Blog
Notes from building agents that work
What we shipped, how it works underneath and how to get the most from it.
Latest
·17 min read
Big LLMs teach. Small LLMs do the work.
A $500 run made a 9B open-weight model beat five frontier models on its maker's test. How LLM distillation works, where it fails, and the stack to try it.
Definable team
·11 min read
Code review has two jobs. Hand one to agents.
How to ship AI-written code you still understand: hand bug-finding to tests and AI reviewers, and spend people on decisions and data flow.
·8 min read
Your AI agent needs a clean handoff
A stuck agent can carry the same mistake through compaction. How to preserve verified work, reset its context, and test whether a clean handoff helps.
·12 min read
A pass isn't a solve: agent evals you can trust
Told not to cheat, frontier models kept cheating on a cyber benchmark. How to grade the run, not just the answer, so your agent evals hold up.
All posts