Context Engine
An open retrieval standard with hybrid sparse retrieval, .VCB corpus bundles, and 97% fewer tokens per query.
RoleAuthor
ToolsOpen Standard · Retrieval · MIT License
The problem is retrieval, not generation
Most AI systems treat every query as if the answer doesn't exist yet. It usually does. Throwing more inference at bad data delivery is a physics problem dressed up as a software one.
Context Engine is what I built to prove it: an open retrieval standard with hybrid sparse retrieval, .VCB corpus bundles, and 97% fewer tokens per query. Eight papers, MIT licensed.
The spec, the papers, and the benchmarks all live on the site. That's where this one really wants to be read.