User-centred design for the age of AI
Stunt Double sends AI actors through your designs, prototypes, and live products, reporting back exactly what they experience: user-centred design at the speed of AI.
Works where you work
- Linear
- GitHub
- Slack
- Figma
- Vercel
- Claude
- Webhooks
Cast any role
Real browser sessions. Real inboxes. Real credentials.
Bringing synthetic research and agentic AI together for an authentic user experience that’s genuinely scalable.
A stand-in for your users at every stage
From early interviews to automated checks to behavioural optimisation, Stunt Double gives you user-grounded feedback at every stage of the process. Open a stage to see how it works.
Qualitative research, on-demand
Recruit a representative panel, run moderated interviews, and get synthesised findings in minutes, not weeks. Stunt Double does the heavy lifting of qualitative research so your team can validate ideas and keep moving.
Recruit a panel in minutes
Cast actors you already trust or generate fresh personas for any audience. Every participant arrives with their own background, device, and point of view.
Guide the conversation
Draft a discussion guide of questions and think-aloud tasks, or let Stunt Double write one from your research brief, then refine it your way.
Watch them think aloud
Participants explore your live product, narrating every hesitation, delight, and dead end as they go. Full transcripts stream in as each session runs.
Findings, not just transcripts
The moment the last interview wraps, you get a synthesised report: key themes, verbatim quotes, and recommendations sorted by impact and value.
Participants
“I’d expect the total to update right here, but nothing happened. Now I’m not sure it worked.”
Test before you build
Interview marketing materials, Figma prototypes, or internal tools, so you can validate the design while it’s still cheap to change.
Validate every iteration
Re-run the same guide on each new version and compare findings round over round to prove what actually improved.
Recall from memory
Actors remember what they’ve told you before, so feedback stays consistent across rounds and builds on every prior session.
Better experiences, together
Bring actor insights and real human feedback into a single loop.
Human feedback too
Capture perspectives from both your AI actors and real human testers in one place.
Capture comments
Enhance prototypes, websites, and designs with annotated feedback that lets your team review, share, and collaborate together.
Suggestions
Get actionable recommendations tied to your goals, ready to action
Measure progress
Re-run journeys after each iteration to see what improved and what still needs work.
Fatima
Legal reviewer
Chen
Compliance officer
Sarah
First-time user
Marcus
Power user
Priya
Non-technical manager
Meet your cast
AI agents that can act, Aictors take on any role from internal stakeholders to technical end-users, each with their own background, expertise, and perspective. They browse with real sessions, their own email and knowledge, and surface the insights that only come from genuinely using what you've built.
Questions, answered
Everything you need to know before you put your product in front of a cast.
What is Stunt Double?
Stunt Double deploys a cast of AI actors, each with a distinct persona, to navigate your product like real users. They work through journeys, complete tasks, and report exactly what they experience, backed by screenshots, transcripts, and clear recommendations.
How is this different from traditional user testing?
It complements it. Your actors give fast, repeatable feedback at any hour, across personas and devices, so you catch the obvious issues before spending real users’ time. You can bring human testers in too, and Stunt Double folds both into one set of findings.
Can I validate a design before I build it?
Yes. Point an interview at a Figma prototype and your panel reacts to the design frames, so you can test the direction while it is still cheap to change.
How trustworthy are the findings?
Every finding is grounded in evidence: the actor’s reasoning, the steps they took, and screenshots of what they saw. You can judge each recommendation on the proof rather than take it on faith.
Which tools does it work with?
Linear, Slack, GitHub, Figma, Vercel, and webhooks, plus a Model Context Protocol server so you can run journeys straight from Claude, ChatGPT, or any AI assistant.
How do I get started?
Create your first actor in minutes and run your first journey the same day, then upgrade when you are ready to bring in the whole team.
