> For the complete documentation index, see [llms.txt](https://help.dscout.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.dscout.com/interview-studies/manage-an-interview-study/scale-an-interview-study-with-ai-moderator.md).

# Scale an interview study with AI moderator

After you've moderated a few interview sessions yourself, you can turn them into an AI moderated study that asks the same questions in the same order and probes the same way. This lets you take a discussion guide you've already proven in real conversations and run it with many more participants, without rebuilding it from scratch.

### Requirements

**Scale with AI moderator** can be used on interview studies that are either launched or closed. However, everything below must be true for the option to appear in a study:

<table><thead><tr><th width="220">Requirement</th><th>Details</th></tr></thead><tbody><tr><td>Dscout AI access</td><td>Your account must have the Dscout AI add-on. If you don't have access, contact your Account Director.</td></tr><tr><td>At least one completed session with a transcript</td><td>You need at least one interview session that has finished, hasn't been excluded from the study, and has a transcript ready. (Transcription happens after a session ends, so a session that just wrapped may take a few minutes before it counts.)</td></tr><tr><td>Study language set to English</td><td><p>The interview study's language must be English.</p><div data-gb-custom-block data-tag="hint" data-style="info" class="hint hint-info"><p><strong>Note:</strong> You can still conduct your AI moderated study in other languages. The requirement here is that the source interview study be set to <strong>English</strong>. You can select other languages in the builder after your AI moderated study is drafted.</p></div></td></tr><tr><td>Your permissions</td><td>You need permission to manage the interview study <em>and</em> permission to create a study in its project. Creating the AI moderated study is a new study, so the same restrictions that apply everywhere else apply here.</td></tr></tbody></table>

{% hint style="info" %}
**Note:** Excluded sessions are skipped. For more information, see [Exclude a session from an interview study](/interview-studies/view-interview-session-recordings/exclude-a-session-from-an-interview-study.md).
{% endhint %}

### Scale a study

Once the above requirements are met, you can scale your interview study from any tab within it. The resulting AI moderated study will be created in the same project as your original interview study.

**To scale an interview study with AI moderator:**

{% stepper %}
{% step %}
Click **Scale with AI moderator** in the top right corner of your study. A modal is displayed.

<div align="left"><figure><img src="https://3976082747-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FIKfn7RcT4CW9XaIbqAdd%2Fuploads%2FBr7eNvSgk8FX57wx2XQ8%2FScale_with_AI_Mod%20(1).jpg?alt=media&amp;token=e1af949c-6fb1-4f46-95b5-895e6ddf709f" alt=""><figcaption></figcaption></figure></div>

{% hint style="info" %}
**Note:** Dscout AI immediately begins reading your completed sessions. You'll see **Reading your sessions…** under **Research objectives** while it works. For more information on what's drawn from your existing sessions, see [How Dscout AI scales a study](#how-dscout-ai-scales-a-study).
{% endhint %}
{% endstep %}

{% step %}
Enter a new **Study name**.
{% endstep %}

{% step %}
Choose a **Recruitment approach**. For more information on the available options, see [Recruitment options in an AI moderated study](/ai-moderated-studies/what-are-ai-moderated-studies/recruitment-options-in-an-ai-moderated-study.md).
{% endstep %}

{% step %}
(If applicable) Set a **Participant limit**. This is only required for some recruitment approaches. When required, it is set to **25** by default.
{% endstep %}

{% step %}
(If applicable) Select a **Target country**. This only applies when auto-recruiting from the Dscout Panel.

<div align="left"><figure><img src="https://3976082747-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FIKfn7RcT4CW9XaIbqAdd%2Fuploads%2FeOJsFXxYUpJpvNKbdisV%2FScale_with_AI_Mod%20(4).jpg?alt=media&amp;token=d1b5c9e1-a850-440b-b6b2-f988a0f6be9d" alt=""><figcaption></figcaption></figure></div>
{% endstep %}

{% step %}
Review the **Research objectives**. Dscout fills these in based on your interview sessions. You can edit these in this modal or in the builder once you create your study. For more information, see [Define objectives and context in an AI moderated study](/ai-moderated-studies/create-an-ai-moderated-study/define-objectives-and-context-in-an-ai-moderated-study.md).

<div align="left"><figure><img src="https://3976082747-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FIKfn7RcT4CW9XaIbqAdd%2Fuploads%2F9vvZwUZr70Q8oEcDJaPC%2FScale_with_AI_Mod.jpg?alt=media&amp;token=8b12dbd6-2fc1-4a88-b119-c897ba008628" alt=""><figcaption></figcaption></figure></div>
{% endstep %}

{% step %}
Click **Draft study**.
{% endstep %}
{% endstepper %}

Now, your AI moderated study is created. This usually takes around a minute, but you can close the modal without interrupting the process. When your new study is ready, you'll find it in the same project as your original interview study.

