Build your first agent with a system prompt

Create a virtual model, define behavior via prompt, connect RAG, and validate via API.

Create virtual model

1. Create a virtual model with a stable slug.
2. Link it to your target deployment.
3. Verify API reachability.

Define system prompt

1. Define role and objective clearly.
2. Set response style and boundaries.
3. Define fallback behavior for missing context.

Connect RAG

1. Assign a knowledge base.
2. Test retrieval quality.
3. Validate responses against reference questions.