Databricks' Unity AI Gateway Now Governs Every Agent Touching Your Data
Databricks' Unity AI Gateway hits GA with hard spend caps, smart routing, and unified governance across agents, MCPs, and coding tools — over a quadrillion tokens already processed.
- GA launch: Unity AI Gateway is now generally available, giving enterprises a unified control plane for AI cost, security, and model access.
- Scale: Over a quadrillion tokens have passed through the gateway in the past year; thousands of customers including Rivian, Asana, and Edmunds are already live.
- Core features: Hard spend caps, PII guardrails, prompt injection protection, MCP server governance, agent tracing, and multi-model routing via a single API.
- Smart Routing (Beta): Dynamically routes requests to the right model based on cost, quality, and budget — reserving expensive models only for tasks that need them.
- Competitive context: Puts direct pressure on standalone AI gateway vendors (LiteLLM, Portkey, Helicone) and platform rivals Snowflake and Palantir building similar solutions.
- What's next: Service policies and agent services remain in Beta; a governance webinar with CTO Matei Zaharia is scheduled for August 13.
Enterprise AI has a sprawl problem. Most large organizations are no longer running one or two models in production , they're managing fleets of coding agents, MCP servers (the emerging standard for connecting agents to enterprise tools), custom bots, and third-party AI apps, all pulling from sensitive data and racking up token bills with no central oversight. Databricks' Unity AI Gateway is now generally available, and it's positioning itself as the single control plane to govern all of it.
The problem that made this inevitable
AI is becoming increasingly multi-model, multi-agent, and multi-vendor. Developers are adopting coding agents, while business users are interacting with enterprise data through AI experiences. Many enterprises are also launching custom agents to automate critical internal workflows. The result is what the industry is calling "agent sprawl" , and it's a governance nightmare.
Three compounding problems are driving urgency here:
- Runaway costs: Organizations are grappling with exponentially growing AI spend, driven by consumption-based token pricing rather than user seats. There's no native cost cap in most AI tools, and no single view of the full bill.
- Security and IP risk: Agents must be trusted with sensitive data to be useful, but they can expose PII and be vulnerable to prompt injection, and agent traces are often full of confidential information that gets inadvertently stored or leaked.
- Vendor lock-in: Organizations are attempting to manage multiple AI tools and thousands of custom agents across disjointed platforms, making it nearly impossible to swap models or adopt new tools without rebuilding governance from scratch.
What Unity AI Gateway actually does
Unity AI Gateway is Databricks' centralized runtime governance layer for AI agents, built as an extension of Unity Catalog. Where traditional governance answers "who can access this asset," the Gateway answers a different question: "what is an AI system permitted to do during a live interaction." That's a meaningful distinction , it's not just access control, it's behavioral control at runtime.