Multi-model AI routing for self-hosted teams.

Genius combines a FastAPI backend, Next.js frontend, PostgreSQL with pgvector, Redis latency tracking, and a model router that can use commercial providers or local Ollama models.

The router checks configured providers, attempts the selected model, follows fallback chains, and records usage so teams can understand cost and reliability.

Document retrieval parses uploaded files, chunks text, generates embeddings, and retrieves relevant context for private workspace search.