The Interview Agent is a powerful AI moderator available in the Discuss platform. I can be accessed and used from the Self-Paced Feedback (Self-Captures) page, and new activities that include AI-led audio tasks can be created. These tasks allow researchers to gather qualitative insights through natural, real-time conversations moderated by AI.
Access and Availability
Access to the Interview Agent is currently available to organizations who:
- Are on a ConsumerReach or ConsumerReach+ subscription, or
- Have added the Interview Agent to their subscription
Users with the interview agent will see AI-led audio tasks as the default task type in the builder. If it’s not enabled, tasks will default to Video.
Creating and Customizing Activities with AI-led tasks
To create an AI-led activity, users begin on the Self-Captures page and select Create New Activity.
Each activity includes a Goal field, where researchers can provide background on the study. This information is shared with the AI and helps align follow-up questions to the study’s objectives. Researchers can also choose from over 50 supported languages for their interviews.
Researchers can also add a Welcome Message. This will be displayed to the respondents as text before they launch the AI-led conversation.
The AI-led Audio task is selected by default for eligible users. Researchers can chain together multiple AI tasks to form a seamless conversational experience, or they can mix in other task types, such as ranking or multiple choice.
Note: Stim cannot be added to AI-Led Tasks at this time.
Task questions are written by the researcher, and follow-ups are generated automatically by the AI. These follow-ups are based on the respondent’s latest answer, as well as the overall context of the conversation.
Users can configure their activity using the existing Activity Settings. However, activities with AI-led tasks must be displayed as Single Tasks.
Note: There is no preview mode at this time, but users can simulate the experience using a test project. A built-in preview is currently in development.
Steps to create and customize an AI-led Activity:
- Navigate to the Self-Captures page from the main menu.
- Click Create New Activity to start a new self-paced session.
- Provide an Activity Name that clearly identifies your project.
- Add a brief but informative Goal statement to help the AI understand the purpose of your study.
- Choose the language your respondents will speak.
- Ensure AI-Led Audio is selected as the task type.
- Within the new activity, select the task editor to begin adding questions.
- For each AI-led audio task, write your task question using clear, open-ended language.
- Add additional AI-led tasks as needed to guide the full interview.
- Optionally add other task types (e.g., ranking, multiple choice) to collect additional feedback.
- Review the task flow to ensure a logical, conversational progression.
- Save the activity and distribute it via the options on the Launch page.
Respondent Experience
When a respondent is sent an Activity with an AI-led audio task, they will be greeted with the Activity Welcome Message and will be prompted to begin the conversation. This will open a chat with the moderator. The respondent will let the moderator know when they are ready, and after a short greeting, the moderator will ask the first Task Question.
The moderator will then ask 1–3 follow-up questions before moving to the next task. If the next task is an AI-led audio Task, the AI-moderator will seamlessly ask the question as part of a single conversation. If the next task is NOT AI-led, the user will go back to the main Activity screen to complete the tasks.
Activities can be set up with multiple blocks of AI-led audio questions. Here is an example of the respondent's experience.
Follow-Up Behavior and Timing
The Interview Agent is designed to probe for additional detail and clarity from respondents. Each task question can have up to three follow-ups, but fewer may be asked if the respondent’s answers are clear and detailed. Follow-up responses are typically delivered within two to three seconds to maintain a natural pace. If anything goes wrong during the session, respondents can cancel out and rejoin at any time.
Reviewing Responses
When reviewing responses, researchers will see the same interface as with other self-paced feedback tasks. They can view results by task or respondent. Audio is displayed in a dual waveform layout: the top waveform shows the respondent, and the bottom shows the AI moderator.
Researchers will be able to automatically translate AI-led responses into English using the existing toggle. Transcripts include speaker labels, so it’s easy to distinguish who said what. AI follow-ups are nested under the task question to preserve the conversational flow, rather than being listed as separate tasks.
Note: Researchers cannot request a redo or create highlight reels from AI-led tasks at this time.
Summarizing and Analyzing Responses
Researchers can use the Self-Paced Summary tool to generate summaries for AI-led audio tasks. The process is the same as with video and open-ended text responses. The system automatically filters out AI moderator dialogue, so all quotes and insights are drawn solely from what respondents say. Researchers can ask their own custom questions (e.g., “What do participants value most about this product?”), Or choose from a bank of Frequently Asked Questions.
Researchers can also combine summaries from AI Audio tasks with responses from other open-ended tasks in a single view, allowing for comprehensive qualitative analysis.
Summaries can be downloaded or saved.
Note: Quotes are currently available for AI-led tasks, but researchers cannot play clips of the source audio on the Summary page at this time.
For more on the subject, you can review this article: Self-Paced Summaries With Genie
Language Support
Supported languages include:
|
Afrikaans Arabic Armenian Azerbaijani Belarusian Bosnian Bulgarian Catalan Croatian Czech Danish Dutch English Estonian Finnish French Galician German Greek Hebrew |
Hindi Hungarian Icelandic Indonesian Italian Japanese Kannada Kazakh Korean Latvian Lithuanian Macedonian Malay Marathi Maori Nepali Norwegian Persian Polish |
Portuguese Romanian Russian Serbian Slovak Slovenian Spanish Swahili Swedish Tagalog Tamil Thai Turkish Ukrainian Urdu Vietnamese Welsh |