Unity AI Gateway Observability: Models, MCP Tools, and Coding Agents

data_pulse — Thu, 29 Oct 2026 17:57:09 GMT

Databricks Unity AI Gateway provides a unified way to monitor and understand AI activity through its built-in Workspace AI Gateway Usage Analytics dashboard and the underlying system.ai_gateway.usage system table.

Some useful capabilities include:

Model usage and performance: Track invocations, token consumption, latency, streaming responsiveness and errors.
Spend insights: Understand model consumption and available cost information.
Request attribution: Use request tags to analyse usage across projects and workloads.
Prompt caching: Examine cached input-token utilization.
MCP observability: Monitor tool invocations, service activity and performance.
Coding-agent observability: Gain visibility into supported coding-agent activity, sessions and model consumption.

One interesting observation: MCP usage records include protocol operations such as initialize and tools/list, not just actual tools/call executions. So counting every MCP event as a tool invocation can overstate the number of tools actually executed.

😄 Turns out, not every MCP event means a tool got to do some actual work. Some were just saying hello (initialize) or checking who's available (tools/list)!

The built-in dashboard provides a convenient starting point, while the system table enables deeper SQL analysis and custom reporting.

I've put together an article covering these capabilities, with dashboard screenshots, SQL examples and practical observations.

📖 Exploring Databricks Unity AI Gateway Observability

Would be great to hear how others are leveraging Unity AI Gateway observability for their AI workloads.

Re: Unity AI Gateway Observability: Models, MCP Tools, and Coding Agents

data_pulse — Tue, 03 Nov 2026 19:59:55 GMT

@jimcarre

system.ai_gateway.usage provides quite useful coding-agent session metadata, including the agent name/version, session ID, model, token consumption, latency and request status etc.

For eg, in my OpenCode validation, I could correlate all successful model requests to the same coding-agent session and follow their timing and token usage using SQL. So it doesn't require complex joins just to understand model activity within a session filtering by session_metadata.client_session_id and ordering by event_time is a good starting point.

The distinction is usage tracking vs. execution tracing. The usage table doesn't provide a complete step-by-step transcript of prompts, file edits, local shell commands or tool results. MCP tool calls routed through Gateway can also appear as separate usage records, but complete correlation depends on the metadata available.

For deeper model/tool execution visibility, Databricks also offers unified tracing (Beta), which can combine Gateway traces with supported client-side agent traces when enabled.

topic Re: Unity AI Gateway Observability: Models, MCP Tools, and Coding Agents in Get Started Discussions

Unity AI Gateway Observability: Models, MCP Tools, and Coding Agents

Re: Unity AI Gateway Observability: Models, MCP Tools, and Coding Agents