trolls.dev
04 The lab — case study

AI User Interviews

Try your questions with AI participants.

AI User Interviews runs simulated interviews using AI moderators and participants. Give it a discussion guide to explore across a hundred-plus interviews a week, then review the transcripts and summarized findings.

Status
Live
Platform
Web
Field
User research · AI
Cadence
100+ interviews / week
Open the platform ↗ All work
aiuserinterviews.com Live
The room
AI What nearly stopped you signing up?
In parallel 24 live
synthesizing 24 interviews… 3 themes
calls scheduled · 0 run while you sleep
Try out a discussion guide before recruiting people for a study. The bet behind it

Why we built it

A way to try questions before a customer study.

Customer interviews take preparation. You need to recruit people, find times to meet, and work out what to ask. A question that looked useful on paper may fall flat once the conversation starts.

We built AI User Interviews to help explore a discussion guide before committing to a study. You can run variations with simulated participants and inspect the conversations they produce.

The AI moderator asks follow-up questions, and the platform summarizes the transcripts. That gives you material to review as you decide which questions need more work.

The answers come from models. Use them to prepare for research with customers, whose experiences may be very different.

What we made Features
  1. 01

    Follow-up questions

    The AI moderator follows your discussion guide and asks follow-up questions based on the AI participant’s answers.

  2. 02

    Run interviews in parallel

    A study can run many simulated interviews at once, without recruiting participants or scheduling calls.

  3. 03

    Read across the transcripts

    The platform groups recurring themes and points of disagreement, with quotes from the simulated conversations.

  4. 04

    Set the study yourself

    You write the discussion guide and define the AI participants. You decide which questions to explore and how to interpret the results.

Design choices

Make the simulations easy to inspect.

Make room for follow-ups

The moderator can ask about an answer that needs more detail. That lets you explore how a discussion guide might develop beyond its opening questions.

Link findings to transcripts

Each finding links to the transcripts and quotes behind it. You can read the conversation to check whether the summary is fair.

Be clear about AI participants

These interviews can help you try out questions and explore possible responses. AI participants cannot tell you what your real customers think; that still requires research with people.

Further reading from the journal

We wrote about designing around a specific use in The ten minutes before the meeting, and about running our own products in Why we only build what we run.

AI User Interviews uses AI participants and AI moderators. Its transcripts are simulated conversations, and findings should be checked through research with real customers.

Colophon End of Case Study 04

Try your interview questions.

Write a discussion guide and run it with AI participants. Read the transcripts to see which questions you want to revise.

Open the platform ↗ More case studies →