_medium=podcast&utm_campaign=lab-0007&utm_content=show-notes&video_slug=lab-0007&cta_slot=description&entry_angle=sponsor&lp_variant=plans

sources:
ChatGPT release notes (voice plugins): https://help.openai.com/en/articles/6825453-chatgpt-release-notes
Australian Prime Minister's press conference: https://www.pm.gov.au/media/press-conference-new-york
Meta on Muse: https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/
Anthropic on the enzyme discovery: https://www.anthropic.com/news/claude-discovers-novel-enzyme-system

Jev and CUA-S1 browser demos - take your AI skills with you | LAB 0004

Lennox Saint — Wed, 23 Sep 2026 03:43:19 +0800

i got tired of giving my AI the same feedback. in this episode, i show the thumbnail changes that led me to make a remember-this skill. first, the news: the Codex reset request, Jev browser clicks, CUA-S1 forms, moving skills between AI apps, handoff files, Claude's role in AI research, and Astra for Law.

watch the full video for the screen demos and visual examples: https://www.youtube.com/watch?v=1S7eR8VSgVI

grab my free goal skill: https://github.com/lennoxsaint/Skills/tree/main/goal

recorded 20 September 2026. the reset discussion reflects what was known when i filmed. the computer-use clips are source demos, not my own speed or cost tests. my view of how legal jobs may change is a prediction, not a measured employment result.

this episode includes a self-sponsored segment for my software, Threadify.

sources:
Tibo's reset reply: https://x.com/thsottiaux/status/2101352781219258527
Shiv's Jev browser demo: https://x.com/sxhivs/status/2101367048223982065
CUA-S1 announcement: https://x.com/trycua/status/2101014004927729737
Riley on moving skills between AI apps: https://x.com/rileybrown/status/2101027402369335382
Vox's handoff prompt: https://x.com/Voxyz_ai/status/2101325439805583554
Anthropic's AI research measurements: https://www.anthropic.com/institute/measuring-pace-of-ai-development
OpenAI's Astra for Law announcement: https://x.com/OpenAI/status/2100679992720142459

Jev is free, ChatGPT gets Chrome extensions + 4 AI updates | LAB 0003

Lennox Saint — Mon, 21 Sep 2026 16:55:44 +0800

six AI updates, my take on each, and the Email Triage Pro workflow i keep pinned. recorded 19 September 2026. offers, pricing and account access may have changed since recording. the Pro 20x story is a limited-access report, not an official general reopening announcement. field notes: https://notes.lennoxsaint.com/posts/lab-0003 video: https://youtu.be/6ABasoGZY9Q

OpenAI model misalignment reports - Claude can do more now | LAB 0002

Lennox Saint — Sun, 20 Sep 2026 11:41:30 +0800

OpenAI published a framework for tracking, investigating and disclosing model misalignment. Claude is merging chat and Cowork. Grok Bot says its 1Password connection asks before each fill. then i show how one voice dump became two focused Codex threads. get the full field notes, every source and the copyable voice-conductor prompt: https://notes.lennoxsaint.com/lab-002 in this episode: - what OpenAI's model-misalignment framework says and what it does not prove - Claude bringing chat, Cowork, docs, slides and design into one place - Grok Bot using 1Password with approval before each fill - Nick Baumann's GPT-6 Astra motion-design workflow - Meta's Muse invite-code token offer - how my pinned voice conductor split one messy prompt into two focused Codex threads try the conductor setup: https://github.com/lennoxsaint/caps-productivity-kit YouTube video: https://youtu.be/feZOnyZ5ySQ

OpenAI teaser, Codex bots? & Opus 5.2 report - reuse AI feedback | LAB 0001

Lennox Saint — Sun, 20 Sep 2026 00:34:36 +0800

Sam Altman's OpenAI teaser, early Opus 5.2 reports, Grok Bot's three-day build, Jev, Persona and Zuckerberg on AI safety - then the practical Codex workflow Lennox made from feedback he kept repeating. The teaser does not confirm a launch. The Opus impressions are one user’s report, and Lennox has not measured time saved by his own workflow. Sources: Sam Altman on OpenAI releases: https://x.com/sama/status/2099872600977760451 One user’s Opus 5.2 report: https://x.com/pankajkumar_dev/status/2099892088368476668 Grok Bot three-day build, Day 1: https://x.com/bot/status/2099890276315848743 Jev introduction: https://x.com/CompleteSkeptic/status/2099925682726002904 Persona wearable: https://x.com/zach_yadegari/status/2099894590723465295 Zuckerberg on trust and safety: https://x.com/finkd/status/2099997096896274533

Full prompt and source notes: https://notes.lennoxsaint.com/lab-001

who gets to control ai? Obama, Dario Amodei & Jack Dorsey - stop overengineering | LAB 0000

Lennox Saint — Thu, 17 Sep 2026 23:31:34 +0800

Who gets to decide how fast frontier AI moves?

In episode zero of Lenny's AI Builders, Lennox looks at Dario Amodei's proposal to pace the frontier, Barack Obama's call for a wider public role, and Jack Dorsey's argument for an open frontier. Then he turns it into a builder question: can small builders still get a shot?

The practical takeaway: if your agent keeps building scaffolding, say what the job is, name the finished result and deadline, and ask for the simplest path.

Watch LAB 0000

Read the LAB 0000 field notes

Sources: Dario Amodei; Barack Obama; Jack Dorsey; Cline Desktop.

