Using Dovetail Magic

Leverage Dovetail's AI magic features to speed up analysis in your notes, cluster and uncover themes from highlights, and summarize notes, insights, and search results. Dovetail’s use of artificial intelligence (AI) in its service:

• Is implemented through use of generic AI models that Dovetail imports and operates within its own infrastructure (hosted in AWS);

• Does not involve any use of customer data to train models (and customer data is not transferred to the AI vendors);

• Produces results that are clearly indicated as being derived through AI; and

• Is delivered as recommendations that can be accepted or rejected by the end users.

AI analysis is enabled by default in projects, but Dovetail lets you control the level you want through our magic automation options. These are configured by managers and contributors per project, where you can choose from fully automated, helpful suggestions or disable AI completely.

Chat allows you to conversationally query everything from sales calls to support tickets, with every answer traced back to the source. It instantly understands your context – whether you're looking at a single transcript, an entire Project or Channel, a specific folder, or even looking across your whole workspace.

Learn more about Contextual Chat in Dovetail →

Automatically find and highlight key moments in your data. Magic highlight will also use your existing tag structure to automatically classify and group highlights for you.

Learn more about magic highlight →

Automatically group highlights with thematic similarities on your canvas. Themes are created from the content of your highlights, not the tags or titles. Titles will also be automatically generated for each group.

Learn more about magic cluster →

Magic insight provides a powerful starting point, handling the initial heavy lifting of synthesis so you can focus on refining, validating, and driving action with your stakeholders. Use pre-defined prompts or create your own to quickly synthesize and structure your data, powered by AI.

Learn more about generating insights from data in Projects and Channels →

Save time identifying key themes in interviews, documents, or customer feedback, and turn them into valuable insights using AI. Add data to your project and insights - including content like PDFs, reels, and transcripts - and we'll automatically generate a summary of the key points.

Learn more about magic summarize for data and insights →

Enable your team to self-serve findings and share evidence-backed insights quickly by automatically summarizing search results, complete with citations.

Learn more about magic search →

Protect your participant's PII by blurring and muting video and audio content, as well as redacting text content within your transcripts.

Learn more about magic redact →

Using LLM and ML techniques, channels continuously analyzes incoming data and classifies it, allowing you to track themes in large data sets, such as support tickets, app reviews, feedback, and more.

Learn more about channels →

Dovetail indicates when AI has contributed to your analysis and content in the workspace. The orange magic shuriken icon will appear as a contributor on summaries, highlights, and any other content generated by AI. If a user edits a summary or accepts a highlight, their avatar will also be added, making it easy to see where human input was involved.

Why should I use Dovetail’s AI features over ChatGPT?

Dovetail’s tailored AI infrastructure means that your data won’t be used to train models for Dovetail or other customers—it’s fully secure. We select a model for the task at hand—whether that’s summarizing an insight or clustering highlights by theme. While ChatGPT enforces character limits, tailored infrastructure means that our AI features can handle whole transcripts and multiple highlights simultaneously.

Using Dovetail’s AI features ensures that all your customer data is in one place, so you don’t need to copy and paste data between tools and risk human error.

Will my data be secure if I use Dovetail's magic features?

We understand research data can contain lots of personal and commercially sensitive information, and participants trust you to keep it safe. That’s why we are committed to keeping this data secure and confidential.

Unlike tools like ChatGPT—which may use your data to train their models—we use tailored processing infrastructure on AWS, meaning your data remains your own. We deploy all our models in the same place it’s already stored. The request is sent to the model, and the response is returned. Models aren’t learning from your data.

You can read more about our data handling practices in our MSA (see in particular section 4, section 6, and section 11), our privacy policy, data processing agreement, and Dovetail trust center.

How is the model trained and deployed?

Dovetail uses a variety of market-leading LLMs. No customer data is used to improve or train our model—all training occurs before the models are deployed. Our models are constantly updated and improved to ensure you get the best experience.

Which Dovetail features use ML vs. generative AI?

Depending on the task, Dovetail’s AI features use machine learning (ML) or a combination of both ML and generative AI.

ML powers Dovetail’s transcription process. It allows us to identify positive and negative sentiments in transcripts and identify and blur faces to protect your users’ privacy. It’s also used to identify and cluster highlights by theme in canvas.

We use generative AI to summarize notes and insights that make it easy for you to keep stakeholders up-to-date. We also use generative AI to label your themes in canvas.

Can I turn off Dovetail's AI features?

AI is foundational to many of Dovetail’s core features, including transcription and sentiment analysis, which means deactivating it for specific workspaces is not possible at this time.

Product specific terms

For more on Dovetail AI, check out our product-specific terms here.

What languages do you support for clustering with AI?

At this time thematic clustering works best with English. However the following other language are supported: Spanish, French, Arabic, German, Italian, Dutch, Russian, Ukrainian, Vietnamese, Japanese, Korean and Simplified Chinese.

We are currently monitoring customer feedback to understand how to improve this feature and the languages we support.

Do you train on user data for semantic search or magic summaries?

No, we use a generic AI model and don’t feed it any training data to ensure user data is kept private.