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

# Mark Aide's draft as good or wrong

> When Aide drafts a reply for your team, you can tell it whether the draft was right.

When Aide drafts a reply for your team, you can tell it whether the draft was right. Your verdict replaces Aide's own review of that draft, and a draft you mark as wrong becomes a test Aide replays every night. The buttons are on Aide's draft note in a conversation, in **Inbox**.

## Where the buttons appear

Aide's drafts are internal notes written by Aide in the conversation. Every night Aide reviews a sample of its drafts against the replies your team sent (see [Quality: see whether Aide's replies meet the bar](https://docs.aide.app/aide-ai-agent/quality-page.md)). Once a draft has been reviewed, its note shows a line such as "Aide's reviewer rated this draft acceptable." with two buttons, **Good** and **Wrong**.

A draft that hasn't been reviewed yet has no buttons. To flag one of those, an owner or admin can use **Add to evals** in the message's menu instead (see [Evals: the tests Aide replays every night](https://docs.aide.app/aide-ai-agent/evals.md)).

## Mark a draft

1. Open the conversation in **Inbox** and find Aide's draft note.
2. Select **Good** if the draft was right. It is saved straight away.
3. Select **Wrong** if it wasn't. A box opens: "This was wrong because…". Write what was wrong in up to 500 characters, then select **Mark as wrong**. The reason is optional, but it is what Aide's test is written from, so a clear one helps.

The note then reads "Marked as good" or "Marked as wrong", with your name and your reason. Select **Change** to give a different verdict.

## What your verdict does

* **It is final.** A teammate's verdict overrides the judge's for that draft, in the headline and the numbers on the Quality page. The judge's own scores are kept beside it, so Quality can show how often the judge and your team agree.
* **Wrong becomes an eval case.** Aide drafts a test from the conversation and your reason, and replays it every night. It is listed in AI Agent > Quality > Evals with the origin **Teammate review**.
* **Good retires that case.** If you change a Wrong verdict to Good, the eval case made from it leaves the nightly run.

## See why Aide scored it that way

Under the buttons, **Scores by criterion** opens the judge's score on each criterion with its reason, and the messages it relied on. **Open review** opens the full review screen in Quality, where owners and admins can change the score on a single criterion and say why.

## Who can do this

Any teammate can mark Aide's drafts as good or wrong. The Quality page, the review screen and **Add to evals** are for owners and admins.

## Only in Aide's inbox

The **Good** and **Wrong** buttons are in Aide's own inbox. They are not in the Aide panel inside Zendesk or Front, or on the Aide card in Gorgias.


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