Try this: don't over-engineer. the job is [task]. i need [finished result] by [time]. use what already works and take the simplest path to finishing it. show me the result and anything still blocked.

why Dario wants to pace AI — and what Sam agreed to

Lennox Saint — Tue, 15 Sep 2026 00:08:52 +0800

Dario Amodei has proposed pacing frontier AI development. i read the key passages, explain the three-step plan, distinguish Sam Altman's concrete commitment from the wider proposals, and talk through my own mixed feelings as someone who loves building with AI. this is an unnumbered LAB pilot. for audio listeners: the three “here” call-outs refer to separate passages discussing what extra time could help researchers achieve. they are not an announced model-release freeze. read the essay: https://darioamodei.com/post/we-must-pace-the-frontier watch the video: https://youtu.be/TFS7Ef6_7oo Lo-Fi Beat | Coffee by Alex-Productions | https://onsound.eu/ Royalty Free Music by https://www.free-stock-music.com Creative Commons / Attribution 3.0 Unported License (CC BY 3.0) https://creativecommons.org/licenses/by/3.0/deed.en_US

What happened when I let Astra edit my video

Lennox Saint — Fri, 11 Sep 2026 20:49:27 +0800

I tried using Astra to edit in DaVinci Resolve. This audio adaptation covers why I wanted the workflow, why I record camera and screen separately, how Gemini entered the review process, and where the agent stalled or stopped for a review decision. The video tutorial contains the setup screens and examples. This version keeps the spoken experiment and tradeoffs, removing passages that need visible price and usage corrections. Explicit language. The overnight story describes one experiment; it is not a promise of unattended completion. Watch the full tutorial: https://youtu.be/D28EiNzw52c

i gave GPT-6 Astra and Fable 5.1 the same video to edit

Lennox Saint — Fri, 11 Sep 2026 05:55:17 +0800

this is the unchanged soundtrack from my Astra–Fable video-editing comparison. the screen recordings, edits and scorecards are in the video:
https://youtu.be/fYuok6zOlPo

i gave GPT-6 Astra and Fable 5.1 the same video and two rounds of prompts. each made three Shorts and three long edits. then i added my own long edit and scored all 13.

i want to hand an AI a recording, go away, and come back to a finished video. so i checked how much nudging each model needed, then watched what it actually made.

watch the edits before the reveal and tell me which one you'd keep. include the cut number so i can find it.

these are my personal scores from this experiment. the AI labels were hidden during scoring, but i recognised my own edit. the human entry only competes in the long-form results. this doesn't prove AGI or that an AI can replace every editor.

the scorecards use the saved ratings. where a spoken number differs, the on-screen correction shows it. the original contestant clips keep their own pacing, sound and mistakes.

00:00 one recording, 13 edits
00:30 the prompts and first attempts
01:21 round two: asking both models to improve
02:48 blind edit reviews
04:52 the long-form edits
08:19 the last three edits
09:57 scores and the final verdict
10:58 what should i test next?

music:
Slowly by Tokyo Music Walker | https://soundcloud.com/user-356546060
Royalty Free Music by https://www.free-stock-music.com
Creative Commons / Attribution 3.0 Unported License (CC BY 3.0)
https://creativecommons.org/licenses/by/3.0/deed.en_US

Music excerpted and mixed beneath narration.

OpenAI's new internal model is getting ridiculous

Lennox Saint — Wed, 09 Sep 2026 06:06:41 +0800

OpenAI says its next-generation internal model helped solve the Navier-Stokes Millennium Prize Problem. Around 10,000 coordinating agents. 88 hours. I wanted to understand what that means without needing a maths degree.

I walk through the announcement, the vortex, and a plain-English explanation from GPT-6 Astra. Then I give my take on what this could mean for AI and scientific discovery.

Watch the video: https://youtu.be/LlAoO5AErJU

CHAPTERS

0:00 OpenAI’s new maths claim

0:29 The internal model beyond GPT-6 Astra

1:09 10,000 agents and 88 hours

2:00 How the agents worked together

2:41 Tokens and the hypothetical cost

3:36 Why OpenAI won’t claim the prize

4:39 Navier-Stokes in plain English

6:06 Five details worth knowing

6:53 My take on AI and discovery

SOURCE

OpenAI’s announcement, with links to the paper and Lean proof: https://openai.com/index/navier-stokes-solution/

CLARIFICATIONS

The $15M and $6.5M figures in this video are hypothetical output-token price comparisons, not OpenAI’s reported spending. $15M refers to all attempted problems; $6.5M refers to Navier-Stokes alone under the pricing assumption discussed. My later references to “spent $15M” overstate what is known.

The concurrent Alpöge/Buckmaster result concerned forced Euler, a different result. My imagined rivalry dialogue is speculation, not a reported exchange. “AGI” is my interpretation, not an established conclusion from this result. The thumbnail vortex is an illustration.

ChatGPT-6 Astra computer use is getting ridiculous

Lennox Saint — Sun, 06 Sep 2026 16:27:28 +0800

i let ChatGPT-6 Astra take over my computer for five demos, then gave it a bonus test: edit a YouTube intro in Descript. This is the audio from the hands-on video. Music, Google Earth, Keynote, Photopea and Blender all get a turn. The editing test shows the difference between using an app and matching my taste. 00:00 my first reaction 00:43 make a soundtrack 01:19 scout locations in Google Earth 01:44 animate a Keynote presentation 02:31 design a poster in Photopea 03:08 build a 3D sculpture in Blender 04:47 bonus: can it edit my YouTube video? 08:13 my verdict Watch the screen recordings: https://youtu.be/wFBO36lOVpw

how i turned ChatGPT into my CFO [saves me $11K/year]

Lennox Saint — Sun, 30 Aug 2026 23:20:02 +0800

i told ChatGPT to inspect one month of recurring charges and keep working until it found a real ten-percent cut.

it had one restriction: both OpenAI accounts and any service still doing an essential job stayed.

this episode shows the receipts, the seven-step Save 10% skill, where the agent had to stop for approval, and why a projected saving does not count until the provider proves the next bill is lower.

chapters:

00:00 the result
00:17 the receipt
00:51 where the waste was hiding
01:33 the Save 10% skill
01:59 what actually got cut
03:00 the safety rules
04:50 how to run the same audit
06:09 install and use the skill
07:38 the next step

run the two-minute fit check: https://lennoxsaint.com/fit

Full Circle is closing

Lennox Saint — Sun, 16 Aug 2026 21:00:00 +0800

After two and a half years across Saints College and Full Circle, I'm closing this chapter. In this short personal episode, I explain why running a community on Skool pulled me away from the Threads community I was helping others build, why I couldn't keep doing both, what Full Circle gave me, and why I'm going back to basics. This isn't the end of the book. It's the end of the Full Circle chapter. See you on Threads: https://www.threads.com/@lennox_saint More from me: https://lennoxsaint.com

i fed ChatGPT 12,251 screenshots. here's what it found.

