Retrieval-Augmented Generation (RAG)

Intric uses the Retrieval-Augmented Generation (RAG) technique to provide answers based on information stored in Knowledge, the platform’s internal knowledge source. When a user asks a question to the assistant, a semantic search is performed in Knowledge to identify the most relevant text passages in the uploaded or integrated data.

This search happens via a vector database (PGVector), where all content has been converted to vectors to enable similarity-based matching. The selected text parts are used as context for the language model, which then generates a response.

The RAG process ensures that answers are not based on general knowledge, but on the organization’s specific content. The result is more relevant, accurate and reliable answers in the assistant—directly based on the uploaded data source in Knowledge.