26 notes filed here
When you add an AI layer to a retrieval feature, data decisions that developers used to make by instinct become explicit contract requirements. Live demos are often the first place you find out which ones are missing.
An agent that plans feature flag removals is useful. The rule about which flags it may never touch is what makes it safe to leave running, and what let the job move to the person who actually wants it done.
Reviewing acceptance criteria for a booking flow turned up a lineage question, an undefined 'stale' and a product overlap before any code was written, and one shared assumption that got through anyway.
Four kinds of failure our test suite couldn't see, which only surfaced once real stakeholders used an AI booking assistant in a room.
A tooling migration, some dependency bumps and a telemetry schema: none of it on a roadmap, and all of it about keeping several frontend services easy to work in.
Building a standardised logging system for an AI assistant service: making consistent structured logging the path of least resistance, and thinking carefully about what an AI service should and shouldn't persist.