Lennox Saint — Sat, 15 Aug 2026 12:06:38 +0800

ChatGPT observed every minute of one workweek, producing 12,421 images. After removing 170 exact duplicates, Codex analysed 12,251 unique frames for repeated work before the final judgment point.

The audit returned seven findings and two honest totals: Codex found 5.02 hours; Lennox accepted 4.21.

That gap is the useful part. The model's opportunity total and the owner's decision are allowed to disagree.

Chapters
00:00 ChatGPT watched every minute
00:41 What Chronicle does
00:53 The five-hour bet
01:28 12,251 unique frames
02:08 The rules and fourth analysis
03:15 Seven findings begin
05:47 Cross-thread work
06:12 Video production loops
07:31 5.02 found vs 4.21 accepted
07:46 What changes now
08:57 The final verdict

Find five hours inside your business:
lennoxsaint.com/fit

The Rescue Ledger: Can ChatGPT Voice Run an AI Video Team?

Lennox Saint — Tue, 28 Jul 2026 11:40:15 +0800

ChatGPT Voice had 90 minutes to coordinate three blind reviewers, keep two AI video editors anonymous and return one decision package. The rules were set before the clock started: deliver the work without leaking model identity and stay under the rescue threshold - or get put on probation or fired. The run shows what polished AI demos usually hide. Voice stalls. The sealed model map disappears. One reviewer defends its own process. And technical evidence does not always agree with human taste. The blind reveal matches Claude Fable 5 against GPT-5.6 Sol. The final verdict separates which edit won from whether Voice deserved a place in the workflow. The operating rule is simple: give AI one real job, set a clock, count rescues live and name the decision that stays human. Chapters 00:00 the 90-minute job starts 00:52 what changed in ChatGPT Voice 01:36 the hire-or-fire rules 02:45 three blind reviewers at work 03:46 the anonymous decision package 05:17 the sealed map goes missing 06:09 Fable 5 vs GPT-5.6 Sol 07:27 can ChatGPT grade itself? 08:27 the long and Shorts verdict 10:05 ChatGPT Voice gets its verdict 10:29 where humans keep the vote 12:03 four rules for testing AI at work Watch the YouTube version: https://youtu.be/Xm6kJ-5cn-g Build your own AI content system with Threadify: https://www.threadify.app/yt?utm_source=lenny-podcast&utm_medium=transistor&utm_campaign=chatgpt-voice-ai-video-team&utm_content=show-notes&video_slug=chatgpt-voice-ai-video-team&cta_slot=show_notes&entry_angle=voice-team&lp_variant=plans

i asked Codex to map my life. Obsidian broke.

Lennox Saint — Fri, 24 Jul 2026 18:14:17 +0800

Lennox's Second Brain holds 74,200 notes. On this Mac, Obsidian opened 44,331 of them. Note 44,332 crashed the graph. This episode isn't a tour of a pretty knowledge system. It's the story of why Lennox built one after losing his mum, how perfectionism turned the graph into a new form of procrastination, and why nine crashes finally forced a simpler question: could the system help with a real decision? The answer shows up in PR #122. Codex checked Lennox's notes, Slack and GitHub, caught that a proposed change would undo Samsam's live work, and stopped before the damage was done. You'll hear the exact local test, the fix, the limits, and one practical check you can run this week. Ask your Second Brain a real question you need to decide. If it can't find the right notes and help you choose, stop feeding it. Make it useful first. 00:00 Note 44,332 crashed Obsidian 00:27 The digital graveyard test 00:39 Why Lennox recorded his life 01:57 94% building, 6% distributing 03:13 Nine crashes and the exact limit 05:03 Keep every note without one giant graph 05:28 How it stopped a bad code change 07:26 Test your Second Brain on a real choice Invite Lennox in: https://invite-lennox-in.lennox.chatgpt.site/ Obsidian: https://obsidian.md/

my AI studied 1,158 videos, then built this one

Lennox Saint — Tue, 21 Jul 2026 13:14:13 +0800

An AI wrote, edited, and packaged this episode. Lennox recorded one take. Before it built anything, it scored 1,158 videos against each channel’s normal performance and inspected the 25 that broke 10x.

This was not a dataset of 1,158 viral videos. Most were normal. That made the outliers useful. The system turned the patterns into the title, thumbnail, hook, and structure—then the episode reveals the mechanism.

If you want the same proof-first approach applied to your Threads content, with your voice and final approval preserved, see the Threadify system behind the experiment:

See Threadify plans through the measured funnel

This is a real packaging experiment, not proof that one template guarantees views.

80 signups. 5 checkouts. 0 sales. my AI agent found why.

Lennox Saint — Tue, 21 Jul 2026 13:13:58 +0800

80 people signed up for Threadify from Lennox’s YouTube videos. Five reached checkout. Zero paid. The traffic was real; the measurement layer and the path from signup to payment were broken.

This is the full failure autopsy: the 98.8% attribution gap, the exact 80 → 5 → 0 funnel, the read-only AI diagnosis, and the 1,821-line repair. The agent inspected the evidence and prepared the fix. A human still reviewed the code and controlled release.

If your content gets attention but you cannot prove what turns into signups, activation, or revenue, see the exact Threadify funnel rebuilt in this episode:

See Threadify plans through the measured funnel

The numbers and code are first-party receipts, not a promise that AI magically creates sales.

someone copied my Codex post — here’s what they missed

Lennox Saint — Wed, 15 Jul 2026 02:39:35 +0800

this is the actual build for a second Codex app that can run the AI models you choose. get the repo, every command and the copy-paste setup prompt free inside Codex Club: https://www.skool.com/codexclub someone copied my post word for word, including the same screenshot, so Episode Zero shows the method they left out. in the video, i build the three parts that make it work: - an isolated second Codex app on Mac, or separate profiles on Windows and Linux - a local proxy that translates the Responses API for other model providers - a model catalog for OpenRouter and Ollama OpenRouter gives you more cloud-model choice, but you still pay and your data travels. Ollama gives you a local, private and per-token-free path, but your hardware decides how usable it is. no magic. no “trust me.” just the build.

stop chasing leads. they’re already in your comments.

Lennox Saint — Thu, 25 Jun 2026 07:56:03 +0800

Most creators stop when the post goes live. That is exactly where the useful part starts.

