Working with personas has long been established in product development – and at the same time is often too flat: nicely designed profiles that are presented once and then rarely used again. Thomas Immich, Managing Director of Centigrade and founder of the AI-supported tool LeanScope AI , is fundamentally rethinking this approach. For him, personas are not static images, but interactive projection surfaces – capable of dialog, adaptive and constantly evolving due to new findings in user research. In an interview with Annett Bergk, he explains how generative AI can accelerate and deepen the use of personas and make them more human at the same time. And why true-to-life text-based descriptions sometimes achieve more than a supposedly perfect profile picture.
Mr. Immich, what makes a “good” persona in the age of generative AI?
It must help to make better feature decisions in the respective product context and be as realistic as possible, i.e. reflect the findings from user research. With AI, personas become alive and interactive, you can talk, discuss and brainstorm with them… and they continue to develop. This makes storytelling and behavior models more relevant and involves teams more closely or distributes target group knowledge in a very intuitive and lightweight way.
What was the impetus for developing LeanScope AI?
LeanScope AI is essentially a Lean UX management tool – the product has been around for around seven years. Originally, it was used to structure the entire product discovery process and bundle human artifacts such as jobs to be done, usage scenarios, user stories and user journey maps in one place, but we later added AI to LeanScope so that we could feed the findings from the user research process back to the team in a more lively way.
Where do you see the greatest need for AI-supported personas today?
Where traditional processes fail due to a lack of emotional connectivity. AI personas translate data into a tangible, human form and inspire teams to think in new ways and come up with more accurate ideas by asking questions.
How does the process work with LeanScope AI?
At Centigrade, we usually start with a completely AI-generated proto-persona as a warm-up that you can talk to directly. The conversation then results in a customized interview guide, which can then be used for interviews with real users. The feedback from the users flows iteratively back into the AI persona – often via an AI transcript from a meeting supervisor – so that this persona can learn and interactively participate in discussions in workshops or virtual focus groups.
What distinguishes your approach from traditional UX processes?
For us, personas are projection surfaces validated by user research, which can speak and promote empathy with gamification elements, instead of being freely invented 3D avatars in the Uncanny Valley.
What do you mean?
A well-formulated, linguistically condensed persona that you can talk to activates the imagination more than a visually charged representation with a photo, hobbies and favorite coffee. Authentic, interactive dialog promotes empathy and curiosity more than visual superficialities.
Greatest added value through AI
The persona has a memory and does not forget, reacts continuously to changes in the project or new findings in user research or product testing and becomes an interactive sparring partner.
Are there limits?
AI cannot discover completely new target groups, it can only combine existing ones. Large language models are not miracle machines, but they can consolidate perspectives and stimulate thought processes. Above all, they remind us of the right things at the right time, things that are otherwise easily forgotten in the many distractions we experience in everyday project work these days For example, a project team learned from the AI persona that their idea for facial recognition could not work because the persona was wearing safety goggles on the factory floor. Everyone was very surprised when they heard this and at the same time glad that the idea was not developed any further.
Where is the topic heading?
AI will soon be able to independently obtain feedback, operate interfaces and test user reactions. This makes it all the more important that humans determine the long-term goals and values of products. For example, an AI could recommend optimizing an app based on the number of clicks – but whether this will help us socially and promote “humanness” is questionable… you Tiktok. We must never relinquish this responsibility for the product.
