Skip to content

AI-powered virtual executive team — a single coherent executive persona backed by 8 specialist agents (FastAPI + Next.js).

4.2 KB
Raw Blame
{
  "section_id": "review",
  "title": "Review Queue",
  "markdown": "**What it is:** A quality-control queue for knowledge *before* the Executive starts relying on it: playbooks, failure cases and reference sources.\n\n**What it does:**\n\n- Lists items waiting for review, with their status (pending, approved, rejected, needs revision) and priority.\n- Opens any item to show its full content and where it came from.\n- Approves, rejects, or flags an item for revision, and attaches corrections without changing the original.\n- Bulk-approves everything pending, or a whole domain at once when you trust the source.\n\n**How to use:**\n\n1. Open **Knowledge** in the sidebar, pick the **Advanced** tab, then **Review queue** (`/knowledge?view=review` opens it directly; the old `/review` link still works). The number on **Advanced**, on **Review queue** and on **Knowledge** in the sidebar shows how many items are pending or need revision.\n2. The queue has three tabs: **To review** (pending and needs-revision items), **All items** and **Annotations**.\n3. On **To review**, click **Approve** on an item, or open its **⋯** menu:\n   - **View details**: opens the item in a side panel.\n   - **Flag for revision**: marks it *needs revision* as a reminder to follow up. A flagged item stays in use until you reject it.\n   - **Priority: High / Normal / Low**: approved playbooks marked High are ranked first when the Executive pulls passages, and are labelled as a verified priority source.\n   - **Reject…**: keeps it out. A dialog lets you add an optional note on why, then click **Reject**.\n4. In **View details** you can:\n   - Read the **Document content** of a playbook or failure case, and click **Edit** to change it there (**Save** / **Cancel**). For a reference source, read its **Source content** passages (**Load 10 more chunks…** shows more).\n   - Write **Reviewer notes**. They save when you click away from the box.\n   - Fix a detail without rejecting the whole item: type in *Add a correction or clarification…* and click **Add**. Active corrections are given to the Executive alongside the item. Each correction's menu has **Turn off** / **Turn on** and **Delete**.\n   - **Approve** and the **⋯** menu sit at the bottom of the panel.\n5. Trust a source? On **To review**, click **Approve all pending**. This approves pending items only; items flagged for revision stay in the queue.\n6. On **All items**, filter by status, domain and type (**Built-in**, **Failure case studies**, **Reference library**). Under **Bulk actions**, pick a domain and click **Approve all pending in *domain***, **Curate *domain***, or **Stop curating *domain***.\n7. **Annotations** lists every active correction across items, each with **Turn off** and **Delete**.\n\n**Where items come from:**\n\n- Playbook and failure-case files you add or edit under **Built-in playbooks** on the Knowledge page.\n- Any domain you choose to **Curate**. The knowledge that ships with Open Executive is trusted by default and never waits here. Curating a domain queues its shipped items and holds them back from the Executive until you approve them; the app asks you to confirm first. **Stop curating** returns the items you haven't reviewed to trusted default. When the queue is empty, **Review a domain yourself** offers the same choice.\n- Company documents you upload go straight to **Company documents** and don't appear here.\n\n**Who can use it:** anyone signed in can review, approve, reject and curate.\n\n**Example:** you add a *discount-policy* playbook under finance. It shows up here as pending. Open it, add the correction *\"Discounts over 15% need CFO sign-off\"*, then **Approve**. The Executive now uses the playbook with your correction attached.\n\n*Tip: when you open a playbook or failure case under **Built-in playbooks**, its review status shows under the filename, with **Approve** right there and **Flag for revision** in its **⋯** menu.*\n\n*Note: approved and flagged (needs revision) items feed retrieval; pending and rejected ones are kept out. Editing an approved or rejected file marks it *needs revision*, which puts it in use until you review it again.*",
  "mermaid": null,
  "generated_at": "2026-10-15T21:00:00Z"
}