Our open RAG tools

Every RAG pipeline consists of a vector database and retrieval tools. We created a library - RAGpy to provide an abstraction layer over different vectorstores (currently only Qdrant is supported). It allows you to create a collection of embedded text chunks and query data.

RAGpy allows to use different types of embedding models - third-party embeddings like OpenAI's or local models.

Use-cases of RAG for AI agents

RAG is widely used to reduse hallucinations and support LLM's output with reference information. It can be utilized by agents on our platform, specifically by adding MCP tools performing queries against a vector database. Read more about our MCP development kit if you want to build such tools.