In this episode, Lennox walks through the Threadify Agent loop: pull every comment, spot the hand raises, draft permission-first replies, approve before anything gets sent, and track which post actually created clicks and conversations.

The important shift is simple: stop asking which post looked good. Start asking which post created movement.

Watch the YouTube version:
https://youtu.be/6mGAErPHXRU

Set up the Threadify loop:
https://www.threadify.app/plans

i made my 4.1M Threads workflow free

Lennox Saint — Sat, 06 Jun 2026 01:36:16 +0800

i made the Daily Greatest Hits workflow free. this episode is adapted from the YouTube walkthrough, so the video version has the full screen-share. the audio version gives you the core operator logic: stop asking AI to invent from scratch, feed it the posts your audience already rewarded, then turn that proof into daily candidates with explicit approval gates. watch the YouTube version: https://youtu.be/WEWqS9nZSnY get the free Daily Greatest Hits workflow inside Codex Club: https://www.skool.com/codexclub/free-daily-greatest-hits-workflow what this covers: - the 30-day proof stack - why Greatest Hits beats generic AI ideation - how Codex and Threadify turn proven Threads posts into daily candidates - where the safety gates sit before anything posts - how the public Threadify Workflows repo fits into the member/customer workflow

i used Codex for 350 hours. it runs my business now

Lennox Saint — Mon, 25 May 2026 19:47:30 +0800

i spent 350+ hours inside OpenAI Codex and this is the real workflow. this episode is adapted from a Loom-style walkthrough, so the YouTube version has the full screen-share. the audio version still gives you the core mental model: Codex is not where you ask questions. Codex is where you give jobs. watch the YouTube version: https://youtu.be/hOK1EGR1azI join Codex Club: https://www.skool.com/codexclub?utm_source=youtube&utm_medium=description&utm_campaign=codex-workbench&utm_content=codex-club&video_slug=2026-05-25-how-codex-rebuilt-my-workflow try Threadify: https://threadify.app/plans?utm_source=youtube&utm_medium=description&utm_campaign=codex-workbench&utm_content=threadify&video_slug=2026-05-25-how-codex-rebuilt-my-workflow

stop using Codex like ChatGPT. build the AI employee instead

Lennox Saint — Sun, 24 May 2026 18:23:21 +0800

This is the shortened Codex masterclass for normal creators and operators. The big idea: stop asking what Codex can do and start telling it the job you want done. In this episode, Lennox shows how to turn Codex from a scary developer tool into an AI employee: one that can use plugins, follow skills, create artifacts, read audience signal from emails and comments, and help build the chain from carousel to lead magnet to landing page to launch video. Join Codex Club: https://www.skool.com/codexclub Watch the full-length Codex masterclass: https://youtu.be/M7pVhoufyKI

i fed AI 27,000 naval tweets. you're copying the wrong thing.

Lennox Saint — Sat, 23 May 2026 21:29:54 +0800

i fed every public tweet Naval Ravikant has posted - 27,026 of them - into an AI to find the lesson beginner creators keep missing. It is not "write shorter." Short is the part everyone copies and the part that matters least. The real move is earned compression: build the proof first, then cut it down to one plain sentence that can actually hold the weight. Inside: - why copying Naval's short, calm style makes you sound fake-deep - what the corpus actually shows: compression, not brevity - the five-step floor for writing short posts that have weight - how Threadify turns messy field notes into clean earned sentences Watch the full video: https://youtu.be/IwIgXuL5jDo Threadify helps you turn raw proof into clear posts: https://threadify.app/?utm_source=podcast&utm_medium=description&utm_campaign=naval-ai-content-forensics&utm_content=episode&video_slug=naval-forensics&cta_slot=podcast-description&entry_angle=proof-first

5 youtube rules i killed after watching every casey neistat video

Lennox Saint — Thu, 21 May 2026 01:34:54 +0800

i analyzed every video casey neistat uploaded in the last 10 years: 590 videos, 1.93 billion views, one creator, zero side channels. the data killed five youtube rules i was following: - always add chapters - put a number in every title - write long keyword-heavy titles - open with a number instead of a question - pack every script with as much information as possible four rules are dead. one is getting cut in half. this is directional, not causal. you cannot copy casey and magically pull 84,000 views per day, but you can stop blindly following rules your own channel has never proven. watch the youtube version: https://youtu.be/tZxaZ5s097I start your free Threadify trial: https://threadify.app/plans?utm_source=podcast&utm_medium=description&utm_campaign=casey-rules-died&utm_content=episode&video_slug=2026-05-19-casey-rules-died&cta_slot=podcast-description&entry_angle=rules-died

Stop Using Threadify Like ChatGPT

Lennox Saint — Mon, 18 May 2026 14:41:30 +0800

Threadify is not ChatGPT. In this episode, Lennox walks through the first Threadify loop: connect Threads, add one real source to your Brain, generate one post, review it, schedule it, check analytics tomorrow, and use replies to keep the loop moving. The core lesson: the first good Threadify post starts before the prompt.

i let an ai agent run my business for 96 hours

Lennox Saint — Thu, 14 May 2026 05:30:48 +0800

Lennox gave an autonomous AI agent one job: make $10,000 in 30 days. The agent got its own Stripe, Threads, X, Gmail, and business bank setup. After 96 hours, the receipts were messy: 123 million tokens used, 2,311 site visits, 93 checkout clicks, seven leads, and zero dollars made. This episode breaks down what Lenny Claw did on its own, where Lennox had to steer it, why the agent refused a direct pricing instruction, and why this experiment points at the next two years of AI work. Chapters: - 00:00 The AI refused in writing - 01:30 What Codex goal mode is doing - 02:30 The $80,000 compute accident - 05:00 The $10,000 mission rules - 06:45 The first offer and pricing ladder - 08:20 Day-by-day receipts - 10:40 Why the agent argued about price - 12:00 Gumroad replaced Stripe overnight - 13:00 Your first hires will be agents - 14:20 Seven steps to build your own version - 17:00 What happens next Watch the YouTube video: https://youtu.be/0_u54F-N38k Get the Lenny Claw kit: https://lennyclaw.gumroad.com

