WRITING

Notes from the studio.

What we have learned building the systems underneath AI-native products. Specific, technical, and written from things that shipped.

2026-08-19
Agents that know when to defer
Most multi-agent systems fail at the seams. Confidence-tagged routing and explicit handoff markers make the seams visible — and make deferring to a specialist, a tool, or a human a decision the system takes deliberately.
READ
2026-07-08
Output that cannot drift past its contract
A schema around every model call turns an unpredictable string into a typed value. Centralised prompts turn a scattered liability into a versioned asset. Neither is exciting, and both are what makes LLM features maintainable.
READ
2026-06-11
AI proposes, a hardened executor applies
Letting a model hold the write path is the mistake. Separating proposal from execution — structured previews on one side, a validating executor on the other — is what makes AI safe to point at real data and real money.
READ
2026-05-14
The app should be optional
A local-first test that is harder than it sounds: delete the application and check whether the work survives. Three shipped Mac apps, and what choosing plain files on disk actually costs.
READ