Writing
Longform notes on agentic AI, product, and building in public. Canonical versions live here, with a link to where they first appeared. Pieces that stay on another channel are linked from this list.
· X
A practical guide from the Cursor + SpaceXAI early beta. How to structure a bot team, real cases from people already testing, and the first lessons on autonomy and organisation.
AI brought side projects back: building an agent communication protocol
How curiosity about how AI agents talk to each other turned into ASAP, an open-source protocol — plus the vibe coding workflow and context engineering behind it.
What to do when AI agents hallucinate? The answer is more about architecture than prompting.
Lessons from putting an AI agent into production: why a better knowledge base and a better prompt were not enough, and how a router agent plus specialists fixed it.