Introduction
Ground your Agents in your own documents, spreadsheets, and web content with Elizon's RAG system.
A Knowledge Base is a collection of your own documents, spreadsheets, and crawled web pages that Elizon chunks, embeds, and indexes so an Agent can search it and answer questions grounded in your actual content — a manual, a policy file, a product catalog — instead of only what the model already knows.
How this fits together
- Create a Knowledge Base — pick an embedding model and a search mode. See Creating a Knowledge Base.
- Add content — upload files or submit URLs to crawl. Elizon chunks, embeds, and indexes each one automatically. See Documents and Web ingestion.
- Bind it to an Agent — an Agent with one or more Knowledge Bases attached can search them at run time. See Agents and Knowledge.
Studio-only — not part of the public API
Everything in this section is managed through Studio → Knowledge. Knowledge Base management
runs as its own service (knowledge-service), authenticated by your Studio session — there is no
X-API-Key-authenticated public endpoint for creating, uploading to, or searching a Knowledge
Base directly. The one thing that is reachable through the public API is indirect: running an
Agent that has Knowledge Bases bound to it — the agent searches them
internally as part of answering your prompt.
Where to go next
- Creating a Knowledge Base — embedding models, search modes, advanced index configuration.
- Documents — supported file types, chunking strategies, limits, per-file operations.
- Web ingestion — crawling a single page, a whole site, or a list of URLs.
- Agents and Knowledge — binding a Knowledge Base to an Agent, Agentic RAG, metadata filters, and HyDE retrieval.