{% hint style="info" %}
**Note:** If a draft fails, you'll see **We couldn't draft the study** with a reason, and a **Try again** button. Nothing is created when a draft fails.
{% endhint %}

Review the draft in the study builder and finish setting it up by adding incentives, consent forms, etc. **Things to keep in mind when reviewing your scaled study:**

* **Task questions will need links.** The AI moderator can't share its screen the way you did in a live session, so participants only see what you attach to the question. You add these in the study builder. For more information, see [Task questions](/study-basics/question-types/activity-questions/task-questions.md).
* **Some questions may be left out.** If your interview covered more ground than an AI moderated study allows, the extra questions at the end are dropped. You'll see how many in the confirmation modal after your study is created.

For more information on finalizing your AI moderated study, see our [AI moderated study pre-launch checklist](/ai-moderated-studies/create-an-ai-moderated-study/ai-moderated-study-pre-launch-checklist.md).

{% hint style="warning" %}
**Review the draft before launching.** Dscout AI writes the questions from what it read in your transcripts. Read them the way a participant would, check that nothing from a specific session was carried over as if it applied to everyone, and confirm the follow-up behavior matches how you want the conversation to go.
{% endhint %}

### How Dscout AI scales a study

Dscout AI scales your interview study using the following:

* **The transcripts of your completed sessions**, oldest first. If you corrected a transcript, your corrected version is what gets read.
* **Your annotations on those sessions.** These are the notes you took, the markers you created, the tags you applied (including any note you left on a tag), and the clips you cut and named. Annotations are the clearest signal of what mattered to you, which is why Dscout AI refers to them.
* **Your study setup copy.** This includes the participant Overview you wrote and the invite message you sent. These usually describe the product and topic in plain words.
* **The research objectives you confirmed** in the modal before clicking **Draft study**.
* **Whether a participant was interviewed more than once.** Follow-up sessions in your interview study are treated as such so Dscout AI knows what topics you chose to revisit.

{% hint style="info" %}
**Note:** AI-generated tags, such as notable moments, are deliberately left out. Those are a model's reading of the session, not yours.
{% endhint %}

Dscout AI does not carry over any of the following from your original study:

<table><thead><tr><th width="240">Not carried over</th><th>Why</th></tr></thead><tbody><tr><td>Links and stimuli on Task questions</td><td>Nothing in an interview recording records which stimulus was on screen and when. The AI moderator also can't share its screen, so participants only see what you attach. Add these to questions yourself in the study builder.</td></tr><tr><td>Questions beyond the AI moderated study limits</td><td>An AI moderated study allows up to 5 Task questions and 30 questions total (though neither Scale nor Multiple choice count toward this limit). If your interviews ran past that, the extra questions at the end of the guide are dropped, and you're told how many in the confirmation modal.</td></tr><tr><td>Any recruitment data</td><td>The new study is a fresh study. You set up recruiting in the builder, starting from the recruitment approach and participant limit you chose in the modal.</td></tr><tr><td>Everything else in the builder</td><td>Incentive, appearance, informed consent, and the rest start from the builder's defaults. Set them before you launch.</td></tr></tbody></table>

### FAQ

<details>

<summary>How many of my interview sessions does Dscout AI read?</summary>

There's a cap on how much transcript one draft can read: roughly eleven hour-long interviews. If you've run more than that, Dscout keeps your most recent sessions in full and leaves the oldest ones out. It will tell you exactly how many were skipped in the confirmation modal.

Whole sessions are kept or dropped, rather than trimming a bit off each one, so the end of your guide never disappears from every session at once.

</details>

<details>

<summary>Which of my questions make it into the new study?</summary>

Your questions, in the order you asked them. Every session contributes: The questions are the union of what you asked across all of them.

Where two sessions word or order the same question differently, your most recent session is prioritized.

</details>

<details>

<summary>What question types can the draft include?</summary>

Drafts can include **Talk out loud**, **Task**, **Multiple choice**, and **Scale** questions. For more information, see [Question types in AI moderated studies](/ai-moderated-studies/what-are-ai-moderated-studies/question-types-in-ai-moderated-studies.md).

</details>

<details>

<summary>How does the AI moderator know when to ask follow-ups?</summary>

How much the AI moderator probes on a given question is based on how often you actually probed on it in your own sessions, not on a guess about how interesting the topic is.

</details>

<details>

<summary>What moderator context does Dscout AI write?</summary>

A short briefing (up to three bullets) covering what is being tested or discussed, how to hold the conversation, and what to steer clear of, drawn from your transcripts. This is what carries the unwritten knowledge a human moderator has into the AI moderator's instructions. It's never shown to participants.

For more information, see [Define objectives and context in an AI moderated study](/ai-moderated-studies/create-an-ai-moderated-study/define-objectives-and-context-in-an-ai-moderated-study.md).

</details>

<details>

<summary>Are my research objectives carried into the new study?</summary>

Yes, exactly as you confirmed them in the modal before clicking **Draft study**. You can keep editing them in the study builder.

</details>

<details>

<summary>What will my new study be named?</summary>

The study takes the name you typed in the modal. The internal name records which interview study it came from and how many sessions it was built from, so two drafts from the same study are easy to tell apart.

</details>

<details>

<summary>What recruiting settings does the draft start with?</summary>

The recruitment approach, participant limit, and target country you chose in the modal. Everything else about recruiting is set up in the study builder.

</details>


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://help.dscout.com/interview-studies/manage-an-interview-study/scale-an-interview-study-with-ai-moderator.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.