We're building the information layer for AI
AI applications need reliable, structured, citeable information from the open web. Fetching that information reliably — at scale, with privacy, across diverse sources — is a hard engineering problem that most teams shouldn't have to solve themselves.
Fetchium is the API that solves it. 17 novel algorithms, 17 backends, built in Rust. From individual developers building their first RAG prototype to teams running production AI pipelines — one API handles every retrieval workload.
What we stand for
Technically honest
We aim to keep public product claims tied to current code, docs, and verifiable configuration.
Privacy by default
Zero telemetry unless you opt in. Your queries are yours.
Open infrastructure
Standards-based (MCP, REST, CLI). Designed to be easy to integrate without locking your application to a proprietary workflow.
Built to last
Written in Rust. 1,100+ tests. Zero warnings policy. Production-grade resilience with circuit breakers, retries, and graceful degradation.
By the numbers
Architecture
fetchium-coreAll 17 algorithms: search orchestration, CEP extraction, HyperFusion ranking, QATBE, AMRS, PIE, intelligence
fetchium-cli26-command CLI: search, research, compare, doctor, provider, export, and more
fetchium-mcpJSON-RPC 2.0 MCP server with 12 tools for Claude Desktop, Cursor, and other MCP clients
fetchium-apiaxum 0.7 REST API server with PostgreSQL, authentication, and rate limiting
apps/webMarketing site and documentation hub
apps/dashboardAPI key management, usage analytics, playground