rasa-setting-up-enterprise-search

To add knowledge base search to your Rasa assistant, configure the EnterpriseSearchPolicy in your config.yml. This enables the assistant to retrieve relevant documents and generate direct answers from your knowledge base.

Yes, Rasa supports connecting to vector stores like Faiss, Milvus, and Qdrant. The EnterpriseSearchPolicy integrates with these stores to retrieve relevant documents for generative or extractive answer generation.

The best way to implement RAG in Rasa is by overriding pattern_search using the EnterpriseSearchPolicy. This configures the assistant to search connected vector stores and retrieve relevant information for custom retrieval implementations.

Configure EnterpriseSearchPolicy in your config.yml to enable both generative and extractive search capabilities. Point it to a vector store like Faiss with documents in your local directory to start retrieving answers.

Yes, you can use Rasa to search a documentation knowledge base and provide direct answers. By integrating EnterpriseSearchPolicy, the assistant searches connected vector stores instead of relying solely on predefined conversational flows.

You need to override pattern_search for Rasa implementations when your assistant must query a vector store for custom information retrieval. This enables dynamic knowledge base search for answering user questions outside predefined flows.