Bili: Designing for Kids Who Can't Read Yet
Four years designing a bilingual learning app for 3- to 8-year-olds, from the first research to the rewards kids play for
Product design
UX for kids
Bili's original character art by Maried.
The short version
One constraint shaped everything: most of our users can't read the instructions, in either language.
Bili is a bilingual literacy app for kids ages 3 to 8, used in public school classrooms and at home, in Spanish, English or both. I've designed it since the first research in 2023, and I still design it today: navigation, reading, games, and the rewards kids play for.
THREE RULES
Three rules for kids who can't read the screen
Rule 1
Tap the thing itself
Five-year-olds can't read button labels, so the sticker, the heart and the door are the buttons.
Rule 2
One color per language
Spanish is red and English is blue, everywhere. Letters for each language sit side by side, so kids never drop a Spanish letter on an English word.
Rule 3
Nothing you earn goes away
Stars and flowers add up and unlock stickers. Nothing is spent, so collecting feels like growing.
Where it started: my 2023 concept prototype.
NAVIGATION
Four big doors on a path
A five-year-old can't read a menu, so navigation runs on pictures, color and sound.
In classroom observations, kids kept asking "now what do I do?" The 2026 home answers with four big doors on a path: Explorar, Leer, Jugar, Conectar (Explore, Read, Play, Connect).
The home, 2023 → 2026.
Read-along: each word lights up as it's spoken.
READING
A read-along in two colors
In our first pilot, families asked for words to light up as they're read.
I designed the read-along so each word lights up as it's spoken, Spanish in red and then English in blue. I chose both colors to pass accessibility contrast, and they match the language colors kids already know from the rest of the app. It shipped in June 2026. Stories also come in Quechua, with their own language switch, and in one pilot a Quechua story became a class favorite.
A Quechua story intro, and a vocabulary word: 'amable / kind.'
Story Factory · Word Lab · The Letters · The Odd One Out · Count with Me · The Thing Counter.
GAMES
Games in three languages
Every game works in Spanish, English or both, and every word can be heard.
Story Factory is a slot machine for silly sentences. Word Lab asks kids whether a word they built is real, and carries an honest note: "Ojo: Este juego fue creado por humanos y robots. A veces nos equivocamos. / Heads up: This game was made by humans and robots. Sometimes we make mistakes." Counting scenes come from real places, like the boats of Xochimilco.
A counting round, start to stars.
FEEDBACK
Feedback a pre-reader can see
A wrong tap glows red and nothing else happens. A right tap glows green.
Every round ends with a culture card and stars, with praise in both languages: "¡Sabía que podrías hacerlo! / I knew you could do it!"
"Students… eagerly respond to phonics and comprehension questions to earn a star." A kindergarten teacher in the first classroom pilot.
The prototype: from the profile to Rewards, opening gift boxes, and every new sticker landing in the album.
REWARDS
Rewards kids actually want
Kids earn stars for right answers, and flowers for activities with no wrong answer, like affirmations and family conversations.
In 2026 I designed what they earn them for: a profile with three doors, a rewards center and a sticker album. Nothing is spent, so totals only grow. Every locked sticker stays visible, so kids can see what's next. Tapping a gift box opens it, and the sticker assembles and goes into the album. The album has no numbers; it's for admiring, not shopping.
I built the clickable prototype the team tested with families, and framed one question for it: do kids understand where their stickers go, and can they find them again?
Profile · Rewards · Sticker album.
Empty states for kids who can't read: goals instead of a blank page.
Saving a favorite: the heart flies to your profile, so kids learn where it went.
Affirmation cards → how do you feel? → a flower.
WELLNESS
Feelings, gently
Early check-ins asked kids yes, no or "I don't know."
An early-childhood evaluation expert pointed out that kids give the answer they think adults want, so it became a five-step thumbs scale, with Bili shrugging for "No lo sé / I don't know." When the team tested my prototype, kids understood the scale without explanation and wanted to keep collecting flowers.
Yes / No / I don't know (2024) → a five-step scale (2026).
BUILDING IT
Designing with the developers, week by week
I worked with two development teams.
Early on, we ran a weekly rhythm: I designed one week, the developers built it the next. To keep screens working from iPad down to phone, I set layouts in percentages of the screen rather than pixels, so columns could drop from two to one without a redesign.
Many design decisions came out of those conversations.
The Wellness map
When the developers struggled to line up its pattern at the edges, I redrew it with a more forgiving texture.
The Dynamic Island
When new iPhones put it in the way, I moved the story buttons into the safe area.
The Yoga cards
When the cards needed more text, flipping one would have pulled a child out of the pose, so I designed a scroll instead.
When a new team joined in 2025, I reconciled every screen in Figma against what was actually built, cleared out old versions, and gave the developers Dev Mode access. Anything not ready was marked as a work in progress so nobody coded against it.
Who’s involved:
Product designer (me), two development teams, the Bili team, an early-childhood evaluation expert, and pilot teachers and families
THE RESULT
Live, awarded, and still learning
Bili is live on the App Store, Google Play and the web, and won the Community Choice award at the SXSW EDU 2025 Launch Competition.
64 learners
Piloted in 3 pre-K and kindergarten classrooms with 29 families, then at home with 24 more families.
496 hours
Logged by 8 students in a 2025–26 rural school pilot.
89%
Of families in the 2026 summer home pilot said Bili's stories reflect their family's background.
In the same pilot, 73% were very or extremely likely to recommend it. It also showed that 36% of kids hadn't earned a flower yet and 21% of families found them confusing. That finding started the rewards redesign I'm working on now.
Read-along shipped in June 2026, and rewards and the sticker album in spring 2026.
What I learned
Start from the hardest constraint
Most of our users can’t read, in either language. Treating that as the brief, not a problem to solve later, is where the three rules came from.
Let the data reopen your own work
The 2026 pilot showed 36% of kids hadn’t earned a flower and 21% of families found them confusing. So I’m redesigning the rewards I designed.
Bring research to the argument
On rewards, I came with research and the team came with their own ideas, and we met in the middle.
