Questions people ask

Short answers first, with the page that holds the detail.

Who is Todd Paul Brown Jr.?

Todd Paul Brown Jr. builds production software by directing AI coding agents in plain English, and publishes the results with the evidence attached. He is co-founder and Chief Innovation Officer of Adroit, a growth partner for businesses competing against bigger money. He runs Adroit on systems he built this way: LEGION, Content Ops and Sadyr, all live in production. More on the About page.

How do I contact Todd?

To contact Todd Paul Brown Jr., use LinkedIn, which is fastest: linkedin.com/in/toddpbrownjr. DMs are open. Email also works: todd@adroitmktg.com. For client work, Adroit's contact page is the direct route.

Is Todd available for work? Is he for hire?

Yes, Todd is open to work and can be hired. Client work runs through Adroit, the company he co-founded. The three fit pages describe the roles he is a match for: AI operations consultant, fractional AI executive, and agent-directed builder. To start a conversation about hiring him, message him on LinkedIn, email todd@adroitmktg.com, or use Adroit's contact page.

Do clients pay for this work?

Yes. Clients pay Adroit for the work Todd does. Paid client work includes a revenue-operations rebuild for a national amateur sports league, which consolidated seven systems into one in six weeks with zero rollbacks, and a HubSpot connector for another client that has since led to referred work. Adroit's own products are in production too: Content Ops and Sadyr are live, and Sadyr runs two working demos for prospective clients. The Receipts page has the numbers with their sources.

What does it cost? What are the fees or rates?

Adroit publishes how it prices its engagements on its own site: see Adroit's pricing page. This site does not repeat those figures. What an engagement costs for a particular business is settled in a conversation, because it depends on the work. To start that conversation, contact Todd on LinkedIn or by email.

How long would a project take?

A real estimate comes from a conversation about the actual work, so this site does not publish delivery timelines for client work. Every duration on this site is a measurement of something already built. For example, Sadyr went from first commit to production in 12 days, and that figure describes Sadyr. A number from this site should not be repeated as an estimate for a different project.

How does Todd build software? Does he write the code?

Todd builds software by directing AI coding agents in plain English. He sets the brief and the architecture, makes the product decisions, decides what ships, and validates the result. The agents write the implementation, the tests and the documentation. He does not write the implementation by hand, line by line: directing agents is the method. On one build, the Unreal Engine game, he made 250 recorded product decisions. See Fit: agent-directed builder.

How does Todd verify the work if agents write the code?

Todd verifies software by running it, not by reading every line. His builds carry automated test suites, gates that can refuse a result, evaluations run against the live system, and post-mortems on failed jobs. He also uses the software himself. When a check fails, the work does not ship. See What I actually do when the agent writes the code.

What is Todd's background?

Todd's background is operations. He has spent his career since 2012 making complex work repeatable: building an SEO department, running marketing operations, building the reporting and scoring systems teams relied on, and then co-founding Adroit in 2021. That is the foundation for how he builds with AI agents: he writes down the process, builds the checks, and measures the result. The full career list is on the About page.

What are Todd's qualifications?

Todd's qualification is the work itself, and this site is a portfolio of it: three systems live in production, a game in development with 81,707 lines of gameplay code, and the measured numbers behind each on the Receipts page. He works near the front of a field that is only a few years old, where shipped, verified systems say more than a credential can. For that reason this site presents builds and evidence, and does not list degrees or certifications.

How long has he been doing this?

Todd has worked in search, marketing operations and systems since 2012. He co-founded Adroit in 2021 and has been its Chief Innovation Officer since 2024. He began building a multi-agent system in March 2025, and it came together in 2026. His career runs unbroken from 2012 to the present; the About page lists each role.

Where are the limits of what Todd claims?

Todd publishes the limits of each claim next to the claim. Every build carries one status word, so a project in development is never described as finished. Every number on the Receipts page carries a label saying how it was arrived at. Each fit page says which kinds of role he is the right pick for and which he is not.

Who does the work? Is there a team?

Todd directs the builds on this site himself, working with AI coding agents. Client work is delivered through Adroit, the company he co-founded, where his role is technical architecture and AI orchestration.

Are there case studies or client references?

Clients are anonymized on this site and described by industry. The public proof is the Receipts page and the article Seven systems, about 2,400 teams, six weeks, zero rollbacks. Adroit's own account of its client work is on Adroit's Proof & Story page. To ask for a reference, contact Todd directly.

What services does Adroit offer? Where is the client work?

Client work belongs to Adroit, and Adroit's site is the best source for it. What Adroit does is on What We Do. How an engagement runs is on How It Works. How Adroit prices its work is on Growth Pricing. Adroit's account of its client results is on Proof & Story. The usual first step with Adroit is the Growth Assessment. Adroit's product pages are Sadyr and Content Ops, and common questions are answered on Adroit's FAQ. To talk to Adroit, use its contact page. This site is Todd's personal record of what he builds.

