Create a study with AI in 60 seconds

See how Versive works as an AI survey generator, turning a plain-language prompt into a full study you can refine and share.

Building a study question by question takes time. Describe what you want to learn in plain language, and the Versive Study Assistant creates questions, settings, and logic in under a minute. You can then ask it to revise any part of the study.

For example, you can turn "I need to find out why trial users don't convert" into a study ready for review.

Describe your research goal, get a complete study

Start a new study and tell the assistant what you're trying to learn. You don't need exact survey wording. A prompt like "I want to understand why free trial users aren't upgrading to a paid plan" is enough to create a study with a mix of question types, a sensible order, and an initial set of settings.

The assistant also works from documents and images you already have. Paste in a research plan or discussion guide and it turns that structure into an interview. Import a Figma prototype and it suggests tasks and questions based on the screens participants will see. You can also provide a reference image as source material.

What the assistant creates

The assistant draws from Versive's full question library, including AI conversational questions with adaptive follow-ups, multiple choice, rating scales, NPS, matrix questions, and ranking. It chooses a type based on what you're asking. A question about willingness to pay will typically use a quantitative format. A question about why someone canceled will typically use an AI question that can probe the answer.

It configures study settings at the same time, including the interview mode (text, voice, or video) and branding. If your prompt implies structure, such as a screener up front or a branch for a specific kind of answer, the assistant can add that as skip logic rather than leaving it for you to wire up afterward.

A worked example

Suppose you type: "I want to understand why free trial users aren't upgrading to a paid plan." The generated study might include a screener confirming that the participant is on a free trial, an AI question about what they hoped to get from the product, a rating scale on whether the product met their expectations, an AI question about what stopped them from upgrading, and a multiple choice question about what could change their mind.

You can revise any part of the study. If it leans too heavily on open-ended questions and you want more structured data to chart later, ask the assistant to swap an AI question for a ranking or multiple choice question. If it missed a screener condition, describe the condition and the assistant adds the corresponding logic rule.

Refine the study by chatting

Review the first version before sharing it. The study editor puts your question list, the question editor, and the assistant into three resizable panels, so you can see each change as it is applied.

Use requests like these to revise the study:

• "Add a screener that filters out anyone who hasn't used the product in the last month."
• "Reorder so the pricing questions come before the open-ended ones."
• "Make the tone more casual."
• "Skip the feature-request question if they rated us below 6."

For a multi-part request like "create a screener section, then add a section on purchase decisions with appropriate skip logic," the assistant breaks the work into steps, shows you what it's doing at each one, and applies the changes in sequence. Every change comes with an undo, so you can try something and back out of it without rebuilding the study from scratch.

Review before sharing

Review these settings before you share the link:

• Interview mode. Text and voice is the recommended default because participants can choose per question. Use text only when you explicitly do not want recordings; use voice-only, preview, realtime voice, or video modes when the research calls for them. See Text, voice, or video: choosing an interview mode.
• Engine. Regular AI, Realtime AI (beta) for a live spoken conversation, or Survey mode for a structured survey with no AI moderation at all. This lives in Settings → General.
• Question types and logic. Check any question types or rules the assistant added without an explicit request. See Survey question types, and when to use each and Skip logic and branching, explained.

Studies are live as soon as you create them. Once the study looks right, you can share the link, embed it in your product, or recruit participants. There is no separate publish step.

When to build manually instead

The assistant is useful for producing a complete starting point. If you already know exactly which questions you want and in what order, the drag-and-drop builder and question library may be quicker for small, precise edits. For a common study type, you can also start from a template and adjust it directly.

You can also create the study with the assistant and make final adjustments by hand. To configure probing on open-ended answers, see AI follow-up questions: probing without a moderator.