I build brands —
and the operating
systems behind them.

DavidOS is the operating system I built for my own life. It runs my studio, keeps my family's calendar straight, tracks the goals I'm actually trying to hit, and handles the production work that used to eat my evenings — so one person can operate like a team.

Under the hood: 83 small AI agents, each with a written job description — what it may read, what it may write, what it must never do. Anything that leaves the machine waits for my approval. Anything that describes me has to cite a source. That's the part worth paying attention to — not that AI does the work, but that it's built so I can trust what it does.

If you're reading this wondering whether your team could run this way — smaller headcount, faster output, nothing shipped unreviewed — that's exactly what I help people build. work with me →

Not mockups — running software and real artifacts, designed by me and produced through DavidOS pipelines. This is what "working with AI" actually looks like when the taste stays human and the production system does the lifting.

Everything I've published, on 4 shelves. Squarespace-era posts migrated; new field notes and process pieces ship from the OS itself.

Field Notes (3)

Written from inside the OS — each one from something that actually happened in the system that week.

07-30

I Taught My Investment Agent to Refuse

The most important function in my portfolio agent throws an error when output starts sounding like advice. Designing the boundary before the feature.

3 min read

07-30

A Claim Ledger for the Person You Market

Every claim I make about my own work now traces to a dated source file — or it doesn't ship. On building a claim ledger for the person you market: yourself.

3 min read

06-12

A Trading Agent With a Human Approval Loop

Building a Robinhood agent that proposes and never executes — prompts, guardrails, and why the approval loop is the product.

9 min read

Writings (4)

Longer arguments about where knowledge work, agents, and design are heading.

08-02

Files Move. Platforms Don't.

I shipped a complete product vision as a folder of files any AI can build from. On provider-agnostic systems, and why the file is the interface that survives.

5 min read

06-12

The End of the Second Brain

Knowledge management was never the goal. Why I'm building an agentic operating system instead of another notes app.

6 min read

06-12

Your Files Are Dead. Here's What Comes Next

Files made sense when humans were the only readers. On moving from documents you organize to context a system can act on.

5 min read

06-12

The Agent Deployer: A New Role Emerges

Between the engineer and the end user there's a new job: the person who deploys, governs, and owns agents in production.

4 min read

In Practice (7)

Applied AI in real brand, content, and product work — with the numbers.

08-19

Huntington: launching a national golf platform, one section at a time

Huntington built a golf platform around World No. 1 Scottie Scheffler. Idea to launch in Sitecore, shipped in stages because the rights cleared in stages.

3 min read

08-18

Mantium: Leading UX design for an enterprise-grade AI platform

Mantium let enterprises build secure AI chat on their own data, no code. A new customer's first deployment averaged 2+ hours; we got it under thirty minutes.

3 min read

08-18

Cardinal Health: Redesigning an enterprise dashboard for clarity, consistency, and scale

Modernizing the Order Express dashboard pharmacy customers rely on — inside brand, CMS, and compliance constraints of a regulated enterprise.

2 min read

08-18

Kahle Nutrition: Strategic UX and web development for a local health practice

Kahle Nutrition & Chiropractic had no website. Now patients can find, trust, and book — strategy, design, launch, and the technical plumbing.

2 min read

08-18

Arloa: Leading product, brand, and growth for a GenAI platform with real-world impact

Arloa helps parents of children with special needs manage IEPs and school communication. Product, brand, and go-to-market — from idea to launch.

2 min read

06-12

Designing Enterprise AI for Non-Technical Users

What Mantium taught me about making model capability legible to people who will never read a prompt guide.

3 min read

06-12

Tripling Content Output With a Four-Person Team

The Claude Code content pipeline behind Arloa — 4 to 12+ pieces a month without adding headcount, and without losing the brand.

3 min read

Under the Hood (4)

The process posts — briefs, iterations, refusals, and how the outputs actually got made.

07-31

One Token, Twenty-Seven Dashboards

I picked a typeface for a website and ended up re-typesetting an operating system — including the contrast failure I couldn't see by eye.

4 min read

07-30

Under the Hood: The Sideline Board

A game-day sideline board built as data plus a renderer — how the same content-in-files pattern behind DavidOS handles live cards.

2 min read

07-30

Under the Hood: The Patch Archive

Rebuilding a military patch archive with an agent pipeline: research, redraws, and a render script that turns markdown records into a browsable site.

3 min read

07-30

Under the Hood: This Site

This site is generated from markdown by the same system it describes. The build script, the content rules, and why there's no CMS.

2 min read