Show HN: Agentset – Open-source RAG with vector DB, embeddings, and API built-in (github.com)

Agentset is an open-source platform for building, evaluating, and deploying retrieval-augmented generation (RAG) and agentic applications, offering end-to-end tooling from ingestion and chunking to vector indexing, embeddings, retrieval and production hosting. It’s model‑agnostic (works with any LLM, embeddings provider, and vector DB), includes benchmarks and evals to track quality, a chat playground with message editing and citation support, typed SDKs and an OpenAPI spec, plus built-in multi-tenancy and preview links/custom domains for hosted apps. The project is MIT‑licensed and backed by a generous free tier (1,000 pages, 10,000 retrievals), making it easy for teams to experiment and scale. Technically, Agentset is built with TypeScript, Next.js, an AI SDK, Prisma, Supabase and Trigger.dev, and can be self‑hosted via a short setup (copy .env, pnpm install, pnpm db:deploy, pnpm dev:web). Its combination of production features (hosting, API, SDKs) with observability (benchmarks/evals) and interoperability (any vector DB/LLM) is significant for the AI/ML community because it reduces integration overhead and accelerates shipping of RAG/agent systems while preserving flexibility around model and infrastructure choices. Contributions are welcome, and the repo includes docs and issues for getting started.