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

# Human Reviews

> Structure expert review and turn feedback into reusable memory

Human-guided reviews are a requirement for judging and improving accuracy in AI-powered systems.

Reviews is the human layer on top of traces. It captures expert judgment as structured, reusable data, instead of feedback scattered across Slack threads and spreadsheets that is hard to manage and reuse.

The value shows up later: every review becomes ground truth you can reuse, so the same expert judgment trains your judges and shortens future reviews.

## How a review flows

A review moves through two stages, then pays off:

* **[Rubric Studio](/reviews/rubric-studio):** define the grading dimensions, instructions, and good and bad examples that reviewers and judges score against.
* **[The operational loop](/reviews/operational-loop):** route traces to reviewers, grade them, and see what came back.

The payoff comes after: every correction feeds your judges and datasets automatically, so expert judgment gets reused instead of repeated. Start by setting up a rubric, then run traces through the loop.

Throughout, the [Review Copilot](/reviews/copilot) helps reviewers grade faster.

## Next steps

<CardGroup cols={2}>
  <Card title="Rubric Studio" icon="clipboard-check" href="/reviews/rubric-studio">
    Define clear, consistent standards for reviewers and judges
  </Card>

  <Card title="The operational loop" icon="arrows-rotate" href="/reviews/operational-loop">
    Route, grade, and reuse expert feedback
  </Card>
</CardGroup>