Why 80% of RAG systems fail in production — and how to avoid it
The hard part of RAG isn't the embedding model. It's the chunking, the metadata, and the eval loop. Here's what we learned from 14 production deployments.
Blog
What we're shipping, reading, and learning. Updated when we have something worth saying.
The hard part of RAG isn't the embedding model. It's the chunking, the metadata, and the eval loop. Here's what we learned from 14 production deployments.
Five guardrails we now ship with every agent — based on production incidents we wish we'd caught earlier.
You don't need a platform team of 20. Here's what we ship for early-stage teams that still want production-grade monitoring.
Boards don't want a 40-slide deck anymore. They want a 90-day execution plan with measurable outcomes.
Get these posts in your inbox.
Subscribe to the newsletter