# Agentset

An open-source, model-agnostic RAG platform with end-to-end tooling for ingestion, chunking, embeddings and retrieval, a chat playground with citations, multi-tenant hosting and a typed API, available hosted or self-hosted.

Agentset is an open-source platform for building, evaluating and shipping retrieval-augmented generation and agentic applications. It handles the full RAG pipeline end to end: ingesting more than twenty file formats, chunking and embedding content, and retrieving relevant passages, while staying model-agnostic so any combination of LLM, embedding model and vector database can be used.

A chat playground supports testing retrieval with message editing and inline citations, and production deployments get preview links, custom domains, a typed SDK and OpenAPI-documented API, and built-in multi-tenancy for serving multiple customers or projects from one deployment.

It can be used as a hosted cloud service with a free tier, or self-hosted following the project's own setup guide. It is built with TypeScript, Next.js, the Vercel AI SDK, Prisma, Supabase and Trigger.dev, and is released under the MIT license.

Website: https://agentset.ai