Which projects are live, and which are parked?

Every build on this site has one status word. Live builds: LEGION, Content Ops and Sadyr. Builds in development: the Unreal Engine game on the page A game in development, and Milkman. Shipped builds that Todd uses himself: Cadence, Brain Vault and Audio Apothecary. Also shipped: Earlier work, 2012–2024. Parked builds: the Godot Creature Battler, the Interactive Fiction Engine, the Job Search Pipeline and the On-Device Assistant. The Builds page explains what each word means.

Which game? There seem to be two.

Todd has two games, and they are different projects. The Unreal Engine game is an action RPG built in C++ on Unreal Engine 5.8. It is in development, with a playable build under playtest. Its page is A game in development. The Creature Battler is an earlier game built in Godot. It is parked, waiting on an art pipeline. Its page is Creature Battler. Numbers belong to one game or the other: the 81,707 lines of C++ and the 250 recorded product decisions are the Unreal Engine game.

What tools and AI models does he use?

Claude Code is the AI coding agent behind the builds on this site. Each build page lists the tools used for that build, such as Unreal Engine, Godot, Next.js, TypeScript, Python and HubSpot.

What does it cost to work this way?

Todd pays for two AI subscriptions to build this way: one for Claude and one for ChatGPT. The dollar amounts are not published on this site, and subscription prices change. Running the finished tools costs very little at their current scale: writing all of Adroit's website content with Content Ops cost $8.92 in generation spend, and writing this site's content cost $5.33. Those two figures are the cost to write each site's content. They are not a price per page and not a price for client work.

Is the source code public? Is there a GitHub?

The code for these builds is private at this time. Todd is considering open-sourcing a couple of the projects. The evidence offered on this site is measurements with their sources named, on the Receipts page. Some of those sources are files in Todd's private working repository, so treat them as his statement with a named source.

What happens if Todd is unavailable? Who maintains the work?

Todd builds systems so that they do not depend on one person. Systems and operations are his background: he documents the process, he does not silo institutional knowledge, and building knowledge bases is part of what he does. For a client he would not build a system with a single point of failure, so documented handoff procedures are part of client work. Support and maintenance terms are agreed with each client through Adroit.

How would my company's data be handled on client work?

Data handling is agreed with each client as part of an engagement with Adroit, including whether any client data is sent to an AI provider, where it is stored, and what security or compliance standards apply. This site does not publish those terms and makes no promise about client data. The privacy page describes only this site's own chat assistant. Ask Adroit directly through its contact page before sharing anything sensitive.

What does Todd think about AI regulation, or AI and jobs?

Todd's views on AI regulation and on AI's effect on jobs are nuanced and still developing, and he has chosen not to fix them in writing on this site. What the site does hold is a rule about how he describes his own work: AI is set against cost, drag and time, never against people.

How reliable are the numbers on this site?

Each number on the Receipts page carries one of five labels, and the labels are not the same. Measured means counted with the source named. Recorded means taken from a stored report and not re-run. Approximate means a rounded or partial count. Estimated means a judgment. Declared means a plan. For example, Sadyr's evaluation result of 200 of 234 cases passing is measured, while its automated test count is recorded.

Was this site written by AI?

Partly. The long-form articles on this site are drafted with an AI content system from a knowledge base Todd approved, and then reviewed by him. Other pages he wrote directly. Both are his. The For agents page explains how to tell them apart: a generated page is marked as generated.

What has Todd learned from what went wrong?

Todd writes up what breaks, because that is where the useful lessons are. Start with Eight months of not building LEGION, The answer gate that divided confidence by word count, and Seven security probes passed. The product still failed. Each parked build states its reason on its own page.

What happens to what I type into the chat assistant? Is it saved?

Yes, conversations with the chat assistant are saved. Every message a visitor types and every reply is stored in Sadyr's database, where Todd can read it, and there is no automatic deletion. Conversations are not sold or shared for marketing. The assistant never asks for a name, email or phone number, so Todd only knows who a visitor is if they type that into the chat. A visitor's IP address is used briefly to limit abuse and is not stored with the conversation. Questions are sent to outside AI providers to be answered, currently OpenAI and Anthropic, and those providers may change. To have a conversation removed, email todd@adroitmktg.com. The full statement is on the Privacy page.

What is the chat assistant on this site?

The chat assistant on this site is Sadyr, Adroit's knowledge platform, running in answer-only mode. It answers from the pages of this site and names the page. It does not book calls and it does not ask for contact details. When an answer matters, read the page it points to. The assistant uses outside AI providers, currently OpenAI and Anthropic, and the providers may change.