Show HN: RagPack – Lightweight self-hosted RAG infra for startups (github.com)

RagPack, a new open-source, self-hosted RAG (Retrieval-Augmented Generation) infrastructure, has been launched specifically for early-stage startups. This lightweight solution allows developers to effortlessly ingest documents from various sources—such as URLs and file uploads—while automating the processes of fetching, parsing, chunking, and embedding. RagPack supports semantic search capabilities where users can query collections using natural language, returning relevant chunks that can seamlessly integrate with any language model (LLM) prompt. The significance of RagPack lies in its accessibility and efficiency, requiring minimal infrastructure—easily deployable on platforms like a $5 VPS or a Mac Mini—making it an attractive option for startups looking to incorporate advanced AI features without incurring hefty operational costs. The framework employs modern technologies, including Go, Fiber, and SQLite, with integration options for embedding models like Ollama and OpenAI, ensuring flexibility for various operational needs. Developers can manage their setup through a REST API or a user-friendly admin UI, drastically simplifying the process of adding semantic search and RAG functionality to existing applications without the complexities of infrastructure management.