Knowledge Bases
===============

Knowledge bases give agents access to your organisation's documents and data, enabling retrieval-augmented generation (RAG). Agents query the knowledge base at runtime to ground their responses in specific content rather than relying solely on the underlying LLM.

.. image:: /images/foundry-kb-list.png
   :alt: Knowledge Bases list page
   :align: center
   :width: 100%

|

OneByZero Neo supports two knowledge base types, each backed by a different storage and retrieval engine:

* **In-House** — documents stored in OpenSearch, fully managed within the Neo platform. You upload files, configure chunking and an embedding model, and Neo handles indexing and retrieval.
* **Bedrock** — integration with AWS Bedrock Knowledge Bases. Documents are indexed using Amazon's managed embedding and vector storage service, with S3 as the supported data source.

Switch between the two types using the **In-House** and **Bedrock** tabs at the top of the Knowledge Bases page.

**Reusability**

A single knowledge base can be connected to multiple agents. Update the knowledge base once and every agent that references it immediately queries the updated content — no per-agent changes needed.

**Access control**

Knowledge bases respect workspace permissions. The `canViewKnowledgeBases`, `canCreateInhouseKnowledgeBase`, and `canCreateBedrockKnowledgeBase` permissions control visibility and creation. In-House knowledge bases are only visible when the In-House feature is enabled for the workspace.

**In this section:**

.. toctree::
   :maxdepth: 1

   creating
   document_upload
   chunking
   retrieval_testing
   using_in_agents