RAG
Why your RAG pipeline answers wrong (and it is rarely the chunking)
Teams spend months tuning chunk sizes while the real loss happens one step later. A short guide to finding the stage that is actually costing you accuracy.
BLOG
Notes from production
Short, specific write-ups on the problems that actually show up when AI features meet real traffic. No roundups, no hype.
FUNDAMENTALS
Token budgets are a product decision, not an infra one
The cost per request is set long before it reaches your billing dashboard. It is decided in the product spec, by people who have never seen a token count.
AGENTS
Shrink the agent, grow the tool
Most unreliable agents are not under-prompted. They are over-optioned. The fix is usually to delete tools, not to add reasoning.
EVALS
Write the eval set before you write the prompt
Forty examples beat any amount of prompt intuition. Here is how to build a set small enough to actually write and useful enough to catch regressions.
STRATEGY
What actually breaks when you swap models
Model migrations fail in the same four places every time. None of them are the ones people plan for.