WilliamFrank

Designed around agents

Bill Woodruff — Wed, 30 Sep 2026 16:43:55 -0700

AI lowered the cost of building software. It also changed what the software should be, and what the company around it should look like.

This episode covers the five decisions that follow once an agent is a user of the software, what changes in the firm's roles, why incumbents end up with an annex beside the old core, what still gets rented, and three counters to the argument.

WilliamFrank is one integrated set of agentic rails for a wealth team: mail, calendar, messages, projects and tasks, people and households, and files, with an agent in every one.

I'm building it in the open with a founding consortium of advisors. Apply at williamfrank.ai.

Read the essay: https://williamfrank.ai/designed-around-agents.html

(00:00) - The fifth part
(00:37) - What changes when the agent is a user
(01:32) - Five decisions
(02:48) - Now the company
(03:39) - Three places the firm changes
(04:40) - Why from scratch
(05:43) - Designed together
(05:59) - What to rent
(06:44) - Three counters
(07:59) - One number
(08:23) - Agentic wealth
(09:19) - The founding consortium

The 100-client ceiling is a software problem

Bill Woodruff — Tue, 04 Aug 2026 17:24:43 -0700

Every good advisor eventually runs into the same number, somewhere around a hundred clients. The practice stops growing because the week is full, and the industry gave the cap flattering names.

This episode argues that most advisors aren't hitting a relationship ceiling. They're hitting an administrative one several floors below it. The overhead ratio barely moves with scale, because every tool the industry bought still needed a person to run it. Agents working from one live, integrated set of facts for a household change that, and three floors move at the same time: who can be served, who can deliver, and who you can build a team from.

WilliamFrank is one integrated set of agentic rails for a wealth team: mail, calendar, messages, projects and tasks, people and households, and files, with an agent in every one.

I'm building it in the open with a founding consortium of advisors. Apply at williamfrank.ai.

Read the essay: https://williamfrank.ai/100-client-ceiling.html

(00:00) - The same number
(00:44) - Which ceiling is it?
(01:36) - The overhead ratio
(02:11) - Every dollar carries its own labor
(02:53) - Why technology didn't move it
(03:15) - What's different now
(03:38) - Only as useful as its reach
(04:36) - Running the math
(05:14) - Where the hours go
(05:43) - Three floors move at once
(06:22) - Who can deliver
(06:54) - Who you can build a team from
(07:32) - Where that leaves me
(08:23) - The founding consortium

Agentic wealth — unlocking the agency of advisors

Bill Woodruff — Sat, 25 Jul 2026 20:03:30 -0700

The advisor shortage won't be solved by recruiting harder. The profession isn't producing advisors as fast as the assets it serves keep growing, and most of an advisor's week goes to the work wrapped around client meetings rather than the meetings themselves.

This episode is about the advisor's agency: which work a person should touch, which work an agent should carry, and what record both of them work from. Agents carry three jobs on the back end of an advisor's business. They run the administrative workflows, organize what is known about a household, and push the work forward before anyone asks. Only the first can run on data that arrives overnight. Work that's testable, repeatable and settled belongs on rails, with every action attributed and nothing deleted. Work that requires knowing a family belongs to the advisor.

WilliamFrank is one integrated set of agentic rails for a wealth team: mail, calendar, messages, projects and tasks, people and households, and files, with an agent in every one.

I'm building it in the open with a founding consortium of advisors. Apply at williamfrank.ai.

Read the essay: https://williamfrank.ai/agency.html

(00:00) - The advisor's agency
(00:25) - The advisor shortage
(01:07) - Where an advisor's week goes
(01:56) - A design problem
(02:14) - The three jobs agents carry
(03:14) - An intelligent client profile
(04:13) - Getting off batch
(04:48) - The limits of old rails
(05:45) - What the rails are for
(06:36) - That's agency
(07:07) - Compliance is not friction
(07:45) - The hundred-client ceiling
(08:56) - Closing the shortage
(09:16) - What I'm watching
(09:34) - The founding consortium

AI lowers the barrier to build — nine weeks, four versions, no code

Bill Woodruff — Sat, 13 Jun 2026 08:02:27 -0700

A few years ago I had something I wanted to build and couldn't. The team, the capital and the specialized skills stood between the idea and a working version of it.

This episode is about what changed. Over the summer I designed one integrated set of agentic rails for a wealth team without writing a line of code. The building was the part I didn't do. The work was deciding: one interface, the agent as a user, nothing deleted, simplicity over borrowed features. AI gives you more of whatever you bring to it. For people who practice a craft like evidence-based investing, the barrier to building what ought to exist is lower than it has been, and it's still dropping.

WilliamFrank is one integrated set of agentic rails for a wealth team: mail, calendar, messages, projects and tasks, people and households, and files, with an agent in every one.

I'm building it in the open with a founding consortium of advisors. Apply at williamfrank.ai.

Read the essay: AI lowers the barrier to build

(00:00) - Something I wanted to build
(00:23) - What got built this summer
(00:47) - Where the work went
(01:41) - The decisions are the product
(02:50) - Taste, rules and judgment
(03:16) - Why no one built this for wealth teams
(03:55) - More of whatever you bring to it
(04:20) - The objections
(05:11) - The first user is us
(05:58) - The founding consortium

Why WilliamFrank — wealth management has no operating system

Bill Woodruff — Tue, 09 Jun 2026 09:02:03 -0700

Tailored financial advice has been rationed for decades by cost structure rather than by intent. Serving a household well takes a certain number of hours of human labor, and firms decline the families whose math doesn't work.

In this first episode I lay out the problem I want to work on. Wealth management has no operating system. A household's facts live in nine systems, and every place a name gets typed a second time is a place the record forks. An expert advisor ends up tracking those facts, or managing the people who do.

WilliamFrank is one integrated set of agentic rails for a wealth team: one record of the household, agents that carry the administration and organize the context, and advisors who decide.

I'm building it in the open with a founding consortium of advisors. Apply at williamfrank.ai.

Read the essay: Why WilliamFrank

(00:00) - The problem I want to work on
(00:56) - Who I am
(01:36) - The stack test
(02:30) - What an operating system does
(03:55) - Complexity as a tax
(04:55) - Tracking facts or managing people
(05:55) - What changed
(07:02) - What WilliamFrank is
(07:55) - Three arguments against
(09:08) - Where that leaves me
(10:43) - The invitation