Mayank works at Colrows on how enterprise AI agents get the context they need to be reliable. He writes about the structural difference between a wiki, a vector store, and a real company brain - and about why Model Context Protocol works only when something on the other end can actually answer typed questions about meaning.

Posts follow the Colrows editorial policy: technical claims are reviewed against the architecture, and citations point to primary sources (Anthropic on MCP, Y Combinator on the company brain, and so on).

Areas of focus

Company brain Model Context Protocol Agent context engineering Institutional memory Enterprise AI

Posts

All posts All authors