The AI-Built App Worked. Then I Opened the Repo.
A year ago, I would have sent it back. This time, closing the gap between useful and releasable took an afternoon.
Read articleWriting
Field notes from building trustworthy AI.
Experiments, failures, and practical notes on local AI, agent systems, and trustworthy software.
My Local LLM Scored 6/6. It Was Wrong Every Time.
Six months of trying to make a 1.2B model useful, and the measurement mistakes I made along the way.
Read article