Oversight — Reporting

Oversight (bar-chart icon in the sidebar) is the reporting area. It's read-mostly: it shows what happened, while changing the settings behind it stays in Settings. Org admins see it organization-wide; Initiative admins see it scoped to an Initiative they manage (select that Initiative in the Workspace dropdown first). Regular members don't see Oversight at all.

Everything follows the Workspace switcher: with All Initiatives selected an org admin sees every figure rolled up; picking a single Initiative narrows every number on every tab.

Interviews

How the Scout interview programme is running. Four figures across the top: average quality score, total sessions, people needing follow-up, and invitation acceptance rate. Below:

• Completed sessions — the 50 most recently scored interviews; click a row to expand the grader's full rubric, reasoning, and per-axis scores.
• Quality by participant — each person's average across their scored sessions.
• Recurring quality issues — the flags the grader raises most often. A flag that keeps recurring usually means the interview script needs work rather than any one person.
• Needs follow-up — individual invitees who haven't finished. With an Initiative selected, a "Process coverage" link beside it opens Discovery, where the same outstanding invitations are grouped per process.

Assistant usage

How people use the in-app assistant: total chats and messages, satisfaction from thumbs up/down, average messages per conversation; charts of daily volume and busiest hours; daily active users (click a bar to see who was active that day); your Content footprint (storage, files, collections); and the most active conversations.

Governance

Whether AI output is inside policy: what share of everything evaluated is covered by the current policy version, open items, which Initiatives have a policy, which constraints trigger most, and a timeline of recent governance activity (click a row to open the underlying receipt). A Manage policies link goes to Settings.

Review queue

What needs a human decision today, split into Pending re-review under updated policy (items last checked under an older policy version — Re-evaluate re-checks them without overwriting history) and Open reviews (items flagged or blocked with no sign-off yet — Open receipt shows which constraints fired and records your decision). This tab always applies to one Initiative, so even org admins pick one. A count shows beside the tab name when items are waiting. See the AI Governance section for the review workflow.

---

The Interviews and Assistant usage tabs have a Period control (last 7, 30, or 90 days). The review queue used to sit in Settings under AI Governance; that Settings tab now keeps only the policy itself.