Databricks is expanding the Unity AI Gateway ecosystem with new partner integrations across AI security, identity, observability, and agent governance, helping organizations bring more control and visibility to AI systems as they move into production.
Whatโs new
- Runtime AI security across prompts, responses, and tool calls: Unity AI Gateway supports guardrails, tracing, payload logs, and policy enforcement across AI interactions, and Databricks is adding integrations with partners focused on runtime protection and observability.
- Partner guardrails for real-time protection: Integrations from partners such as Alice, CrowdStrike, and Palo Alto Networks are aimed at inspecting prompts, responses, and MCP or tool activity for risks like prompt injection, unsafe content, and sensitive data exposure.
- Identity governance for AI agents: Unity AI Gateway is also being positioned as a place to extend identity-aware controls to agents and tools, with partner support around agent identity, authorization, and accountability.
- One control layer across AI systems: Databricks describes Unity AI Gateway as a centralized runtime governance layer for models, agents, MCP servers, and AI tools, with controls for access, observability, tracing, guardrails, and cost management.
- Built on Unity Catalog: The gateway extends Unity Catalog governance into runtime AI interactions, so organizations can govern both the assets behind AI systems and the way those systems behave in production.
๐ Read the full post here
