AI Activity Timeline is the chronological view of everything the AI fleet did. Pages, decisions, tool calls, verifications, approvals, kills, demotions. Filter by tenant, agent, severity, time. Export for audit. The single page that answers "what did the AI do?" without you having to assemble the answer.
Six event kinds. Page (a signal arrived). Decision (an agent decided what to do). Action (a tool was called). Verification (an outcome was checked). Approval (a human signed off). Kill (a kill switch was pressed). Each kind has its own color so a glance tells the story without reading every row.
Five filter facets you can stack: tenant (which tenant's activity), agent (which agent), severity (sev-1 only), time (last hour, last 24h, custom), and kind (only verifications, only kills). Saved filters cover common operator queries: "yesterday's sev-1 activity" or "this week's human approvals."
Every row links to its Decision Bundle. Click and see the prompt, response, tool calls, and verifier outcomes for that one event. The timeline is the map; the bundle is the territory. Useful for audits ("show me the prompt behind this decision") and postmortems ("what was the agent thinking at 14:22?").
For SOC2/HIPAA/ISO reviews, export the filtered timeline as a CSV plus a signed Decision Bundle archive. Signature included so the auditor can verify nothing was edited in transit. Default windows: this quarter, last quarter, this year. Custom windows supported.
Subscribe to Nova AI Ops on YouTube for demos, tutorials, and feature deep-dives.
When auditors ask, the activity timeline is the answer. When ops asks, it is the same answer. One source.