How do I use Ask AI?

Ask AI is Marvin’s powerful search function, supercharged by AI. You can use it to ask a range of questions about your research data – across projects, notes, users and lots more.

For example, "What do customers say about our onboarding?" or "How many customers mentioned our new launch in surveys between October and January?".

This article contains tips on how to get the best results with Ask AI.

You can use Ask AI effectively at every stage of your research project.

Before you begin your project, you can ask questions about your existing research to form a hypothesis.

You can also identify participants who you’ve interviewed before and meet the criteria for your new project. Use the information to write project overviews and discussion guides.

Type in questions in the Ask AI bar on your home page.

Or you can also go to the Discover Research tab.

When you’re gathering data through interviews, surveys, usability tests, etc., use Ask AI on individual files. It will understand the context of the recordings, images and documents to summarize and match insights to your discussion guide and labels.

Under the tab "Q/As", you’ll see key points from the call tagged to specific questions.

Click "My topics" to see insights grouped by topic, theme or use-case. The default topics available for each call are listed under "Pre-made topics". "Feedback for Researcher" will contain helpful guidance and suggested questions for the person moderating the call.

"Painpoints & Concerns" and "Key Takeaways" will group relevant information from the call under these topics.

Once you’re ready to synthesize your research, you can uncover trends, find themes in the data, prove or disprove your hypothesis, locate quotes, etc. You can use this information to create reports that can be shared with the rest of your team. You can do this from the Analyze tab.

You can also use Ask AI while analyzing notes across a bunch of projects in "Cross project analysis".

To get relevant answers more easily, make sure you apply the right filters. Marvin will then be able to provide more targeted results. You can filter by project, file tags, research areas, time and even the file type.

Whenever you Ask AI a question that mentions any file tags in your repository, it will automatically apply them and prioritize the linked data.

For example, if you ask, "How was customer sentiment about XYZ feature in March 2025?", Marvin will check for tags of XYZ feature as well as March 2025 and apply these filters in the results.

See how Marvin applied the "Plan: Enterprise" file tag when we asked "What do enterprise users say about Marvin?".

In case you do not want to apply these filters, use the option in the top right corner to "Run query again". This time, Marvin will generate an answer balanced for relevance and recency, and without any file tag filters.

If you still need to regenerate the answer, you can try asking for results based on "All time data".

You can turn off automatic file tag detection completely. However, we recommend that you keep it on to get more targeted answers.

File tags are a crucial step of identifying and categorizing data and insights in your repository. The more robust this tagging, the more detailed results you can get from Ask AI and Deep Research.

You can ask follow-up questions on an Ask AI result to refine the answer. You can use a range of prompts. Here are some suggestions:

This is not an exhaustive list. We encourage you to explore Ask AI till you have the insights you need.

If you apply a filter after the answer is generated, Marvin will create a new branch of the answer. You can switch between these branches by using the arrows in the top right corner of the answer.

Marvin will also retain a history of your Ask AI results. You can access these quickly from the advanced search page. Find searches categorized by time period: last 7 days, 7-30 days, and older than 30 days.

You can navigate to the advanced search page from the search bar.

Want to learn about the kind of prompts you can use on Ask AI? Check out our help center article.