Future of UX | Your Design, Tech and User Experience Podcast | AI Design

chevron_right

#143 Synthetic Users, AI Interviews, and the Future of UX Research with Julian Della Mattia

whatshot 23 snips

Feb 12, 2026

Guest

Julian Della Mattia

Julian Della Mattia, senior user insights manager at DuckDuckGo who builds research infrastructure, discusses synthetic users and AI in research. He explains what synthetic users are and when they help. He covers risks like polished, misleading answers, AI-moderated interviews, bias and validation, and how researchers’ roles may shift toward orchestration.

47:52

forum

Ask episode

web_stories

AI Snips

view_agenda

Chapters

menu_book

Books

auto_awesome

Transcript

info_circle

Episode notes

volunteer_activism

ADVICE

Use Synthetic Users As A Pre-Stage

Use synthetic users as a low-risk pre-stage for polishing protocols and early prototypes.
Combine synthetic practice with real-user testing before making product decisions.

volunteer_activism

ADVICE

Build An AI User Gym For Training

Create an internal user gym with AI-generated personas to train non-researchers.
Pair the gym with documentation and vetted question banks to raise research quality across teams.

insights

INSIGHT

AI Answers Tend To Be Too Polished

AI-generated responses often sound polished and implausibly idealized.
Human answers reveal messy context and emotions that matter for real insights.

Get the Snipd Podcast app to discover more snips from this episode

Julian's background and research focus

02:41 • 3min

chevron_right

Research changes and persistent challenges

05:28 • 3min

chevron_right

How AI surfaces research trade-offs

08:13 • 2min

chevron_right

Defining synthetic users and how they're made

10:27 • 2min

chevron_right

Useful cases for synthetic users

12:52 • 3min

chevron_right

Risks: polished answers and lack of messiness

16:11 • 4min

chevron_right

Data needs and limits of training synthetic users

19:52 • 2min

chevron_right

Bias and stereotypes in AI-generated personas

21:47 • 4min

chevron_right

Practical advice when company mandates synthetic users

25:25 • 3min

chevron_right

AI-moderated interviews: strengths and limits

28:06 • 4min

chevron_right

Empathy and engagement gaps with AI moderators

31:43 • 2min

chevron_right

Future integration: context and company knowledge

33:33 • 2min

chevron_right

Preparing researchers for an AI-assisted future

35:41 • 4min

chevron_right

Opportunities: freed time and wider research adoption

39:22 • 3min

chevron_right

Researchers as validators and toolsmiths

42:19 • 2min

chevron_right

Resources and next steps for research ops

• Mentioned in 6 episodes

Research That Scales

Kate Towsey

#1524

• Mentioned in 27 episodes

Four Steps to the Epiphany

Steve Blank

The Four Steps to the Epiphany by Steve Blank is a comprehensive guide for entrepreneurs, focusing on the customer development process rather than the traditional product development model. The book highlights the importance of understanding the market environment and customer needs before developing a product. It outlines four key steps: customer discovery, customer validation, customer creation, and company building. Blank argues that startups must learn and adapt continuously, unlike large companies which have established customer bases and market knowledge. The book provides insights from failed startups to illustrate the critical aspects of achieving and sustaining success in the entrepreneurial journey.

In this episode of Future of UX, I’m joined by Julian Della Mattia, Senior User Insights Manager at DuckDuckGo.

I first came across Julian through his conference talk on synthetic users, and it immediately stood out. Not because it was hyped, but because it was thoughtful, critical, and refreshingly honest. At a time when AI research tools are popping up everywhere, Julian brings a grounded perspective on what actually helps, what quietly misleads teams, and why judgment still matters more than speed.

In this conversation, we talk about:

What synthetic users really are and what they’re not
Where AI-generated users can be useful and where they become risky
Why “plausible” answers are often more dangerous than wrong ones
AI-moderated interviews and what they do well (and badly)
Bias, validation, and responsibility in AI-supported research
How the role of UX researchers is shifting from execution to orchestration

This episode isn’t about replacing research with AI.
It’s about learning how to use AI as a tool without losing context, empathy, or decision quality.

If you work in UX, research, or product, and you’re trying to make sense of how AI is changing research practice right now, this episode will give you a much clearer mental model.

Resources mentioned:

Follow Julian's work: Substack

AI for Designers: 5-week Bootcamp 💻

→ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Sign up for the ⁠waiting list⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

👉 0€ AI mini Training for Designers 👈

→ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Learn 3 strategies to leverage gen AI as a designer in just 30 min⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

💌 Newsletter

→ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Free Weekly UX insights and updates: UX Newsletter⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

Follow the Future of UX:

🎨 Linkedin: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Future of UX Podcast⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

📸 Instagram: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Future of UX Podcast⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

🍏 Apple Podcast: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Future of UX⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