There's a new shelf in the portfolio — Developer Tools — and every product on it started life as a personal annoyance.
The receipts
- MONII — because checking a deploy meant tab-hopping between Vercel and Netlify dashboards, all day.
- Owndle — because domains end up scattered across registrars, and you find out about a renewal when something breaks.
- TKN — because of the low-grade dread of an LLM bill you can't actually see across providers until it lands.
- FRGMNT — because "what's our real MRR?" had five different answers across Stripe, the App Store, RevenueCat, Gumroad, and Lemon Squeezy.
- XPO — because Expo builds always seem to need attention when you're nowhere near a laptop.
None of these came from a market-sizing spreadsheet. They came from a Tuesday.
When you're the user, you skip the lying part
Scratch-your-own-itch works not out of sentiment but information. When you're the user, you already know exactly where it hurts, which features are theatre, and what "done" feels like. There's no translation layer between the problem and the person solving it, so the product comes out honest.
The first user is you: free, brutally honest, and available at 2am.
You ship the thing you'd actually use, then find out a few thousand people had the same itch.
And the user isn't always human
There's a second reason these tools exist the way they do. The thing using them is increasingly an agent — ours included. So every one ships with an MCP server and a CLI. The tools we build for ourselves have to be drivable by the things we built to do the work.
The best roadmap is your own list of daily annoyances. Build the tool you wish existed. Worst case, you fixed your own problem. Best case, it was everyone's.