Ragpi

Ragpi is an open source AI assistant API that answers technical questions from sources a user configures. It builds knowledge bases from documentation websites, GitHub issues and repository README files, stores them in a vector database, and uses an agentic retrieval augmented generation approach to fetch relevant documents for each query. It is aimed at developers and teams who want a documentation question and answer service they can run themselves and call over REST. The workflow has three steps. A source is created through the sources endpoint with a connector, such as a sitemap connector that points at a documentation site. Synchronization of documents then runs automatically and can be monitored through the tasks endpoint. Finally the chat endpoint takes a list of sources and messages and returns an answer. The assistant can also be used through Discord or Slack integrations. Documented features include hybrid search that combines semantic and keyword search with Reciprocal Rank Fusion, response generation through a selected LLM provider, flexible connectors for docs, issues and READMEs, and basic OpenTelemetry tracing. The project is open source and documented at docs.ragpi.io. No pricing is published, and running it involves the user's own infrastructure and chosen LLM provider.