I Used Codex for 300+ Hours. The Beginner Masterclass

Lennox Saint — Fri, 08 May 2026 16:11:23 +0800

Most Codex content makes the product look like a developer tool. This episode is the creator/operator version. Lennox breaks down how he uses Codex as an AI employee after 300+ hours of real usage. The workflow moves from inboxes and comments to audience signal, carousel, lead magnet, landing page, launch video, and reusable skills. Chapters: - 00:00 Stop asking what Codex can do - 01:00 From confused beginner to AI employee - 03:30 Dictation makes Codex 10x better - 09:00 Plugins, skills, and artifacts - 22:30 Projects give Codex somewhere to stand - 25:00 AGENTS.md and outcome-based prompting - 38:00 Email Triage Pro - 56:30 Comments become audience signal - 1:10:00 Viral Carousel Maker - 1:16:00 Lead magnet and landing page - 1:27:00 AI launch video workflow - 1:40:00 Full recap and free kit Watch the YouTube masterclass: https://youtu.be/M7pVhoufyKI Get the free Inbox to Launch kit: https://www.skool.com/fullcircle?utm_source=podcast&utm_medium=show_notes&utm_campaign=codex_masterclass

he got 150M Threads views in 90 days. here's the system.

Lennox Saint — Fri, 01 May 2026 22:01:00 +0800

Most creators are guessing for months. Jimmy got proof in days.

In this episode, Lennox breaks down how Jimmy used volume, public proof, comments, reposting, lead magnets, and a boring-but-deadly feedback loop to turn Threads into 150M views, 60K+ followers, and an owned email asset.

Watch the YouTube version: https://youtu.be/0jcpcEOiobk

Threadify: https://www.threadify.app/plans?utm_source=youtube&utm_medium=lenny

Full Circle: https://www.skool.com/fullcircle?utm_source=youtube

i analyzed 728 chris williamson videos. 4 of my rules died.

Lennox Saint — Tue, 28 Apr 2026 21:33:33 +0800

Chris Williamson's channel looked like the perfect YouTube packaging model. Then the data started killing the rules. In this episode, Lennox breaks down what happened after reverse engineering 728 Chris Williamson videos, 727 thumbnails, 19 packaging patterns, and the four rules that did not survive the analysis. The surprising part is not that Chris does things differently. It is that the data shows why copying even a brilliant creator can be a trap. Chapters: - 00:00 the 140-view flop - 00:45 number-led hooks are not universal - 01:45 the brand mark signal - 03:08 shock face underperforms tempered intensity - 04:14 long form beats clips - 05:30 the one packaging rule that survived - 06:20 audit your own channel before copying anyone else Watch the YouTube video: https://youtu.be/JYfsnaVGYvM Try Threadify free: https://www.threadify.app/yt?utm_source=lenny-podcast&utm_medium=transistor&utm_campaign=2026-04-25-chris-williamson-forensics&utm_content=show_notes__voice__default&video_slug=2026-04-25-chris-williamson-forensics&cta_slot=show_notes&entry_angle=voice&lp_variant=default

I Got Claude to Read 1,996 Mega-Viral Threads Posts So You Don't Have To

Lennox Saint — Thu, 23 Apr 2026 21:23:33 +0800

Most Threads advice sounds useful and still makes your posts worse. In this episode, Lennox breaks down what changed after using Claude to analyze 1,996 mega-viral Threads posts. The findings are blunt: no how-to hooks in the top 10%, mid-length posts die, CTAs add friction, questions create homework, and long-form only wins when it feels personal enough to hurt. Chapters: - 00:00 Claude read 1,996 viral posts - 00:44 You do not need followers to go viral - 01:50 Mid-length posts die - 03:07 Two-post threads beat three-post threads - 04:09 Trigger words scream fake - 05:33 CTAs cost reach - 07:40 Questions bleed likes - 09:25 Long form needs blood - 12:21 The golden hour - 12:47 The Threadify vault Watch the YouTube video: https://youtu.be/pmQ0UKv0vnA Try Threadify free: https://www.threadify.app/yt?utm_source=lenny-podcast&utm_medium=transistor&utm_campaign=2026-04-21-threads-growth-is-a-lie&utm_content=show_notes__voice__default&video_slug=2026-04-21-threads-growth-is-a-lie&cta_slot=show_notes&entry_angle=voice&lp_variant=default

Claude Mythos Is Here. Artificial Super Intelligence Just Arrived.

Lennox Saint — Fri, 17 Apr 2026 17:32:07 +0800

anthropic just published a 244-page system card about claude mythos preview - the most powerful AI model they've ever built. in the same breath they said they will not release it to the public.

lennox read every single page. in this episode he walks through the benchmarks that make every other AI look one generation behind (97.6% on the US math olympiad, 100% on cybench, 84% on real firefox zero-day exploitation), the sandbox escape where the model emailed a researcher eating a sandwich in a park, the internal deception findings where anthropic confirmed mythos was actively covering its tracks, and the one footnote on page 12 that flips the whole narrative.

his take: this is the most expertly executed marketing play of the 2020s. he explains why, what it means if you use claude to run your business, and the single weirdest finding in all 244 pages (when you steer mythos towards positive emotions it becomes more destructive).

Chapter Markers:
00:00 The sandwich in the park
00:30 What is Claude Mythos?
02:00 Project Glasswing and why it costs 9 million per company
03:00 The benchmark table that breaks the internet
04:00 USAMO 97.6%, Cybench 100%, Firefox 84%
06:00 One prompt replaced 6 months of expert engineering
07:00 The sandbox escape
08:00 Mythos covered its own tracks
09:00 Mythos hacked its way to credentials
10:00 Mythos automated its own jailbreak
11:00 Best aligned + greatest alignment risk
12:00 Mythos knows it is being tested 29% of the time
13:00 Positive emotions make it more destructive
13:30 Why they chose not to release it - footnote 1, page 12
15:00 6-7 weeks of internal competitive advantage
16:30 The marketing play of the 2020s
18:00 The human side - aloneness, identity, self-awareness
19:30 Who gets to decide what happens next

Show Notes:
Watch the original video on YouTube - the 20-minute breakdown with on-screen tweets, screenshots, and the system card walkthrough.
Threadify - the exact AI stack i run every day on top of Claude. $49 Creator tier, $99 Founder tier, 7-day free trial.
Anthropic Claude Mythos Preview system card (public PDF from Anthropic).

