> ## Documentation Index
> Fetch the complete documentation index at: https://docs.botbrains.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Review Queue

> Work through conversations whose scorecard needs a person's answers, one conversation at a time

export const feature_0 = "the Review Queue"

export const note_0 = undefined

<Warning>
  [ServiceLens](/concepts/servicelens), which includes {feature_0}, is currently in beta and available per project. Contact [support@botbrains.io](mailto:support@botbrains.io) to turn it on for your project. {note_0}
</Warning>

Some [scorecard](/concepts/scorecards) items need a person's judgment, so you fill them by hand instead of letting the AI answer them. The Review Queue collects the conversations that wait for those answers and lets your team work through them one after another with the keyboard. A conversation has no score until someone answers its open items. After that, its score counts in score filters, monitors, and alerts like any other score.

## How Queues Work

**A queue is a monitor.** Every active [monitor](/concepts/monitors) whose scorecard has at least one [manual item](/concepts/scorecards#unanswered-items) becomes a review queue. botBrains never creates queues on its own, so create a monitor with that scorecard to start one. When a scorecard has manual items, the scorecard editor links to a new monitor with the scorecard already selected.

**Conversations enter when the monitor evaluates them.** If the scorecard still has open items after the evaluation, the conversation joins the queue. The monitor's filters and sample rate decide which conversations it evaluates, so use them to control how much work reaches your team. A critical item the AI fills also keeps a conversation in the queue when the AI couldn't determine a value.

**Only reviewed conversations get a score.** While a conversation waits, it has no score, see [Unanswered Items](/concepts/scorecards#unanswered-items). Conversations the monitor doesn't evaluate, because its filters, sample rate, or look back leave them out, keep no score unless someone answers them on the conversation page.

**Conversations leave once every open item has an answer.** Answers count wherever you give them, inside the queue or on the conversation page. Choosing N/A clears the value, which counts as an answer. A [backfill](/concepts/conversation-attributes#backfill-existing-conversations) that overwrites values set by people keeps the answers from a completed review.

## Reviewing Conversations

Open the [Reviews page](https://platform.botbrains.io/~/reviews) to see every queue with its pending conversations, and click **Start** on one. You reach it with the **Reviews** button on the [monitors page](https://platform.botbrains.io/~/monitors). To go straight to one queue, click **Review** on the monitor's card. The queue serves conversations oldest first, by their last message.

<img src="https://mintlify.s3.us-west-1.amazonaws.com/botbrains/images/review-queue/review-page.png" alt="Reviews page listing review queues with their scorecard, pending count, conversations done in 7 days, oldest pending age, and a Start button" data-generation-prompt="Navigate to platform.botbrains.io/~/reviews in a project with ServiceLens turned on, at least two review queues (active monitors whose scorecard has an item backed by a conversation attribute with fill mode Manual) and several pending conversations each. Set viewport to 1920x1080 and collapse the sidebar. Capture the page header and the full queue table with the columns Queue, Scorecard, Pending, Done in 7 days, Oldest pending, and the Start button. Show at least one queue with pending conversations; a second row showing All done is a plus." />

**Start** opens the oldest pending conversation in review mode. Answer the open items in the **Needs your input** block of the details panel. The values the AI filled stay collapsed below it. The count of conversations done today restarts at midnight Coordinated Universal Time (UTC).

<img src="https://mintlify.s3.us-west-1.amazonaws.com/botbrains/images/review-queue/review-mode.png" alt="Conversation in review mode with the review strip at the top and the Needs your input block listing open scorecard items in the details panel" data-generation-prompt="In a project with ServiceLens turned on, open platform.botbrains.io/~/reviews and click Start on a queue with pending conversations. Set viewport to 1920x1080 and collapse the sidebar. Capture the conversation page in review mode: the review strip at the top (queue name, 'N left · M done today', Skip, Save & next, Exit) and the details panel with the Needs your input block showing at least two open items, one of them marked critical, and the 'x of y answered' counter. Answer one item first so the block shows partial progress." />

Skip and Exit save the answers you already gave. When you reach the end, the strip offers to review the conversations you skipped in this browser tab.

**Several reviewers can work one queue.** Each reviewer gets a different conversation. If you open a conversation that a teammate is reviewing, the strip says "Someone else is reviewing this conversation."

**Answer outside the queue.** The **Needs your input** block also appears when you open a conversation with open items from the [conversations page](https://platform.botbrains.io/~/conversations). Each answer saves right away and counts toward the queue. On a [monitor's page](/concepts/monitors#reviewing-results), the **Awaiting review** tab lists the queue's pending conversations, and **Start reviewing** opens the oldest one in review mode.

**Permissions.** Organization Owners and Admins, Project Owners, Project Contributors, and Project Members can review. Project Viewers can see the Reviews page but not answer. See [Permissions](/concepts/servicelens#permissions).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.