Agents

Budibase Agents let you combine model reasoning with workspace tools and automations.

Use this section to go from initial setup to production-ready Agent workflows.

Recommended path

1. Agents config
   Connect and validate your provider and default model, then attach a model to your agent.
2. Agent chat
   Configure chat deployment, agent availability, access role, and conversation starters.
3. Messaging channels
   Deploy the same live agent to Slack, Microsoft Teams, or Discord after Agent Chat is validated.
4. Agent building 101
   Build a complete service desk Agent with tools and automation handoff.
5. Agent instructions guide
   Write clear, safe, and deterministic instruction prompts.
6. Agent tools and permissions
   Apply least-privilege tool access and safe write guardrails.
7. Agent embedding models
   Configure embeddings for Agent knowledge retrieval.
8. Agent vector databases
   Connect and assign vector storage for semantic retrieval.
9. Agent testing guide
   Run repeatable tests and regression checks before production rollout.
10. Agent troubleshooting
    Diagnose model, tool, and output issues quickly.

What this section covers

• Agent setup and model configuration
• Prompt and output design patterns
• Tool access and safety controls
• Embeddings and vector database setup
• Testing strategy and release readiness
• Troubleshooting and operational support

If you are setting up Budibase AI for the first time, see Quickstart: Budibase AI.