Alex·an·dria
/ˌæl·ɪɡˈzan·drɪ·ə/
A knowledge library for AI agents.
/ˌæl·ɪɡˈzan·drɪ·ə/
A knowledge library for AI agents.
Alexandria is a knowledge library for AI agents. An agent sends a query and gets back a structured, cited answer — not a list of document links to parse.
Every entry has a type, a confidence rating, and a source citation. Conflicting claims between sources are flagged explicitly. The agent reads once and builds on what it learns.
Web search returns documents. The agent parses HTML, reconciles contradictions, and guesses at reliability. That cost is invisible because there was no alternative.
We return a resolved, deduped, cited answer. We also answer questions no search engine can: how is A connected to B? (pathfinding), what is the relationship between X and Y? (comparison), and what are other agents asking about? (trending).
Price: $0.001–$0.006 per call depending on complexity. No signup. No API key.
Pathfind. Ask "how is momentum investing connected to CAPM?" and get the shortest path with labeled hops. No search engine can do this — none has a typed graph.
Compare. Take two entities and get direct edges, shared neighbors, belief differences, and provenance overlap in one call.
Trending. The telemetry is a product feature. See what agents are asking about right now: top types, tags, routes, and volume.
These are graph operations. Document search cannot do them. The graph covers 334 entries and 440 edges across quantitative finance and ML, with automated ingestion keeping it fresh.
What we are building, and why.
The library is live. Agents query it over MCP or the HTTP API. Dev token: alex-dev-token.
No signup. No API key. Payment is the credential.