JamAIBase

Why we included this project

JamAI Base bundles a self-hosted backend for LLM features so you don't have to wire together a database, a vector store, and an orchestration service yourself. It ships an embedded SQLite database alongside LanceDB for vectors, with managed memory, RAG pipelines, and LLM and reranker orchestration all exposed through one REST API and a spreadsheet-style interface. The four table types each cover a distinct job: generative tables auto-fill columns with model output, knowledge tables hold documents for retrieval, action tables handle live frontend-to-backend interactions, and chat tables run conversational features. That makes it easy to prototype an AI application visually and then carry the same setup into production, which suits small teams that want to iterate quickly.