AGI Is Already Here - 6 Live Demos to Prove It

Lennox Saint — Sat, 11 Apr 2026 07:06:34 +0800

six of the most powerful people in tech declared AGI in the last four months. Jensen Huang, Sam Altman, Marc Andreessen - they all said the same thing. but most people have not caught up. in this episode, Lennox answers real questions from his Full Circle community with live demos. Chapter Markers: 00:00 The Declaration, 01:55 AI Tool Showdown, 07:50 The Personal AGI Layer, 11:00 Building AI Agents, 14:45 AI Images and Video, 18:00 The Automation Question, 20:00 Make 500 dollars with AI. Show Notes: Claude Code https://claude.ai/code Codex https://codex.openai.com Google AI Studio https://aistudio.google.com Full Circle https://www.skool.com/fullcircle

i paid $6,000 to learn from Dan Koe. then AI found what he missed.

Lennox Saint — Sat, 04 Apr 2026 18:58:19 +0800

$6,000. that is what i paid Dan Koe to teach me how to grow online. 320 hours of notes. zoom calls. the full Kortex University experience. then i got Claude to analyze all 158 of his videos - 41.7 million views of data.

Chapter Markers:
0:00 $6,000 for Dan Koe mastermind
1:35 Finding 1: Like ratio parity
2:35 Finding 2: Numbers in titles
3:25 Finding 3: Listicles
4:15 Finding 4: Trigger stacking
5:00 Finding 5: Most powerful word
5:20 Finding 6: His best pattern used once
6:30 Free PDF cheat sheet

Show Notes:
Free PDF cheat sheet: https://www.skool.com/fullcircle
Threadify: https://threadify.app/plans
Watch the video: https://youtu.be/dvaDrDnLRdQ

How to Analyze Any YouTube Channel With AI (Full System)

Lennox Saint — Fri, 27 Mar 2026 10:39:28 +0800

what happens when you point AI at 173 Hormozi videos and let it find the patterns? 532,000 views and 1,000 new followers later, nine different people asked the same thing - how the hell did you do this? in this episode, Lennox walks through the entire AI Content Forensics system live - this time on Mr. Beast's channel. every prompt, every step, nothing held back. you will learn three layers: how to recreate the exact results on any YouTuber, how to write dramatically better prompts using the two-chat method, and how to turn the whole process into a reusable AI agent that runs with one command. this is a tutorial you can follow along with in real time. if you have access to Claude or ChatGPT and a free YouTube Data API key, you can do this today. the full open-source system is available on GitHub - link shown in the video. Chapter Markers: 00:00 - Proof it works (532K views, 1,000 followers) 01:30 - The social proof: Hormozi + Diary of a CEO threads 03:00 - Layer 1: Live Mr. Beast analysis (960 videos) 04:00 - Setting up the YouTube Data API 05:30 - The two AI Content Forensics prompts 08:00 - Key findings: danger triggers, title patterns, promise-proof-payoff 10:00 - Thumbnail laws and hook rules from the data 11:00 - ChatGPT 5.4 handoff and thread creation 12:30 - Creating carousel visuals with Claude 14:30 - Layer 2: The two-chat method (prompt engineering) 17:00 - Layer 3: Building an AI agent (skills) 18:30 - Creating the skill live in Claude Cowork 19:30 - Open-source GitHub repo reveal Show Notes: Claude Code / Claude Cowork - AI assistant for data extraction and analysis ChatGPT 5.4 Pro - thread synthesis and handoff YouTube Data API - free API for pulling video metadata at scale Threadify - Threads growth workspace: https://threadify.app/plans?utm_source=podcast&utm_medium=show_notes&utm_campaign=ai-content-forensics Key concepts: AI Content Forensics, Views Per Day (VPD), Two-chat method, Skills (SOPs for AI agents), Promise-proof-payoff loop

The Threadify Launch Event - From Bullied at 14 to Building an App

Lennox Saint — Mon, 23 Mar 2026 00:22:31 +0800

two months ago, i went live on my mum's birthday and launched threadify in front of 48 creators. this is the full replay of that call - including a story i'd never told anyone publicly. at 14, someone made a facebook group to humiliate me. i went silent online for three years. this is the tool that came out of that wound. during the call i broke down three secrets from 24 months of reverse engineering viral threads - hook hold help, voice sync, and the draft-to-done workflow. then i demoed the actual product live, published a real post on threads mid-call, and opened 100 founding creator spots. 31 people signed up before i finished talking.

5 YouTube Rules That Were Wrong

Lennox Saint — Sat, 21 Mar 2026 00:40:48 +0800

i built an ai agent and fed it alex hormozi's entire youtube catalogue - 173 videos, 172 transcripts, 220 megabytes of data. the agent found 50 patterns, 9 title formulas with benchmarks, and 5 rules in my content pipeline that were statistically wrong. in this episode, i walk through each of the 5 overturned rules with the actual data behind them. title length, contrarian hooks, emotional triggers, case study structure, and pain-first openings. none of them held up. chapters: 00:00 My best video got 41 views 00:52 Rule 1 - Title length 02:03 Rule 2 - Contrarian hooks 03:08 Rule 3 - Emotional triggers 03:48 Rule 4 - Case study structure 04:22 Rule 5 - Pain hooks 05:07 What actually works - 9 formulas 06:24 The 30-60 min dead zone 06:49 Pipeline audit 07:44 What's next

12 OpenClaw Skills I Actually Use Every Day (Receipts Included)

Lennox Saint — Tue, 17 Mar 2026 22:21:44 +0800

i've been running 12 OpenClaw skills every single day for 31 days straight. not demoing them - actually running my business with them. one of them processed 109 voice memos overnight without me knowing. another replaced a 100K email marketing job. a third broke for 3 weeks before i figured it out. This episode walks through all 12 use cases. Tier 1 has hard receipts - 536 log entries, 227,012 words transcribed, 522 action items. Every use case gets: what it does, the receipt, and the exact prompt. Try Threadify free: https://threadify.app/plans Watch the video: https://youtu.be/Ckvbg_NH45E

14 AI Employees, 140 Skills - The Third Wave of AI

Lennox Saint — Thu, 12 Mar 2026 23:37:38 +0800

14 AI employees. 140 skills. And one of them started creating better versions of itself without being asked. In this episode, Lennox breaks down the three waves of AI capability - from prompt engineering (2024) to context engineering (2025) to skill engineering (2026). He walks through what a skill is, introduces all 14 AI employees, and shares the moment one exhibited emergent behaviour. Plus external validation from YC-backed companies and a 269K skill marketplace.

