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

# Build and Edit Your Study in the Studio

> Use the studio chat to build and edit an AI Moderated Interview or AI Moderated Prototype Test: every step, setting, and follow-up you can change, and what the chat can't do.

**The studio chat builds and edits your AI-moderated study for you: describe what you want in plain language, and it changes the study directly.**

## What Is the Studio Chat?

When you open the studio from an [AI Moderated Interview](/support/support/studies/ai-moderated-interviews) or [AI Moderated Prototype Test](/support/support/studies/ai-moderated-prototype-testing), the **Setup** tab shows your study as a canvas: the steps participants go through, in order. The chat sits beside that canvas. Instead of filling in forms, you tell the chat what you want, for example "add a rating question after the second task" or "don't ask more than two follow-ups on the pricing question", and it makes the change.

Interviews and prototype tests are built the same way. A prototype test is a study that includes a **Prototype test** step, with tasks and questions inside it. Everything in this article applies to both unless it says otherwise.

## Why It Matters

**Build a full study from a sentence, or from the guide you already have.** Describe your research goal, or upload an existing research guide, and the chat drafts the plan, the questions, and the flow.

**Change anything without hunting for the setting.** Question wording, answer options, follow-up limits, prototype links, and moderator instructions are all one request away.

**See every change as it happens.** Edits land on the canvas right away, and the chat tells you exactly what it changed.

## How the Chat Works

1. From your study's **AI Moderated Interview** step in Great Question, click **Open studio**. You land on the **Setup** tab with the chat open beside the canvas.
2. Type what you want, or upload a research guide (see [Start From a Research Guide](#start-from-a-research-guide) below). You can describe the whole study at once or make one change at a time.
3. While it works, the chat shows **Thinking…**, then a collapsible **Thought** step with its reasoning. Open it to see how it interpreted your request.
4. The chat applies the change and confirms it in plain language, for example: "Added right after the mic setup, before the prototype: 'How would you describe your experience with \[topic]?' Rest of the flow is unchanged." After larger changes it also draws the study as a flowchart in the chat so you can check the order.
5. Keep going in the same conversation until the study is right. There is no separate save step.

### Start From a Research Guide

If you already have a discussion guide, test script, or research brief, you don't need to retype it. Upload it to the chat and it builds the study from the document, carrying over your goals, questions, and tasks.

1. In the chat box, click **+** and choose **Upload document**. You can also drag the file into the chat box.
2. Add a short message saying what you want, for example "Turn this guide into an AI-moderated interview" or "Build a prototype test from this script, using [https://example.com/prototype](https://example.com/prototype) as the prototype".
3. Send the message. The chat drafts the study on the canvas and tells you what it built.
4. Review the result and refine it in the same conversation, the same way you would any other change.

The chat accepts **PDF**, plain text (**.txt**), Markdown (**.md**), and images (**PNG**, **JPG**, **GIF**, **WebP**), up to **25 MB** per file. Word documents and Google Docs aren't supported directly, so export them as a PDF first. Images work too, for example a screenshot of a guide that lives in another tool. The chat reads the image to build the study; it doesn't add the image to what participants see.

<Tip>
  If your guide includes things the studio doesn't support, such as branching or time limits (see [What the Chat Can't Do](#what-the-chat-cant-do)), the chat can't build them as written. Check its confirmation and the canvas to see how it handled those parts.
</Tip>

### Working With the Chat

A few things make the chat faster to work with:

* **Edit this step in chat.** Select a step on the canvas and click **Edit this step in chat**. The chat knows which step you mean, so you can just say "make this shorter" or "add a follow-up about cost".
* **Clickable questions.** When the chat needs more detail, it may ask up to four short questions as a card you can click through instead of typing answers.
* **Several chats per study.** The chat picker in the top bar lists the study's five most recent chats, named from their first message, and includes **New chat**. Coming back to a study reopens its latest chat.
* **Closing the chat.** Close it from its header, or drag its resize handle all the way left, to give the canvas the full width. Reopening picks up the same conversation.
* **Resending a message.** If you edit and resend an earlier message, the chat discards the replies after it and answers again. Changes it already made to the study stay in place.

<Note>
  If you close the browser tab while the chat is working, it finishes the change anyway. Reopen the studio to see the result.
</Note>

## What You Can Edit

### Study details

| What | What you can change |
| - | - |
| **Study title** | The name of the study. |
| **Welcome screen** | **Title participants see** and **Description participants see**, the first screen participants land on. |
| **Agenda** | The list of items shown on the welcome screen so participants know what to expect. Each item has a title and an optional description. |
| **Research plan** | Your **goals**, **research questions**, and **success criteria**. The AI moderator uses your goals as the purpose of the whole study, so a clear goal leads to better follow-up questions. |

### Steps

You can ask the chat to add, remove, reorder, or rewrite any step. Every step has a title and description. Here is everything else you can set on each type.

| Step | What you can set |
| - | - |
| **Open-ended** | The question. Participants answer in their own words. |
| **Rating** | The question, and labels for the low and high ends of the scale (for example "Very difficult" and "Very easy"). The scale is always 1 to 5. |
| **Single choice** | The question and its answer options. Participants pick one. |
| **Multiple choice** | The question and its answer options. Participants pick any that apply. |
| **Yes / No** | The question. |
| **Prototype test** | The link to your prototype or website. Every prototype test holds at least one task or question. Prototype tests only. |
| **Task** | What the participant should do (the instructions), plus **Success criteria** and **Failure criteria** that describe what finishing, or failing, the task looks like. Tasks live inside a prototype test. |
| **Device permissions** | Whether **Microphone**, **Camera**, and **Screen share** are each required, optional, or hidden. |
| **Contact details** | Only the title and description. This step collects the participant's name and email. |
| **End** | The closing screen's title and description. This is where your goodbye message goes. |

Answer options need at least two choices, with no blanks or duplicates.

