Skip to content

Release 9.2.35.1

This release adds a Usage by model view to the AI Governance Monitoring page—rolling up token usage across your models with drill-down to the agents behind each one—and a page-level filter bar that lets you narrow the whole page by agent, model, and date range. The AI Assets Collector also expands coverage for Microsoft 365 and Azure Foundry agents, and surfaces a clear warning when a platform does not expose traces or token usage.

AI Governance

Usage by model on the Monitoring page

Usage by model on the Monitoring page

  • The Monitoring page now includes a Usage by model panel that rolls up token usage across your AI estate, highlighting the top models by tokens with a View all option for the rest.
  • Select a model to expand an inline breakdown by agent—showing agent, owner, tokens, and sessions—so you can see which agents are driving each model's usage.
  • Each model offers an Open in Model Registry shortcut that jumps straight to that model in Inventory → Models. For deeper per-model history, use the Usage tab on the model's details page.
Filter the Monitoring page by agent, model, and date

Filter the Monitoring page by agent, model, and date

  • A new filter bar drives both the Usage by model panel and the agent table together, so everything on the page reflects the same selection.
  • Filter by agent or owner search, select one or more models, and choose a date range—the last 24 hours, 7 days, 30 days, or a custom start and end date.

AI Assets Collector

Microsoft 365 AI agents inventory and traces

Microsoft 365 AI agents inventory and traces

  • The AI Assets Collector now discovers Microsoft 365 AI agents and adds them to the AI Governance inventory.
  • Trace collection is supported for Microsoft 365 AI agents, so agent activity can be reviewed alongside other connected platforms.
Azure Foundry traces and token usage

Azure Foundry traces and token usage

  • The AI Assets Collector now collects traces and token usage for Azure Foundry AI agents.
  • Foundry agent activity and consumption are available in AI Governance for monitoring and governance workflows.
Warning when traces or token usage are unavailable

Warning when traces or token usage are unavailable

  • AI Governance now shows a warning banner when traces or token usage cannot be collected for an agent because the connected platform does not expose those APIs.
  • For example, Microsoft Copilot Studio does not provide traces or token usage APIs, so those metrics are unavailable for agents on that platform. The banner makes this platform limitation clear instead of showing empty or misleading values.