Designing for learning means the product is judged by whether someone actually learns, not by how the screens look. From 2022 to 2026 I was Product Designer and AI Builder at MindCET, Israel's EdTech innovation center, building products like Chameleon.AI for mobile, web and iPad. These are the lessons that survived contact with real kids, real teachers and real classrooms.
Education products have a built-in conflict: the person using the product is rarely the person buying it. Kids learn, parents pay, teachers administer, schools procure. Every one of them has an opinion about what the product should be.
What four years taught me: serve the learner first, or nobody else gets served at all. A school can buy a platform and a teacher can assign it, but if the child disengages in four minutes, the renewal conversation next year is already lost. The learner's experience is the retention strategy.
That doesn't mean ignoring adults. It means the interface earns the adult's trust by working for the child, not the other way around.
Working on Hachi Eitan, the Hop! Channel game for children ages 3-7, made this concrete. Young kids trust what stays the same: same characters, same rooms, same faces, every screen. So consistency wasn't polish, it was the foundation the whole experience stood on.
That principle turned into a production system. Fifteen interactive stories, two playable characters, more than 500 illustrations, and every face had to stay on-model. We built an internal AI tool for creating, editing and approving illustrations, generated one character's complete set first, got it approved, and only then produced the second. The tool existed to protect the feeling that the world is stable.
Adults forgive inconsistency and call it a quirk. Kids experience it as the world breaking. The older I get in this field, the more I think adults just learned to tolerate broken worlds.
Every education product I respect was shaped by watching real learners use it. With Hachi Eitan we tested with every kid we could reach, then with parents in an early-adopters group. The result: whole parts of the flow removed, voice-overs rewritten, the path to starting an activity trimmed to the minimum.
And one place where we added friction on purpose. Choosing an emotion is the meaningful moment of the game, so the first tap only focuses the card and plays the voice-over; the second tap confirms. Reduce friction to move, add friction to decide. No internal review would have found that. A five-year-old found it in minutes.
If your testing plan starts after launch, you don't have a testing plan. You have a hope.
STAGA spent three years moving a music-education platform from teacher-led software to student-driven creation. The design problem was the pedagogy problem: how much structure does a student need to start, and how much freedom do they need to make something theirs?
The answer, every time we tested it, was: less structure than the adults were comfortable with. Given a real creative tool and a reason to care, students go further than the lesson plan. The design's job is to build the on-ramp and then get out of the way.
Sprint Lab taught the same lesson from the other side: a high-school science course that starts from the student's curiosity instead of the curriculum's order. Curiosity first, structure following behind. It works.
On Hachi Eitan, the emotional-learning content came from psychologists - SEL.IL, the social-emotional learning center from Reichman University. They wrote the journeys and choice options from the TV episodes; I designed how children experience them. Mid-chapter direction checks, end-of-chapter reviews to verify the emotional messages landed.
That loop was the best part of the project, and it's the model I trust: pedagogy experts own what is taught, design owns how it's experienced, and the two meet constantly. A designer who claims to own the learning science, or a pedagogy team that hands over a PDF and disappears, both produce the same thing: a product where content and experience drift apart.
I build with AI every day, and I design AI products for learners - Chameleon.AI at MindCET among them. AI compresses production: the 500-illustration pipeline, prototypes in days, one person covering design and build. What it doesn't change is the core loop: understand the learner, build something real, watch them use it, change your mind.
If anything, AI raised the stakes on taste and judgment. When anyone can generate a hundred screens, the skill that matters is knowing which one respects the learner.
Four years in, that's the sentence I'd put on the wall. Tools change every quarter. The learner doesn't.
My education work is collected on the EdTech page, and how I work with teams is on the hiring page.
That learning products are judged by whether someone learns, not by how the screens look. The practical lessons: design for the learner first even when someone else pays, treat consistency as trust, test with real learners early, cut ruthlessly, and let pedagogy experts own the content while design owns the experience.
MindCET is Israel's EdTech innovation center, part of the Center for Educational Technology. From 2022 to 2026 I was Product Designer and AI Builder there, designing and building learning products for mobile, web and iPad, including Chameleon.AI.
Assume zero reading, treat hierarchy as a sequence instead of a layout, and keep the world consistent because consistency is how children decide something is trustworthy. Every interaction has to survive being tapped by someone who cannot read the label.
Yes, if the product is built for it. STAGA's music platform moved from teacher-led software to student-driven creation, and the design work was exactly that shift: giving students enough structure to start and enough freedom to make something theirs.
With real learners, early, on a real build. Watching a child or a teacher use the product for five minutes teaches more than a month of internal review. Expect to remove features, rewrite copy, and occasionally add friction on purpose at moments that deserve thought.
Two roles. As a builder's tool: AI pipelines kept more than 500 illustrations consistent across the Hachi Eitan game, and I build with AI every day. As a product material: products like Chameleon.AI put AI in the learner's hands, where the design job is keeping it honest, paced and age-appropriate.
Yes. Since July 2026 I work freelance with founders, schools and education teams, from discovery to a working MVP. The fastest way to reach me is WhatsApp or email.