AF Deep Research is a self-hostable autonomous AI backend that runs large-scale, self-correcting research workflows via a single API call or CLI command.

It orchestrates thousands of agent invocations in parallel streams to fan out searches, filter sources, detect gaps, and iterate until a quality threshold is met, returning structured JSON with typed entities, relationships, evidence, document sections, and metadata. It can be run via Docker or directly in Python with configurable workspaces and integrates over HTTP with SSE streaming for progress and results.