StarStarter

The brief

Anxiety is one of the most common mental-health conditions in the world, and most of the people it touches never receive treatment for it. Therapy that works exists. Access to it doesn’t.

Arcade Therapeutics, co-founded by a neuroscientist, was built to change that, to put a clinical tool directly into the hands of the people who need one. Their flagship intervention is Attention Bias Modification, a research-backed exercise where two faces appear on screen, one happy and one angry. They disappear, and the happy face leaves behind a small spaceship. The player swipes the ship to collect stars, six minutes at a time, until the brain learns to look toward warmth and away from threat.

They had the research, the clinical results, and a working prototype. What was missing was a game people actually wanted to come back to. That was the brief: keep the science untouched, and build a game around it that someone reaching for it on a hard Tuesday morning would love.

The approach

Most of what I had to learn was new on two fronts. Mobile development at this scale, the IAP plumbing, the App Store and Play Store release pipelines, the dozens of small mobile patterns you only feel when something breaks on someone else’s phone , was new. And so was weaving therapeutics into a game without bending either one.

The constraints were the usual ones, plus one that wasn’t: budget and team size shaped what we could ship in a given quarter, but the medical core of the game, the specific stimuli, the timing, the order, was not something we could touch. That meant every gameplay change had to be designed around a piece that was, by clinical necessity, immovable. The interesting part of the job was finding the room to make a game inside that.

For texture and feel I kept coming back to retro space games and pinball machines, the same loops of points, ships, small satisfying motions, and a soundtrack that didn’t apologise for itself.

The work

Most of the codebase eventually passed through my hands. Tutorial and onboarding, power-ups, the Ship Shop, the subscription and IAP system, narrative scenes, unlockable content pages, an auth refactor, the content manager, levels, maps, animations, and on the other side of the keyboard, mockups, wireframes, Jira tickets, dev notes for the App Store and Play Store releases. A few threads ran through most of it.

The old tutorial taught the mechanic. The new one taught the point of the mechanic, why this small focus exercise was about to matter to this player in particular. We redesigned it around two screens of warmth and one of explanation, in that order. Warmth first because the science doesn’t work if the player doesn’t feel safe; explanation last because by then they already trusted us. Daily active users rose 150% in the month after launch.

Before the shop, every session ended in a number with nowhere to go. We added a small economy of named ships, Sonic, Flash, Shield, each with a different power-up profile: a magnet that became an aura that became a ship ability, a faster cruise, a wider grab. Earned points could be spent on upgrades, and a small buy-animation rewarded the spend. It gave the game a loop that hadn’t existed before: play, earn, choose, return.

Halfway through the subscription rewrite I learned the company didn’t have one kind of customer. It had two: enterprise users (clinics, research programs) and direct App-Store users, with different access rules, different price logic, and different renewal paths. The system needed to gate, price, and renew for both without mixing them up. I rebuilt around two parallel entitlement paths and the dual offers, promo codes for one, App Store IAP for the other, and migrated the IAP layer to RevenueCat in the same wave. It’s the most useful thing I’ve had to learn so far on this project.

Two more layers ran on top of the core loop. A narrative arc gave the player a sense of accomplishment as they saved small creatures across the map, visual moments that paid off the time they’d put in. Underneath that, a library of short content pages about anxiety management unlocked as the player progressed , small articles in real coping language, written with the clinical team. The first rewards the player for playing. The second turns the play into something they can carry into the rest of the day.

How it landed

StarStarter shipped to the App Store and Google Play and has been clinically validated across seven NIH-funded trials to reduce social anxiety in over 90% of users in our cohort. APA Labs awarded the project a Silver Digital Badge for clinical evidence, user experience, and data privacy.

Downloads at launch were up 30% over the previous version, and we kept that lift. The tutorial redesign tripled daily actives in its first month; the gameplay refactor lifted day-one retention by more than half.

What surprised me most was the small, persistent cohort of power users, people who’d played the whole game through two, three, four times over. Not the average user, but the one who’d already gotten what they came for and kept coming back anyway. That is the part I keep watching.