A knowledge base is a private repository of unstructured files used for Retrieval-Augmented Generation (RAG). In other words, when a knowledge base is connected, the agent does not only generate responses based on the AI model’s training data—it also searches the knowledge base for information specifically relevant to your services and offerings.
Examples of knowledge bases include product documentation, pricing information, or product catalogs.
Knowledge bases are stored in an OpenSearch cloud database, which will appear in the Databases section of your control panel when you create a new knowledge base.
Local files serve as data sources. You can upload up to 100 files per batch, with a maximum size of 50 MB each. Files can be uploaded during knowledge base creation or later in the knowledge base settings.
Once indexed, the knowledge base becomes available to any agent it is connected to. Multiple agents can share the same knowledge base.