Designing for kids who can't read yet
Designing for kids who can't read yet means replacing every word in the interface with voice, motion and characters children trust. This is the story of Hachi Eitan, a social-emotional learning series and interactive game from Hop! Channel for children ages 3-7: how the project unfolded, what testing with real kids broke and fixed, and how an AI pipeline kept more than 500 illustrations consistent across two characters.
The brief
It started with a brief that sounds simple and isn't: Hop! Channel wanted to help Israeli children ages 3-7 build emotional resilience. The content would touch their reality directly - a dad away in reserve duty, sirens, loss - and had to make all of it gentle enough for a five-year-old. The initiative, Hachi Eitan, launched in February 2026 as a TV series, an interactive game, and tools for the parents and educators around the kids.
My side of it: the game interface for mobile and desktop, every visual, and the characters and their whole world. The content side belonged to SEL.IL, the social-emotional learning center from Reichman University, working with psychologists.
First, watch everyone else.
Before drawing anything, I ran competitive research on apps for the same ages. What do three-to-seven-year-olds actually enjoy? Where do other products lose them? Most of my future users can't read - not "read slowly", can't read - so I needed to see how the good ones replace text entirely.
The answer that kept coming back: they don't replace it, they route around it. Voice-overs carry the meaning. Visual cues carry the navigation. Motion tells a child what is alive and tappable better than any label ever will. Buttons come big, clear, and one after another, because hierarchy for a four-year-old is not a layout, it's a sequence.
Then the platform we were building on added its own technical constraints, and the design had to live inside them. That's the job.
Building a world kids can trust.
The series follows Eitan, a five-and-a-half-year-old, and his doll Toni, who comes to life in his imagination as a supportive inner voice. Around them sits a whole world: the rooms of the house, a garden, family members, secondary characters. Kids this age trust what stays the same, so consistency wasn't polish, it was the whole foundation: same characters, same places, same faces, every single screen.
The game lets kids play as Eitan or as Tamar, a girl his age, across 15 journeys. Each journey is 3 chapters with 2 choice junctions, each chapter 3 illustrations. The journeys work like a very simple simulation: they show children what their choices mean.
Which created a production problem. Fifteen journeys, two characters, hundreds of illustrations, and every face had to stay on-model, down to facial expressions in video. This is where AI did real production work. I built a dedicated internal tool for creating, editing and approving illustrations. We generated Eitan's complete image set first, got it approved, and then used the tool to produce the same set for Tamar. That one decision saved hours of manual work. The final library: more than 500 consistent illustrations across 15 interactive stories.
The content loop.
The division of labor with SEL.IL turned out to be the best part of the project. Their psychologists took the TV episodes and wrote the journeys and choice options from them. I created the matching visuals. Then the loop: a mid-chapter meeting to approve the direction, and an end-of-chapter review to check that the emotional messages they wanted actually landed. A great way to work. Everyone came out happy.
Then the kids arrived.
We tested internally with every kid we could reach, and after that with parents in an early-adopters group. Watching a pre-literate child use your interface is the fastest UX education there is. You see exactly where they get stuck, where they tap randomly, and where they surprise you completely.
What changed after testing: fewer clicks, rewritten voice-overs, and a flow trimmed until starting an activity took the absolute minimum. We removed whole parts of the flow. One button after another, no detours.
Except in one place, where we deliberately added a click.
Choosing an emotion is the meaningful moment of the game, and a meaningful choice deserves a pause. So selection works in two steps: the first tap focuses a card and plays a voice-over naming the emotion. The second tap confirms. A child can't end up somewhere by accident, and gets time to actually think. Reduce friction to move. Add friction to decide. That principle survived every round of testing, and it's the thing I'd most want another designer to take from this project.
What stays with me.
Designing without words changes how you think about interfaces. Assume zero reading. Treat hierarchy as a sequence, not a layout. Keep the world consistent, because consistency is trust. Test with real children early, because they will break your assumptions in the first five minutes.
I'm not allowed to share usage numbers for this project, so I won't pretend to. What I can share is the story, the principles, and the game itself - which shows all of it better than metrics would. The series and the game also grew into something wider: tools and video guides for parents and educators, so the emotional vocabulary kids learn on screen keeps going at the dinner table and before bed.
Kids who can't read yet, clearly.
How do you design an app for children who can't read?
Replace text with voice-overs and visual cues, keep one clear action per screen, and test with real children early. For moments of meaningful choice, a two-step interaction (first tap focuses and narrates, second confirms) prevents accidental choices without frustrating anyone.
What is Hachi Eitan?
A social-emotional learning initiative by Hop! Channel, launched February 2026: a TV series, an interactive game and tools for parents and educators, helping children ages 3-7 build emotional resilience. Developed with psychologists and SEL.IL from Reichman University.
How was AI used in the project?
A dedicated internal tool for creating, editing and approving illustrations kept more than 500 images consistent across two characters and 15 interactive stories. Once one character's image set was approved, the same set was generated for the second character, saving hours of manual work.
How did you keep a whole world consistent with AI?
By treating consistency as a process, not a single prompt. The world was much bigger than two characters: places, the rooms of the house, a garden, family members, secondary characters, and every pose and facial expression each of them might need. Each piece had its own approved reference image, and everything after it had to match. We generated Eitan's complete set first and got it approved before touching Tamar's, so there was always an approved reference the rest of the library could be checked against.
What did the internal illustration tool do?
It was the creation, editing and approval pipeline for the whole illustration library. Because generation, fixing and sign-off lived in one place, producing Tamar's set after Eitan's was approved took hours less than building it from scratch, and every approved image stayed on-model, down to facial expressions in video.
What age group was this designed for?
Children ages 3-7, most of whom can't read yet, together with the parents and educators who help them talk about feelings.