Independent studio · Google Play
NEW LIGHT designs, ships and tends games and apps for Android. Unity for games, Kotlin for apps, and a stubborn habit of testing on the cheapest phone in the room before the nicest one.
Three titles in rotation. Each one started as a single verb.
Live on Google PlayA hand-painted puzzle adventure about relighting a valley.
Forty-one hand-built rooms, each one a single mechanical idea turned over until it clicks. Our first title to cross a million installs.
Live on Google PlayA one-thumb arcade roguelike that reads a run in 90 seconds.
Built for the commute: one input, no tutorial, and a run that resolves before your stop. Ninety-one percent of sessions are played one-handed.
In soft launchStack weathered stones against a rising tide.
A physics puzzler where the water is the timer. Currently in soft launch across three markets while we tune the difficulty ramp.
Utilities built for the moment the signal drops.
Four things, done properly, for Android.
We work as one team on one title at a time. If you need a hundred people, we are the wrong studio — and we will tell you that in the first call.
We take a single verb and turn it over until it is a game. Prototype in week one, playable vertical slice in six, and a production plan you can actually staff.
Unity for games, Kotlin and Compose for apps. We build for the phone most people actually own, not the one on our desks — mid-range Android is the target, always.
Store listing, testing tracks, staged rollout, policy review. We have shipped enough releases through the Play Console to know where a submission stalls and why.
After launch is the actual job. Weekly content rotation, crash triage, and honest instrumentation — retention and run completion, not vanity installs.
Four stages. Most ideas are supposed to die in the second one.
One idea, one verb, no documents longer than a page. We build the smallest thing that can prove the idea is fun or useful.
A playable build on a real device in the first week, then iterate against people who are not us. Most ideas die here, which is the point.
Content, polish, store assets, closed testing tracks. Weekly builds to the same testers so regressions surface while they are still cheap.
Staged rollout, then the long part: crash triage, content rotation, and reading the retention curve instead of the install count.
We would rather have 2 million installs we can explain than 20 million we cannot.
No energy timers, no rewarded-video gates on progress, no dark patterns in a children's category. It costs us revenue. It also means our one-star reviews are about difficulty curves, which is a problem we know how to fix.
Taking on one new title for 2026
