Harshit works at Colrows on the boundary problems - the places where the semantic execution layer meets agents, retrieval pipelines, and the analytics tools enterprises already have. He writes about why RAG is necessary but not sufficient for enterprise AI, how self-serve analytics survives a CISO review, and why the control plane has to enforce policy at compile time rather than after a query has run.

Posts follow the Colrows editorial policy.

Areas of focus

RAG vs semantic layer Self-serve analytics Semantic search Semantic control plane Compile-time governance

Posts

All posts All authors