Governance and audit evidence for Copilot agents and Power Platform flows.
Copilot Studio agents and Power Platform flows act across your Microsoft 365 environment. Route their calls to external systems through Panopticore to enforce policy and produce evidence auditors and insurers can verify independently.
Copilot agents are acting. Audit teams can't prove what they did.
Agents are calling tools
Copilot Studio agents execute tool calls against Microsoft 365 data, and Power Platform flows take actions across systems. Those calls move data and change systems.
Audit evidence doesn't exist yet
Most organizations cannot show what an agent did, which policies were in effect, or whether those policies were enforced. Operational telemetry was not built to answer those questions for an auditor.
Where Panopticore sits in the Microsoft stack.
third-party
services
Panopticore intercepts egress, applies policy, routes approvals when needed, and records tamper-evident evidence.
What happens when policy says no.
Copilot agent attempts an action that violates policy, e.g., calling an external API with production data.
Panopticore evaluates the request against policy, blocks it with a 403, and records the decision, the reason, and the policy that triggered it.
If the policy requires approval, Panopticore routes the request to an approver in Slack. Only a signed approval token, checked at execution time, lets it through.
The full session is recorded in a cryptographically signed Evidence Binder: identity, action, policy decision, approval chain, and verification artifacts.
HTTP/1.1 403 Forbidden
Content-Type: text/plain; charset=utf-8
request denied by policyHow Panopticore differs from Microsoft's Agent Governance Toolkit.
Microsoft's README says the Agent Governance Toolkit "enforces governance at the application middleware layer" and that "the policy engine and agents share the same process boundary." It hooks into agent frameworks through their extension points: LangChain callback handlers, CrewAI task decorators, and the Microsoft Agent Framework middleware pipeline.
Panopticore operates at the network layer in a separate trust boundary from the agent and requires no framework changes. The Evidence Binder is self-contained and offline-verifiable by any third party.
Both are useful; they answer different questions.
| Panopticore | Agent Governance Toolkit | |
|---|---|---|
| Trust boundary | Separate from the agent | Same process as the agent |
| Integration | Network layer, no SDK changes | SDK callbacks, framework decorators |
| Audit evidence | Evidence Binder, produced outside the agent's process | Signed records, produced inside the agent's process |
| Enforcement | Deterministic block/allow/approve | Policy evaluation in-process |
We made the same argument to NIST.
From our formal comment on AI agent security:
“The governance mechanism and the entity being governed share the same trust boundary. This is analogous to relying on an application to enforce its own access controls without an external authentication service, a pattern that the security community abandoned decades ago.”
Microsoft ecosystem partnerships
Panopticore works alongside Microsoft's own governance tools. We're exploring partnerships with Microsoft Inner Circle partners, system integrators, and ISVs building on Copilot Studio and Power Platform. If your clients need governance for their agent deployments, get in touch.
Partner inquiries →“Panopticore delivers the controls and evidence that compliance teams, auditors, and often insurers need before agents can move from pilot to production.”
Further reading
An architectural read of Microsoft's open-source Agent Governance Toolkit, and how it differs from Panopticore and Microsoft Agent 365.
Pre-deployment review, model evaluation, and prompt engineering all happen before the agent acts. Runtime governance happens when it acts.
Start with one Microsoft workflow.
Request early access with your Microsoft scenario and we'll scope a design partner engagement.