For a task, you can also ask for a step that runs just before the task begins or just after it's completed, for example a quick rating question as soon as the participant finishes.

### Moderator settings and follow-up questions

Every question, prototype test, and task can have its own moderator settings. These control how the AI moderator behaves on that step:

| Setting | What it does |
| - | - |
| **Goal** | What the moderator is trying to learn on this step. The moderator uses it to decide whether an answer is complete. |
| **Max probes** | The most follow-up questions the moderator can ask on this step. Set it to **0** for no follow-ups. The canvas shows it as "up to 2 probes". |
| **Instructions** | Directions for the moderator on this step, in order, for example "keep follow-ups short" or "ask about cost if they mention switching tools". This is also how you change the moderator's tone. |

Settings on a **Prototype test** step apply to every task and question inside it, so one set of moderator settings on the prototype test is usually all a prototype study needs.

**How follow-ups work.** The moderator treats **Max probes** as a ceiling, not a target. It only follows up when an answer needs more depth, and it won't probe an answer that already explains the why. As it gets close to the limit, it starts wrapping up the step. Asking a participant to clarify something, or acknowledging what they said, doesn't count as a probe. The moderator never reads its goal or instructions aloud to participants.

## What the Chat Can't Do

Some things aren't part of a studio study today. If you ask for one, the chat will tell you it can't.

| Not available | What to do instead |
| - | - |
| **Branching or skip logic** | Every participant goes through the same steps in the same order. |
| **Language, session length, or time limits** | These aren't settings. You can ask the moderator to keep a step brief through its **Instructions**. |
| **Moderator name, voice, or persona** | Steer tone through each step's **Instructions**. |
| **A rating scale other than 1 to 5** | Use a **Single choice** question with your own options. |
| **Recruiting, invitations, participant emails, or screeners** | Click **Back to study in Great Question** to handle these there, the same way you would for any study. |
| **Undo or version history** | The latest change always wins. To reverse something, ask the chat to change it back. |

The chat also checks every change against a few rules before saving it. Every study needs a **Device permissions** step and ends with an **End** step, and tasks must sit inside a prototype test. If a change would break one of these rules, the chat adjusts it or explains why it can't.

If you ask for a question that could lead participants toward an answer, the chat suggests a more neutral version once. If you'd still like your original wording, confirm it and the chat uses it as written.

## Editing a Live Study

You can keep editing after your study is launched, and changes apply to participants who start after that. Be careful with deletions.

<Warning>
  Deleting a step permanently deletes the step **and every response collected for it**. This can't be undone. If a study already has responses, rewording a step is usually safer than removing it.
</Warning>

## Best Practices

**Start with the goal, not the questions.** Tell the chat what decision the research will inform. It writes better questions, and the moderator asks better follow-ups, when the goal is specific.

**Upload your guide instead of summarizing it.** If you already have a discussion guide or test script, the full document gives the chat more to work with than a summary.

**Make one change per message when it matters.** Large requests work, but small ones make it easier to check the confirmation and spot anything the chat interpreted differently than you meant.

**Use Edit this step in chat for precise edits.** Pointing the chat at a step avoids ambiguity when several questions look alike.

**Set probes on purpose.** Use **0** for simple factual questions, and 2 or 3 where you want the "why" behind an answer.

**Preview after every round of edits.** **Preview** runs the study exactly as a participant sees it, including the live AI moderation, without recording anything or saving results.

## Troubleshooting

| Issue | What's Happening | How to fix |
| - | - | - |
| "Unsupported file type" when uploading a guide | The chat accepts PDF, text, Markdown, and image files only. | Export Word documents or Google Docs as a PDF, then upload the PDF. |
| "File is too large" when uploading a guide | Files can be up to 25 MB. | Remove embedded images from the document or split it, then upload again. |
| "Couldn't attach that file" | Your browser or a security tool blocked the upload. | Remove the attachment to keep chatting, then try again or paste the guide's text into the chat. |
| The chat won't save a prototype link | Links must start with **https\://**, use a normal web address rather than an IP address, and can't include a username or password. | Use the prototype's public share link. |
| A Figma Make prototype doesn't show on the canvas | Figma Make files can't be embedded. | Publish the app in Figma and use its **.figma.site** link. |
| The canvas says "This prototype can't be shown here" | The site doesn't allow itself to be embedded in another page. | No action needed. Participants open the prototype in a new tab, with the study panel kept on top. |
| The chat says it can't send invitations or set up recruiting | Recruiting happens in Great Question, not the studio. | Click **Back to study in Great Question** and use your study's usual recruitment settings. |
| A change the chat confirmed isn't showing | The canvas may not have refreshed yet. | Reload the page. If it's still missing, ask the chat to show the current study flow. |
| Changes in the studio aren't reflected in Great Question | The study didn't sync back, most often because of a temporary connection issue. | Reopen **Open studio** and make the change again. If it keeps failing, contact your Customer Success Manager. |

***

**Still need help?** Contact us at [support@greatquestion.co](mailto:support@greatquestion.co) — median response time is 19 minutes during [support hours](https://greatquestion.co/features/about-our-support-team).