How to read a trace
A practical guide to finding the failures your agent's answers hide, and turning them into evals.
Antaripa Saha
Loading blog…
The neatlogs blog.
Latest story
All stories
4 stories to explore
How to read a trace
A practical guide to finding the failures your agent's answers hide, and turning them into evals.
11 min read
Self-healing is self-defeating
How an agent that fixes every mistake can slowly erase your strategy.
9 min read
Your strongest LLM judge should grade the fewest traces
How to make LLM evals cheaper without actually making them worse
16 min read
A short history of machines that fix themselves
What a century of automation can teach us about self-healing agents, and the work that refuses to disappear.
11 min read