CASE STUDY
JustGrow: an AI-empowered parenting companion
One app for the whole job of early parenting: tracking, memories, and family planning.
Problem
Trackers held the baby’s data, but the advice lived somewhere else
When a parent wondered “is this a sleep regression?”, the answer came from search engines and forums that had never seen their baby’s logs. Connecting the two was one more job for parents already short on sleep.
I joined when JustGrow was just an idea. The founder owned direction; I owned UX, the Flutter frontend, parts of the backend, and the agent prompts.
Challenge
Parents had no time to learn a new app, so switching had to cost nothing
JustGrow asked them to move tracking, photos, and family planning into one app. To be worth switching, it had to be as convenient as the apps they already used, with smoother flows and fewer taps. Every feature, and the screen that came with it, had to earn its place.
Solution
One chat to log everything, and a relay race to hand off care
Parents log a feed or a nap in plain language, and AI agents turn it into events. Gleam Calendar lists those events and their durations, so parents and caregivers can see the day at a glance.
Screens on this page are rebuilt in HTML with JustGrow’s design-system tokens, so retired, shipped, and proposed designs read as one set. Some are interactive; try them.
Those same events power the handoff. The founder wanted passing care between caregivers to be easy, and invisible work to be visible. I designed the handoff as a relay race: pick the children and your partner, hand over, and the receiver reviews recent events before accepting. I built it end to end, including edge cases such as two simultaneous handoffs and handoffs that expire before anyone accepts them.
Launch
We launched a fully functional tracker, in English and Chinese
The launch included tracking with the chat agent built in, growth charts to visualize the data, and the care handoff. We checked the localization by hand instead of relying on machine translation, so both languages use the words parents actually say. For grandparents abroad, times always show in the baby’s timezone.
Everything starts in chat. One sentence becomes an event, and the ☰ menu or a link in the reply opens it on the calendar, where the rest of the family sees it too.
Iteration
I designed a replacement that asks nothing
In an App Store review, a night nurse called the handoff “perfect for overnight care.” But analytics showed few families used it. Every handoff needed both caregivers to remember a step: one to hand over, the other to accept. I had designed for an idealized parent whose only concerns were the baby and the app. Real parents are juggling far more.
A better entry point would have repeated the same mistake, just cheaper. Every tracking event already recorded who did it. The app didn’t need parents to declare a handoff; it could read what they were already logging.
I proposed a replacement that lives in chat. Credit comes from existing events, and a weekly digest shows the family total plus a private count of your own. No new screens, no new agents. The founder retired the handoff; I handed off the replacement plan before I left, so I didn’t get to build it myself.
Beyond tracking
I consolidated: one place for memories, one design system
Photo sharing, from uploading to browsing, became Memory: one timeline where the family posts photos, videos, and notes. With privacy settings, it doubles as a private journal. I built its media pipeline end to end: storage on Google Cloud Storage, upload and access permissions, thumbnail generation, and caching.
In addition, tokens, type, and components now live in one system, with a log of deliberate exceptions so no one “fixes” them later.
Me doesn't feel well
journalImpact
200+ signups in nine months, with no paid marketing
Takeaway
Wanting an outcome isn’t the same as being willing to pay for it
Parents wanted to share the load, but the handoff asked for their attention at the worst moment. Now I weigh what a design asks of people before I scope it.