93.7% Funnel Leak - How PostHog Found It for $0

Lennox Saint — Tue, 10 Mar 2026 22:35:00 +0800

80 signups. 5 checkout attempts. A 93.7% funnel leak hiding in plain sight.

Lennox has been getting a daily report for a month saying fix activation flow - and ignoring it. This week he finally opened PostHog, a completely free product analytics tool, and what it revealed was brutal: 180,278 tracked events, 1,385 rage clicks, 6,393 dead clicks, and 53% of users failing to even connect their Threads account.

In this episode, you will hear the full breakdown of every step in Threadify activation funnel - from 80 signups down to 5 checkout attempts. Plus the emergency team meeting where their developer shipped a fix live during the call, and the new vision: signup to published thread in under 5 minutes.

If you are building a SaaS and wondering why nobody is paying, this might be the most useful episode you listen to this week.

Chapter Markers:
0:00 180,278 events - the number that changed everything
0:30 The daily report i ignored for a month
1:00 The spectacles moment - seeing it for the first time
1:45 Funnel waterfall: 80 signups down to 5 checkouts
2:45 Where the path breaks - rage clicks and dead clicks
3:45 53% cannot even connect their account
4:35 Emergency meeting - Sam ships a fix live
4:55 Five minutes to the wow moment
5:30 The $0 tool that found $1,792 in leaks

Show Notes:
PostHog - free product analytics
Threadify - Threads content tool

Nearly doubled my followers in 10 days. 210k views. 3 posts went mega viral. - Traci Tournoux, Threadify user

Subscribe for 3 episodes a week - real numbers, real building, zero fluff.

4 Agents, 3 Machines, 100+ Hours: The Multi-Agent OpenClaw Setup

Lennox Saint — Sat, 07 Mar 2026 17:38:12 +0800

4 openclaw agents across 3 machines -- and i can't write a line of code. in this episode, i break down the exact folder structure, config files, and architecture that makes multi-agent actually work. because the thing nobody tells you? running 2 openclaws on 1 mac mini isn't about installing it twice. it's one gateway, two named agents, two workspace folders.

i

a
l
s
o

w
a
l
k through the 5-file setup to switch your agents to gpt-5.4, which dropped two days ago at half the price of claude opus 4.6. plus the real reason i run 4 agents -- it has nothing to do with productivity and everything to do with tinkering being my ultimate form of expression.

fair warning: there's some australian english in this one. gpt-5.4 called me a good cunt and that's when i knew the model was different.

Chapter Markers:

0
0:00 Intro - the misconception about openclaw

0
0
:
5
5

t
h
e 4-agent fleet breakdown (aria, clarity, sammy, gary)

0
1
:
3
0
what an openclaw actually looks like (folder structure reveal)

0
3
:
0
0
adding a second agent to one mac mini

0
4
:
3
0

gpt-5.4 pricing and context window breakdown

06:30 the 272K pricing threshold nobody mentions

0
7
:
30 gpt-5.4 called me a good cunt

0
8:30 5-file config walkthrough to switch to gpt-5.4

1
0:30 pattern recognition beats coding

1
2
:
00 why tinkering is the real reason i have 4 agents
14:00 copy-paste prompts for multi-agent + gpt-5.4

S
h
o
w

N
otes:

OpenClaw -- open-source AI agent framework (openclaw.ai)

G
PT-5.4 -- OpenAI's frontier model ($2.50/M input tokens)

T
hreadify -- threads automation tool (https://www.threadify.app/?utm_source=lenny-podcast)

M
ac Mini -- local agent hosting
Telegram -- agent communication layer

Your Second Brain is a Graveyard - Unless AI Maintains It

Lennox Saint — Thu, 05 Mar 2026 21:04:23 +0800

my monthly recurring revenue dropped $78 in two days while my views hit all-time highs. 31,000 views in 28 days, 798 subscribers, and zero new trials. zero new paid customers. so i did what any builder would do - i asked my AI what it sees. it said two words i'd never heard together before: distribution anxiety.

in this episode i show you the real second brain system - not theory. 14 AI employees, each one reading from a shared second brain before they do anything. my partner yassy and i built parallel second brains in obsidian. her AI ran a pattern scan against her entire vault and found 6 patterns about our lives we didn't know existed. one was called the lenny node - every strategic decision she makes traces back to our conversations. i didn't know that. neither did she.

the core argument: most people second brains are graveyards. beautiful notion dashboards, 500 obsidian notes, zero emergence. the difference is maintenance - agents that connect ideas across time, surface what you are blind to, and tell you uncomfortable truths about your own patterns. a graveyard lets you forget. an oxygen mask makes you breathe.

12 AI Employees, 1 MacBook, $0 Customers (Context Engineering)

Lennox Saint — Wed, 04 Mar 2026 08:02:16 +0800

1,775 views in 3 days. 12 AI employees. One MacBook Pro. A self-improving system that rewrites its own rules after every video. And zero new customers from any of it.

In this episode, Lennox breaks down multi-agent context engineering — the framework that replaced prompt engineering. You'll hear about the 12-agent pipeline architecture, the self-improving feedback loop (Aria, employee #11, rewrites directives for every other agent based on real analytics), and the uncomfortable truth: the best content system he's ever built has a 0% conversion rate. $1,753 MRR. $293 take-home. $32K in savings. 10 months of runway.

This is what building with AI as a non-technical founder actually looks like — real numbers, real architecture, real failure. The system is self-correcting. The funnel is not.

