> ## Documentation Index > Fetch the complete documentation index at: https://help.voxdash.com/llms.txt > Use this file to discover all available pages before exploring further. # AI Assistant in the Editor > Ask the AI Assistant to review, shorten, retype, or check the pronunciation of your questions. Choose what it works on, review each suggestion, and accept or reject changes. The **AI Assistant** works inside the Questionnaire Editor. You can ask it about a question, a page, chosen sections, or the whole questionnaire, and it replies with **suggestions** that you review before anything changes. > [!NOTE] > > AI requests use your organization's **AI credits**. The estimated cost is shown before you send a request. See [AI Credits](https://help.voxdash.com/AI-&-Analysis/AI-credits.md). ## Open the AI Assistant You can open it in two ways: * Select the **AI** icon (sparkles) at the bottom of the property panel rail on the right of the editor. * Select **Ask AI about this question** on a question in the canvas. ![AI Assistant panel in the editor](https://mintcdn.com/voxdash/B7FWt7C1UKiKcVrB/images/QB/d-ai-panel.png?fit=max\&auto=format\&n=B7FWt7C1UKiKcVrB\&q=85\&s=b64120a209dd4f9b60b76d0808b9e6e1) ## Choose What the AI Works On The line under the chat tabs shows the current **scope**, for example *This question · Q1*. Select **Change** to pick another scope: | Scope | What the AI looks at | | - | - | | **This question** | The question selected on the canvas. | | **This page** | All questions on the current page. | | **Choose sections…** | One or more sections you pick. | | **Whole questionnaire** | Every question in the questionnaire. | When you have applied filters in the editor, you can also work on the **filtered questions** only. ![Scope menu of the AI Assistant](https://mintcdn.com/voxdash/B7FWt7C1UKiKcVrB/images/QB/d-ai-scope.png?fit=max\&auto=format\&n=B7FWt7C1UKiKcVrB\&q=85\&s=3f993812170105b35fe36c10e1818aea) > [!TIP] > > In a new chat, selecting something on the canvas switches the scope to it automatically. Once a request is running, the scope of that chat is fixed. ## Ask the AI Type your request in the box at the bottom — for example *"Make this question easier to read aloud"* — or pick a **quick action**: | Quick action | What it does | | - | - | | **Suggest question type** | Recommends a better-suited question type for the question. | | **Check how it sounds** | Checks how the text will sound when read aloud by text-to-speech and flags words that may be mispronounced. | | **Shorten questions** | Proposes shorter wording to reduce interview length. | | **Review the questionnaire** | Reviews the questions in scope and flags issues and improvements. | Other actions may be available depending on your plan, such as suggesting SPSS labels or explaining logic in plain language. Before you send, the cost button next to **Send** shows the estimated cost of the request. ## Review Suggestions The AI answers with **suggestion cards**. Each card describes one proposed change or finding, for example *rewrite*, *change type*, *merge*, *remove*, *add*, *flag*, or a *tip*, and is marked **Critical**, **Improvement**, or **Info**. For each suggestion you can: * **Accept** — apply the change to the questionnaire. An **Undo** notice appears right after. * **Reject** — discard the suggestion. * **Dismiss** — hide a finding without changing anything. * **Don't flag again** — stop the AI raising the same finding on this question. * **Show on canvas** — highlight the affected question. * **Preview** — see how the canvas would look with the change, and open the preview in a tab. When a request returns many suggestions, use review mode: move with **Next** and **Previous**, or use **Accept all** / **Reject all**. Keyboard shortcuts are listed in the panel. > [!NOTE] > > If the questionnaire changed after the AI made a suggestion, the card may be marked **outdated**, or say that the question is missing or the list changed. You can still choose **Apply anyway** where that is safe. Suggestions that save interview time show the estimated seconds saved. ### Pronunciation Check **Check how it sounds** returns a card that lists words the voice may mispronounce. For each word you can: * **Say as…** — enter how the word should be spoken, * **Say in another language** — have the word pronounced in another language, * **Save** — store the correction as a custom pronunciation for text-to-speech. Saved corrections appear in the [Text-to-Speech](https://help.voxdash.com/Questionnaire-Builder/tts.md) panel. ## Runs, Progress, and Cost Large requests run in steps. While a request runs, the chat shows its progress, and you can **Stop** it. When a request ends early or only partly succeeds, the chat says so and offers **Try again** or **Redo**. You are **charged only for what was processed**. The final cost appears in the chat when the run finishes. ## Chats and History * Each request lives in a **chat tab**. Select **+** to start a new chat, or **×** to close one. There is a maximum number of open chats; close one to start another. * Select the **history** icon to see earlier chats on this questionnaire. Filter by scope (**All**, **Question**, **Section**, **Selection**, **Questionnaire**), by action, or show only chats that **Need review**. ![AI Assistant chat history](https://mintcdn.com/voxdash/B7FWt7C1UKiKcVrB/images/QB/d-ai-history.png?fit=max\&auto=format\&n=B7FWt7C1UKiKcVrB\&q=85\&s=a8de77ffe2bb6162792a105074a30760) ## When the AI Is Unavailable | Message | What it means | | - | - | | Not included in your plan | Your subscription doesn't include AI actions. Select **Upgrade** to see plans. | | Plan limit reached | Your plan's limit for this questionnaire has been reached. Select **Upgrade**. | | Not enough AI credits | The organization's credit balance can't cover the estimated cost. Top up AI credits. | | Locked or read-only version | AI changes can't be accepted on a locked, published, or previewed version. Create or switch to an editable version. | | Action unavailable | The action is temporarily unavailable or not supported for the selected scope. | ## Readable Logic for Exports The **Excel data map** export can include plain-language descriptions of your logic. When you configure that export, a panel shows how many questions already have a readable description and lets you have the AI write the missing ones. See [Export](https://help.voxdash.com/Questionnaire-Builder/Export.md). This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.