Microsoft Copilot Studio
A low-code platform for building, customizing, and deploying AI agents that can automate tasks, answer questions, and integrate with enterprise data sources across Microsoft 365 and other channels.
Capabilities
- Agent Authentication ControlSecure Agents Users
Sets whether an agent requires user authentication — none, authenticate with Microsoft Entra, or manual — and enforces web channel security, so a published or autonomous agent does not expose corporate data to anonymous users. Reduces unauthenticated-access and data-exposure risk.
- Agent Runtime Protection StatusObserveSecure Agents
Shows makers a per-agent protection status — Protected, Needs review, or Unknown — with security analytics on blocked prompt attacks, authentication gaps, and admin data-policy violations, so misconfigured or unprotected agents are caught before and after publishing. Reduces the risk of shipping insecure, unmonitored agents.
- Tool and Connector User CredentialsSecureGovern Agents Users
Runs an agent's tools, connectors, and flows under each end user's own credentials by default instead of shared maker-provided credentials, and warns makers before publishing if they weaken it. Reduces the risk of over-privileged connectors and broken per-user attribution.