Agentic AI Platform Change Log

Agent capability

Agent accounts in Copilot Studio let agents work in Microsoft 365 with their own credentials, so people can assign them work in Teams and Outlook, mention them in Word, Excel and PowerPoint comments, and hand off tasks the agent continues between interactions. Tools run under the agent's own approved credentials and scoped permissions rather than a user's or maker's, and agents keep business process memory across the people they work with.

Agent actions become attributable to a governed identity that admins can scope and manage, which matters for processes that outlive any one employee. Microsoft says more access policy, cost and lifecycle controls are planned.

Human approval / guardrails

Microsoft's September 23 release notes add Outlook actions that let Microsoft 365 Copilot delete, move, copy, and categorize email and manage inbox rules and folders through natural language. Copilot requests confirmation before acting on more than five messages or changing inbox rules. The capability covers classic and new Outlook for Windows, web, Mac, and mobile.

Teams evaluating delegated mailbox management gain a defined approval boundary for bulk actions and rule changes. Administrators can assess whether those confirmation requirements fit their operating policies before expanding use.

Observability / auditability

Microsoft introduced new data export capabilities for the Copilot and Agent 365 Dashboards, allowing administrators to export de-identified, row-level metrics. The Consumption Dashboard was also updated to support tracking GitHub Copilot AI Credit usage.

Administrators gain deeper visibility into agent usage and user-level interactions without compromising privacy, enabling better budget management and decisions for AI investments.

Agent capability

Microsoft introduced an AI agent that automates the generation of unit tests for developers. The agent builds the workspace, runs the test suite, checks assertion specificity, and performs mutation testing to ensure test validity before integrating into CI/CD pipelines.

Engineering teams can accelerate software delivery and improve code quality by offloading the creation and validation of unit tests to an autonomous agent.

Security / enterprise

Microsoft expanded its Zero Trust for AI strategy with new tools and guidance specifically designed to secure AI agents and DevSecOps pipelines. The update provides frameworks for verifying agent behavior and protecting underlying data before agents reach production.

Security and compliance teams gain native tools to govern AI agent deployments, ensuring that autonomous workflows adhere to enterprise Zero Trust policies and do not expose sensitive data.

Agent capability

The GitHub Copilot CLI reached general availability, turning the command line into a terminal-native coding agent that can autonomously run multi-step workflows inside the developer's own environment.

Extending Copilot from the IDE into the terminal opens autonomous code generation and infrastructure tasks at the command line, where much real engineering work already happens.

Observability / auditability

Microsoft took the Azure Copilot Observability Agent to general availability. Built into Azure Monitor, it correlates logs, metrics, traces, and operational context to autonomously investigate cloud incidents and surface root causes.

Autonomous incident investigation moves SRE and IT teams off manual dashboard triage toward contextualized root-cause analysis, directly targeting mean time to recovery.