Create a RAG Agent

RAG is the fastest way to make an agent useful on your own data without training a model.

Goal

Create an agent that:

• answers from your documents
• cites or grounds responses through retrieval
• stays focused on a specific domain

Step 1: Create a Knowledge Base

In the dashboard:

1. Open Knowledge Bases
2. Create a new knowledge base
3. Give it a clear scope such as “Support Docs” or “Internal Policies”

Step 2: Ingest Documents

Upload a small, clean set first:

• product docs
• runbooks
• FAQs
• internal procedures

Avoid dumping everything in on day one.

Step 3: Create or Open an Agent

Create a new agent or edit an existing one. Use a system prompt that explicitly tells the agent to rely on retrieved knowledge when answering domain-specific questions.

Step 4: Attach the Knowledge Base

Associate the knowledge base with the agent in the dashboard.

Step 5: Test Retrieval Behavior

Ask questions that should be answerable only from the uploaded material.

Look for:

• grounded answers
• stable domain behavior
• fewer hallucinated specifics

Step 6: Refine

If quality is weak, adjust in this order:

1. source quality
2. document scope
3. chunking and retrieval setup
4. prompt guidance

Related Pages

• Knowledge Bases
• Document Processing
• Advanced RAG