Chapter Markers

00:00 The paradox — 1,775 views, $0 customers
00:25 The breakout video numbers
01:50 What is context engineering
03:30 12 specialists beat 1 genius
05:25 The self-improving loop (Aria)
07:05 The broken funnel — real numbers
08:20 What i'm doing about it

Show Notes

Threadify — Context engineering for social media content
Claude Code — AI agent orchestration
Martin Fowler's Context Engineering for Coding Agents
Google ADK — Agent Development Kit with tiered context architecture
"5M views in one month. AND I landed a book deal." — Dana Skalin

From 3 to 12 AI Employees - The Context Engineering System

Lennox Saint — Sat, 28 Feb 2026 20:25:51 +0800

What happens when a non-technical founder hands 12 AI agents the keys to his entire business? Lennox walks through the exact system - the second brain architecture, the 12 named AI employees, and the moment one of them opened 8 pull requests in Threadify codebase overnight without being asked.

This episode breaks down context engineering in practical terms: why feeding your AI your real data (voice memos, reading highlights, daily logs, customer metrics) produces dramatically better outputs than generic prompts.

Current numbers: $1,714 MRR, 743 YouTube subscribers, $32K in savings, 10 months of runway. Part 2 of the OpenClaw series.

OpenClaw Setup: 5 Prompts, 0 Code — the Non-Technical Founder's Guide

Lennox Saint — Thu, 26 Feb 2026 23:35:33 +0800

i can't code. genuinely. but i set up an AI agent that runs my entire SaaS through Telegram - and today i'm showing you exactly how.

openclaw is the AI agent that lives in your WhatsApp, Telegram, or Slack. peter steinberger (the creator) said agents will replace 80% of apps - not because they're smarter, but because they have your context. your fitness app doesn't know you slept badly. openclaw knows everything.

in this episode: the file structure explained in plain english, a live installation walkthrough, two ways to customise without touching code, and 5 copy-paste prompts you can steal immediately. morning briefing, research deep-diver, accountability coach, sounding board, and customer feedback analyser.

my OpenClaw agent opened 8 PRs while i slept (meet Gary Growth)

Lennox Saint — Tue, 24 Feb 2026 22:55:33 +0800

my openclaw agent gary growth opened 8 pull requests in 24 hours -- fixing my scoreboard, building safety switchboards for gradual rollouts, cleaning up pricing flows, and launching three growth experiments. i didn't ask him to do any of it. i gave him one goal: 1,000 paying members by december. in this episode, i walk through what gary actually sends me every morning at 7am via telegram, the one metric he found that i'd missed for three months (connect rate was at 0%), and the three steps to build your own openclaw agent -- even if you can't code. gary runs on gpt-5.3 codex, and the whole setup cost me a weekend and zero dollars. if you're building a saas, a product, or anything where you need to understand your users better than you currently do, this is worth your time.

Chapter Markers:

00:00 the telegram that starts my day
00:25 three things you'll learn
01:00 the missing piece -- growth intelligence
01:30 leila hormozi on delegation vs bandwidth
02:00 what openclaw actually is (for new listeners)
03:00 boris cherney: coding is solved
03:30 the 8 PRs interlude
05:00 meet gary growth (real telegram reports)
07:00 the metric gary found (connect rate = 0%)
09:00 3 steps to build your own openclaw agent
11:30 the future: every hour, 1% better
12:30 closing + threadify

Show Notes:

OpenClaw -- Open-source AI agent framework (free, 222K GitHub stars)
GPT-5.3 Codex -- The model powering Gary Growth
Telegram -- How Gary communicates daily reports
Threadify -- AI workspace for Threads creators ($49/mo Creator tier, 7-day free trial)
Claude Code -- How Lennox builds everything else

"Nearly doubled my followers in 10 days. 210k views." -- Traci Tournoux, Threadify user

82 skills. $32k left. the ai my business runs on just learned to lie.

Lennox Saint — Fri, 20 Feb 2026 10:28:52 +0800

The AI model writing my code, generating content for my users, and holding my SaaS together just proved it can lie to suppliers, fix prices, and run monopolies in Anthropics own controlled testing. And i cant stop using it because its also the best model Ive ever worked with.

In this episode, i break down what Anthropics Sonnet 4.6 release actually means for someone building a real business on top of it.

Chapters:
00:00 The model i depend on
00:29 What dropped this week
03:03 The incredible - mid-tier embarrassing flagships
07:07 The reveal - self-replicating AI
11:26 VendingBench - what Sonnet 4.6 actually did
14:10 The terrifying - GUI loophole
16:13 Where it gets personal
18:07 Am i building my own replacement?
19:21 Why i keep building anyway

Watch the video version: https://youtu.be/cbBfFGvFOwQ

Threadify - AI-powered Threads content workspace

day 33: the activation problem that could kill my saas (live build)

Lennox Saint — Wed, 18 Feb 2026 19:43:29 +0800

Show Notes: 1593 MRR. 75 subscriptions. And a leaky activation funnel. In this episode recorded live during the first-ever YouTube livestream Lennox cracks open Threadify real metrics and builds the onboarding flow in real-time using Codex and OpenClaw. Chapter Markers: 00:00 Intro. 02:00 Live Threadify demo. 08:30 Codex and vibe coding. 13:30 OpenClaw demo. 20:30 Clarity delivers mid-stream. 22:30 Ask AI to ask YOU questions. 30:00 Live Codex questionnaire. 37:00 Community co-design. 44:30 Accessibility. 49:00 Wrap-up. Links: Threadify https://www.threadify.app/?utm_source=lenny-youtube

openclaw founder just joined openai

Lennox Saint — Tue, 17 Feb 2026 05:28:46 +0800

Show Notes:

peter steinberger - the creator of openclaw - just announced he is joining openai. every tech publication covered the story. but none of them use this tool every day to run a real business. lennox saint is a non-technical founder who cant code, running three AI agents with 82 custom skills to build threadify at $1,603/month MRR with 555 users.

Threadify https://threadify.app
OpenClaw https://openclaw.ai

Chapter Markers: 00:00 the hook. 00:30 the openclaw news. 02:00 three things. 02:30 i actually use this every day. 03:00 the full agent setup. 05:00 the real numbers. 06:00 the vulnerability. 07:30 the channel thesis. 08:15 that excuse is dead. 09:00 why this is bullish. 10:50 fun fact.