# Agent observability: definition and example

URL: https://aiagentsecuritysolutions.com/glossary/agent-observability/
Updated: 2026-11-16

Agent observability is the collection of prompts, responses, tool calls and outcomes that show what an agent did and why. It supports audit, investigation and cost control. It records activity; it does not by itself stop an action.

## Example

Operant says Microsoft Purview already records Copilot interactions for audit and compliance, and that its integration turns that record into security telemetry attributed to the user, app surface and resources referenced.

## Sources

- [Operant: coverage for Microsoft Copilot and embedded AI assistants](https://www.operant.ai/art-kubed/announcing-operants-coverage-for-microsoft-copilot-and-every-ai-assistant-embedded-in-microsoft-365) (operant.ai), read 2026-11-16

## Related pages

- [Agent trace](/glossary/agent-trace/)
- [Workspace agent](/glossary/workspace-agent/)
- [Guardrail policy](/glossary/guardrail-policy/)
- [AI Agent Security Metrics and Monthly Reporting](/guides/ai-agent-security-metrics-and-reporting/)
- [All glossary terms](/glossary/)
- [Agent lifecycle](/glossary/agent-lifecycle/)
- [Agent orchestration](/glossary/agent-orchestration/)
- [Guides](/guides/)
