Inside Cohort 1, Week 2: 103 parents, two personas, and the limits of an AI user

Our second Beginner UX (AI) Design class was about user research, and it started with a gap every designer runs into.

A business always has a picture of its users: who they are, what they need, why they'd pay. Some of that picture is right and some of it is guesswork. Natalia's framing for the week was that user research is how you find out which parts are which. The students were about to interview famagic's founders, so the timing was good. They'd hear the business's view on Friday, having just learned how to check it.

Design thinking, and why it isn't a straight line

Natalia introduced the Design Thinking process commonly used in UX: Empathise, Define, Ideate, Prototype and Test. Although often presented as five stages in sequence, the process is iterative rather than linear. Teams move back and forth between stages as they learn from research, test assumptions and gather feedback, refining their understanding of the problem and potential solutions along the way.

She shared an example from her own work. On an enterprise platform, her team built a solution based on the marketing team's research, only to discover through one-to-one user testing that it solved the wrong problem. This highlighted why UX research is essential to validate assumptions about user needs rather than relying on what teams believe users need.

Empathy maps, personas and the AI version

Next came the tools that turn interviews into something a team can use. An empathy map splits what you heard into what the person says, thinks, does and feels. Natalia's point was that quotes are worth gold here. A user's exact words, repeated across several interviews, are your evidence when you need to argue for a direction.

Personas come from the patterns. A persona is a fictional person who stands for a real group with the same needs and motivations. Natalia showed one from a corporate learning platform she'd worked on. The team tried to group users by age, then by job, and neither worked. What split them was motivation: people who learn without being pushed, people who are neutral, and people who aren't engaged at all. Seniority and department made no difference.

She then addressed the elephant in the room: synthetic users, or AI personas. They're quick, affordable and useful when access to real users is limited, but rely on existing data and can miss unexpected insights. Real users can reveal problems teams didn't know to look for. Her advice was to use AI and real-user research together, understanding the strengths and limitations of each.

Interviewing a classmate about their living room

The practical part was a short exercise: pair up and interview each other as if you were redesigning the other person's living room. The brief was about the experience of the room, not how it looks. How do you use the space, what annoys you, what are your habits?

Each pair had about five minutes per interview, which was tight on purpose. Most people came out saying they'd left questions unasked. Natalia said that's what a real interview feels like. You can't ask the client to stay longer, so you decide in advance what you most need to know.

One question stood out. Someone asked whether the flat was rented, because that changes everything you could suggest. Natalia picked it up straight away. In a digital product, the same question is about constraints: what's technically possible, and how far you're allowed to go.

Then she grouped the answers on the spot, and two rough personas appeared: people who want the room for unwinding after work, and people who use one space to work, eat and relax. It was a living-room-sized version of the exercise the students now have to do for real.

Which research method, when

The theory section covered research in two ways: primary (you collect it) or secondary (someone else already did), and qualitative (why and how) or quantitative (how many). Then came the methods: interviews, surveys, usability tests, focus groups, diary studies, card sorting, A/B tests, unmoderated testing and analytics.

To make it stick, the class went through five scenarios and held up their answers. Why do people quit a fitness app after a week? A survey to find the pattern, interviews to find the reason, and analytics to see where they drop off. Where do people struggle to send money in a banking app? A usability test, every time. Natalia told the class that a beginner course won't make them researchers. She wants them to know which method fits which question, because that's what makes a designer sound professional in a meeting.

Her favourite example of secondary research came from the Evening Standard. While she was working on a dating app, she read an article in it that listed rival apps she'd never heard of. Good research isn't always formal. You just have to know where it came from and whether you can trust it.

What the survey said

famagic didn't start from guesswork. Before the course began, they surveyed 103 parents and shared the anonymised answers with the class. It's 50 questions, a mix of multiple choice and free text, and it's hard to read as a spreadsheet. Natalia ran it through AI tools and ended up with a 20-page report from different prompts. Three findings stood out:

• Parents already use apps to organise family life, but they use several, and keeping them in step is a job in itself.
• School information arrives from everywhere: email, WhatsApp, paper letters and notices on the wall at pick-up.
• Nobody's sure who owns what. One parent carries most of the mental load and the other wants to help but keeps missing things.

famagic already uses AI on that second problem. Snap a photo of a school letter, a notice on the nursery wall or a party invitation, and the app picks out the dates and to-dos and turns them into calendar events and tasks you can add with a single tap. It works for whole schedules too, like a music school timetable or a year of swimming lessons, and for screenshots of emails and messages. It'll even turn a handwritten shopping list into items on your list. There's also an AI meal planner that works around allergies and what the children will actually eat, and a personal AI family assistant is in the final stages of development.

From the third finding, Natalia drafted two personas: the parent who manages the mental load, and the co-parent who wants to share it. Why only two, when the app also serves grandparents, childminders and babysitters? famagic asked the team to focus on parents for this project.

The rest of the week

The students analysed the survey themselves with whichever AI tools they liked, started turning the two personas into a team version, and got set up on UXtweak, which they'll use for usability testing later in the course. On Thursday and Friday the teams met industry mentor Bobby King for a practical introduction to UX research, including a live usability-testing exercise. On Friday they interviewed famagic's founders. Natalia's brief for the interview was to start with "tell us about your product", listen, ask follow-up questions, and make sure everyone asks at least one question of their own.

Week 3 is information architecture.

Want to try this week's tools on a product of your own? The empathy map builder and the persona builder are free. For a fuller list of methods and when to use them, read our guide to UX research methods. Or see what's in the Beginner UX (AI) Design course.