I don’t just have opinions about AI.
I ship them.
Eight systems, built end-to-end and in active use. Every one started as a workflow I was already doing by hand. The demos run on fictional data — click through everything.
They stack.
These aren’t eight unrelated demos. A personal context layer feeds an ops layer, which feeds the products — the same architecture I’d build for a company’s market intelligence.
Durable memory
A structured knowledge base so every tool starts from real context instead of a blank session. Nothing above it works without this.
Daily judgment
Systems that read the context layer and act on it — briefings, tracked commitments, ranked outreach, feedback on my own data.
Things other people use
Full editorial products with their own audiences, CMSes, and reader apps — where the first two layers stop being personal.
Frontier
A source-first AI intelligence product for people building with AI: model releases, labs, open models, developer changes, and verified reporting in one editorially organized interface built around signal instead of recycled takes.
Neighborhoodr
A living feed for where you live — posts, events, and a neighborhood pulse from the people who know it best. A rebuild of the network I started on Tumblr in 2009, which grew past 25 cities.
Chronicle News
A modern take on Circa — the mobile news app I helped build that broke stories into structured atoms readers could follow over time. Rebuilt with an editorial CMS and a reader app.
Market Signal
A market-wide screener on Alpaca’s movers and most-active endpoints, wrapped in a terminal-style board: index tape, top gainers and losers, three free news sources, and a fresh opinionated market read written every 30 minutes.
AnthonyWiki
A personal knowledge wiki compiled from journals, notes, and messages — a writer’s pass over raw personal data that synthesizes it into structured articles. The context layer that lets every other tool reason from durable memory.
Chief of Staff
A daily-briefing and project-tracking system built on top of AnthonyWiki. Morning briefings, open items across projects, 1:1 prep, and a running memory of people, projects, and commitments.
Networker
A local-first tool for deciding who in a professional network is worth reconnecting with — turning a fuzzy relationship graph into a ranked system for outreach, follow-up, and second-degree discovery.
HealthDash
I text what I eat and what exercise I do to Poke, which feeds Notion. HealthDash turns that into a dashboard with coach-style commentary — AI giving feedback on the data, not just charting it.
Want this fluency
on your team?
The same instinct that builds these is what makes an AI-era content operation actually run. Build notes for most of them are public on GitHub.