Cortex
Compiles organizational knowledge into verified, queryable intelligence.
In the constellation
Cortex highlighted in the live map — hover or tap any node to explore.
Why it exists
An organization’s knowledge is buried in years of files, repos, and documents — duplicated, inconsistent, and impossible to query straight. Search gives you back a pile of documents and leaves the reading to you. A chatbot gives you an answer, but you can’t tell where it came from or whether it’s true.
What you actually want is a system that knows your organization better than anyone — and can prove every answer it gives.
What Cortex is
Cortex is a truth compiler. It reads your organization’s raw material and compiles it into verified, structured facts — each one tagged, dated, and carrying a trail back to its source. Ask a question, and you get an answer grounded in those facts, with citations. If no verified fact supports a claim, Cortex doesn’t make it.
The operating principle is “AI proposes, the engine decides.” AI does the reading and the drafting; a verification layer checks every output against the source material before it’s stored or returned. The result is an answer system whose answers are defensible — not a guess wearing a confident tone.
What it does
Verified facts, not documents
Compiles prose into structured, verified facts with provenance — so a query returns an answer, not a stack of files to read.
Grounded answers with citations
Every synthesized answer is backed by verified facts and cites them. Click any claim back to the source that supports it.
Point-in-time truth
Tracks both when something was true and when it was learned — so you can ask what was known then versus what’s known now.
Audited synthesis
A separate adversarial pass challenges each answer for accuracy before it’s returned — the AI is checked at both write-time and read-time.
The line between judgment and machinery
AccelMars draws one hard line through every product: what an AI decides, and what runs deterministically. Cortex sits on the boundary — and keeps the two honest.
What the AI does
- Extract structured facts from prose
- Synthesize a grounded answer from them
What the engine guarantees
- Verifies every fact before it’s stored
- Provenance, citations & dedup
- Deterministic storage and retrieval
The verification layer is the point: AI is proposed-from, never trusted. Hallucinations and unsupported claims are caught by the verifier — not papered over with a different prompt.
Live in production — two pipelines (compilation and query) running against thousands of compiled facts. Composition into the unified Platform API is the next step.