A curated, quality-gated medical evidence knowledge base — not an index over everything that exists, but a governed base only sources meeting a quality bar enter. It's what grounds the analysis behind Agent HERO and every other domain agent on the platform.
Most retrieval systems are built to cover as much as possible and leave relevance to query time. MeSI is built the other way: a document has to meet a defined quality bar — provenance, currency, source credibility — before it enters the knowledge base at all. The result is a smaller, more trustworthy base than a generic web-scale index, and one every agent that queries it can cite with confidence rather than caveat.
Peer-reviewed literature and open-access full texts; clinical trial registrations and published trial reports; real-world evidence studies; regulatory, HTA, and payer documents; post-market safety and adverse-event data; and pricing, reimbursement, and utilisation data. Coverage spans 975 curated sources across 26 countries and the EU today, growing as new markets and therapeutic areas are onboarded.
MeSI doesn't produce outputs on its own — it's the evidence substrate the platform's agents reason against. Agent HERO queries it to source and justify every parameter in a health-economic model; the platform's other domain agents draw on it the same way for regulatory, medical affairs, and pharmacovigilance work. See the module registry for how it fits into the wider orchestration engine.
Where it runs: MeSI deploys the same way as the rest of the platform — sovereign cloud, on-premise, or air-gapped, inside your own environment. Nothing in the knowledge base trains a public or foundation model.
Most retrieval tools index everything available and let the model sort out relevance at query time. MeSI filters at ingestion — a document has to pass a quality bar before it ever enters the knowledge base.
Coverage today spans 975 curated sources across 26 countries and the EU, growing continuously. Ask about your specific market and indication — that's exactly what a walkthrough is for.
The composition of the knowledge base is part of what makes it valuable, so it's walked through live with serious evaluators rather than published as a full manifest.
Yes — sovereign cloud, on-premise, or air-gapped, inside your own infrastructure, like the rest of the platform.
Our proprietary AI orchestration platform for healthcare: one engine, a registry of reusable task modules and domain agents, and a governed knowledge base — deployed inside your walls and run by your team.