AI governance with Unity AI Gateway

Unity AI Gateway is the Databricks governance solution for enterprise AI. Built on Unity Catalog, it extends governance beyond your data and AI assets to the runtime interactions between models, agents, MCP servers, and tools. Control which AI services teams can use, route and manage AI traffic, set guardrails, and monitor usage from one control plane.

Databricks serves large language models (LLMs) natively through Foundation Model APIs, so you get pay-per-token access to models such as GPT, Claude, and Gemini with no infrastructure to run. You can also connect external model providers and route to their APIs through the same access control and traffic management.

Get started

Set up and apply AI governance across your AI assets, traffic, and service behavior.

• • AI governance guide
  • An end-to-end path for administrators to govern access to AI assets, traffic and cost, and the content of requests and responses.
• • Tutorial: Govern a coding agent's GitHub MCP access
  • Restrict a coding agent's access to GitHub MCP tools using Unity Catalog permissions and a built-in service policy.
• • Tutorial: Implement guardrails on a model service with service policies
  • Implement guardrails on a model service using built-in and custom service policies.
• • Tutorial: Track foundation model spend by user, team, or project
  • Identify LLM spend drivers by attributing cost to users, teams, or projects with service tags and request tags.

Control access to AI services

Register AI assets as Unity Catalog securable objects, then grant and revoke access with the same privileges you use for tables and volumes. Agents are governed through these same securables: an agent is registered as a Unity Catalog model, and the tools it calls are governed as MCP services, functions, and connections.

• • Model APIs
  • Query foundation models served natively in system.ai with no setup, and govern access through Unity Catalog privileges.
• • External model providers
  • Connect providers such as OpenAI and Anthropic with bring your own key (BYOK), and govern them through one control point.
• • Models
  • Govern registered models, including Databricks-served foundation models, with Unity Catalog privileges.
• • Foundation model permissions
  • Restrict which Databricks-served foundation models your organization can access, account-wide or per group.
• • MCP tools
  • Govern MCP servers registered as Unity Catalog securables, with tool filtering and service policies.
• • Custom tools
  • Govern the Unity Catalog functions that agents use as tools, with the same privileges you use for data.
• • HTTP connections
  • Govern the Unity Catalog connections used to reach external APIs and MCP servers.
• • Create custom model APIs
  • Optional: Add your own governed endpoints when you need models beyond the ready-to-use system.ai defaults.

Route and manage AI traffic

Unity AI Gateway routes requests to your model and MCP services from a central control plane, so you can manage capacity, availability, and spend across providers.

• • Apply rate limits
  • Enforce consumption limits on model services and MCP services to manage capacity and cost.
• • Configure traffic splitting and fallbacks
  • Distribute requests across multiple model destinations and add failover to increase availability.
• • Manage budgets
  • Monitor spend and set per-user thresholds and hard caps.

Set guardrails and access policies

Service policies, also called guardrails, control how each request and response proceeds, based on its content and on who is making the call.

• • Service policies for AI securables
  • How service policies govern the content of requests and responses to AI services, using built-in and custom policies.
• • Create and attach a service policy
  • How to write a service policy function and attach it to an MCP service or model service.

Monitor usage, cost, and risk

Track activity, spend, and outcomes across all Unity AI Gateway services.

• • Observability overview
  • Choose the right tool to monitor AI usage, cost, performance, and activity across services.
• • Monitor usage
  • Track requests, token usage, and latency across Unity AI Gateway using system tables.
• • Analyze cost
  • Attribute Databricks cost to services, target models, principals, and tags.
• • Audit requests and responses
  • Log requests and responses to Unity Catalog Delta tables for monitoring and debugging.