I build tools for problems I know firsthand.
Thirteen years in GTM strategy and operations shaped how I think about systems. Now I use AI to build products for the problems I keep running into.
A multi-source market news briefing built around what matters to me.
I was tired of jumping between different websites to keep up with the companies and topics I follow, so MarTech Intel brings me all Marketing/Technology news in a daily digest. It even has a view of how the story changes over time. Built to turn a scattered reading habit into a reliable stream of relevant information.
A place to keep my family’s history together.
A family tree anyone can start, with relatives invited to contribute. Corrections keep their history, and the information stays within the family. Built to give everyone a shared picture without one person maintaining and resending a spreadsheet.
These projects started with problems I wanted to solve. AI agents write the code. I set the direction, make the decisions, and keep improving the results through everyday use.
Every project has a GitHub repo you can fork. Dynasty Analyzer, Family Tree, and Job Search also have apps anyone can sign up to use.
Market intelligence for a GTM org that doesn't have one.
Receipts and email become a gear ledger that answers back.
Spending and habits, tracked and then actually interpreted.
One Telegram thread, routed to whichever agent should answer.
Trade and roster valuation for Sleeper dynasty leagues.
This site. Written, designed and deployed by agents.
Expands one lesson at a time, then grades the photo that comes back.
Tracks open roles against my own criteria, and keeps every link real.
Start a private family tree and invite your relatives to contribute.
Grades my own side projects and tracks what each one costs to run.
Each one cost at least a quarter to learn.
An agent with bad context makes confident, wrong decisions fast. MarTech Intel starts with the sources and how they fit together.
Talk the problem through, get a plan, then build. Adding a multi-tenant design later taught me to think about scale from the beginning.
Camera Agent uses one model to expand a lesson and a different one to grade the photo that comes back.
A model saying “done” is a claim. Tests, independent/subagent review, and using the product are how I check that claim.
You’re your own best user. Build, test, break, repeat. That’s the simplest path to a better product.