Matt Barbelli.
Traditional user research is often too slow and expensive to keep pace with modern design, but AI personas are closing the gap, writes Matt Barbelli, founder and managing director of Wonderful.
Getting feedback from real users has always been one of the most valuable parts of the design process. But it has also always been one of the slowest elements of it.
Done properly, it takes time, coordination and budget. You have to recruit participants, schedule sessions, run interviews, then pull together the findings. While the outcome is almost always worth it, because of the effort involved, it tends to happen at key moments rather than throughout the entire design process.
In between those moments, a lot of decisions are still being made on instinct, stakeholder opinion or limited data. Not because design teams want to, but because there has not been a practical way to test everything as they go. That's starting to change, thanks to, surprise surprise, AI.
While, happily, AI is not replacing user testing, it’s changing when and how often it can happen. The shift taking place is testing being something occasional to something that can be done whenever it’s needed.
At the centre of this is the use of AI driven personas.
When we talk about these personas, we’re not talking about quick prompts or surface level role play. An AI persona is a structured representation of a real user type, built from a combination of research, analytics, stakeholder input and past user interviews. It includes key details such as motivations, behaviours, goals, frustrations, levels of digital confidence and context of use.
These personas are not generated in isolation. They are grounded in real inputs and shaped into a consistent format that reflects how different users think and behave. In practice, they are developed as dedicated, interactive models tailored to each project, allowing design teams to test ideas against a consistent set of user perspectives throughout the design process.
For example, a persona might represent a first time user accessing a healthcare platform. They may be anxious about their results, unsure of their login details, using a mobile device and not particularly confident with technology. The persona reflects how that person thinks, what they are trying to achieve and where they are likely to struggle.
The AI layer allows teams to engage with these personas in a dynamic way. Designers can walk through a journey and ask how they would respond at each step, testing messaging and see whether it feels clear or confusing. Different approaches can be explored and gaining a better understanding of how they might land for different types of users.
The AI is not the source of truth, but it’s a layer that helps bring existing understanding to life. This means ideas can be tested much earlier and concepts can be explored before too much time has been invested. Journeys can be walked through from different perspectives to spot friction or confusion and messaging can be tested to see how it is likely to land.
What used to take weeks can now happen in minutes. But the real benefit is not just speed, it’s the ability to explore more options. Instead of committing to a single direction and validating it later, teams can test multiple approaches and refine them as they go. It creates space for better design thinking, not just faster output.
That said, it is important to be clear about what this is and what it is not. I’m not saying AI personas are a replacement for real users. They do not bring the same emotional nuance or unpredictability that us humans deliver. While they can reflect bias depending on how they are set up, they should not be treated as absolute truth.
What they offer is direction. They help sense check ideas, surface potential issues and build confidence in a direction. But they work best alongside real user testing, not instead of it. The most effective approach is a combination of both. Use AI early and often to shape thinking, then bring in real users to validate and challenge. This is where the real shift is happening.
We’re moving away from long, linear processes and towards something more continuous. Ideas are tested as they are formed. Assumptions are challenged earlier and decisions are made with more context and less guesswork. The result is not just speed, it’s better outcomes.
When designers can test more often, better decisions are made. When better decisions are made, reworking is avoided and when that happens, more focus is placed on building stronger, more considered digital products.
User testing has always mattered, but what is changing is the ability to do it at the pace modern digital work demands. While AI personas do not replace real users, they make good user thinking available whenever it is needed and that changes everything.