Creating Knowledge Bases

A knowledge base is a collection of documents indexed for vector search. Each KB has its own embedding configuration and can be mapped to specific queues or conversation types.

Creating a Knowledge Base Walkthrough

Steps

1. Navigate to Knowledge Bases

Open the Agent Assist admin portal and go to Knowledge Bases in the sidebar navigation.

2. Create a New Knowledge Base

Click Create Knowledge Base and fill in the following fields:

imageCreate Knowledge Base form with name, description, and embedding model fields

The knowledge base creation form

Field	Required	Description
Name	Yes	A descriptive name for the KB (e.g., "Product Support", "Billing FAQ"). This name is used to identify the KB in queue mappings and analytics.
Description	No	A brief summary of what the KB contains. Helps other admins understand its purpose.
Embedding Model	Yes	The model used to generate vector embeddings for document chunks. Options depend on your configured AI integrations.

3. Configure Chunk Settings

Chunk settings control how documents are split before indexing:

Setting	Default	Description
Chunk size	512 tokens	Maximum size of each document chunk. Larger chunks provide more context per result but reduce search precision.
Chunk overlap	50 tokens	Number of tokens shared between adjacent chunks. Overlap prevents important information from being split across chunk boundaries.

TIP

For technical documentation with detailed procedures, a chunk size of 768-1024 tokens works well. For FAQ-style content with short answers, 256-512 tokens is more effective.

4. Save and Add Documents

After saving the KB configuration, you can begin uploading documents and configuring external sources.

WARNING

Changing the embedding model after documents have been indexed requires a full re-index of all documents in the KB. Plan your embedding model choice before uploading large document sets.