This is a live, shared instance of TekmerDB. Name and launch your own scoped engine below, then insert some data and watch it score confidence and catch contradictions between sources automatically — mechanically, not guessed by an LLM.
TekmerDB is domain-scoped — it has to be told upfront which EU AI Act risk category it's operating in (Healthcare, Finance, Employment, and so on). Every claim you submit is checked against a similarity gate for that domain, and anything that drifts outside it gets rejected. That keeps confidence scoring meaningful and specific to the regulatory context, instead of one fuzzy pool of unrelated facts.
Once your instance is running, insert data by uploading a PDF, pasting raw text, or hand-crafting a single claim with its own confidence and source. Every insert is checked for corroboration or contradiction against what's already there. To retrieve, use semantic search or list all sources right in the UI — or connect any MCP client straight to your instance and do the same thing programmatically. Give it a try.
| Name | Domain | Alive | |
|---|---|---|---|
| … | |||
This instance will automatically shut down after 60 minutes, ending your session.
Upload a small PDF or paste text and TekmerDB will extract claims and score confidence mechanically — not guessed by an LLM. When two claims conflict, both are flagged, not silently resolved.