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

# Reviewing before you send

> The pause that lets you read every recipient's script before a single video is made — what to look for, and how to fix what you find.

**Where:** the batch card, once a gated batch has paused

Text is cheap and video is expensive. That is the entire argument for the
<Tooltip tip="An optional pause after the scripts are written, so you can read and edit every recipient's text before anything is rendered."><strong>narrative review gate</strong></Tooltip>: it stops a run after the scripts are
written and before anything is rendered, so a wrong claim costs you a correction rather than fifty
finished videos.

Switch it on in the campaign wizard. For a first campaign, and for any campaign to people who
matter, leave it on.

## How it goes

<Steps>
  <Step title="The batch pauses and waits">
    Every person is personalized up to their script, and then the batch parks. It waits indefinitely,
    and the card's main action becomes **Review narratives**.
  </Step>

  <Step title="Read every script">
    One card per person, with the personal parts in editable boxes: the opening, the optional context
    scene, and the closing. The core story between them is signed off and is not editable here, which
    is the point.
  </Step>

  <Step title="Edit what needs editing">
    Type straight into a box and it saves as you go. **Undo** puts back the original wording.
    **Regenerate** writes that person's script again when you would rather have a fresh take than
    hand-edit. **View target info** shows the data the text was written from, so you can check a claim
    against its source without leaving the page.
  </Step>

  <Step title="Approve and continue">
    Releasing the batch lets voicing, animation, rendering and assets run for everybody.
  </Step>
</Steps>

<Frame caption="The review page: one card per recipient, with the personal parts editable and the core story fixed between them.">
  <img src="https://mintcdn.com/latentcast/t0A2kXds21lJHQNY/images/guides/narrative-review.png?fit=max&auto=format&n=t0A2kXds21lJHQNY&q=85&s=ee0da614c64ebfc63d2aa028439ade2c" alt="The batch narrative review page" width="1440" height="900" data-path="images/guides/narrative-review.png" />
</Frame>

<Note>
  **What you approve is what gets spoken.** Your edits are voiced exactly as written. The system does
  not regenerate the script behind you after you approve.
</Note>

## Reading a batch of scripts well

Twenty-five scripts is a lot of similar text, and it is easy to skim. Four things repay attention:

* **Anything asserted about the person that you cannot find in their row.** This is the single most
  important catch, and this page is the last place to make it cheaply.
* **Sameness.** If every opening has the same shape, the batch will read as automated to anyone who
  compares notes with a colleague.
* **Tone against the relationship.** Warm familiarity to a cold contact reads worse than a neutral
  greeting.
* **The thin rows.** Someone with sparse data gets a shorter, plainer opening. Check that it still
  stands on its own rather than sounding truncated.

## Reviewing outside the product

**Export Excel** downloads this batch's scripts, for a review pass with someone who is not a
LatentCast user. Edits still have to be made in the product before you approve.

## Afterwards

The review page stays available after approval as a read-only record of what was said to whom, which
is useful when someone asks months later.

## Next

<CardGroup cols={2}>
  <Card title="The manifest and sending" href="/docs/guides/campaigns/manifest-and-sending">
    What a finished batch hands you.
  </Card>

  <Card title="Campaigns and batches" href="/docs/guides/campaigns/campaigns-and-batches">
    The run itself, and its controls.
  </Card>
</CardGroup>