Guillaume Bray
I built Pause.ai because I was losing hours to algorithms, and nothing existing could stop the loop. Daily-total blockers blocked innocent reuse. Friction-based blockers I'd disable instantly. I needed something that detected the actual problem: reopening the same app five times in a minute, scrolling without purpose, the compulsive pattern itself.
“Make every block feel deserved.”
That’s the principle. When the algorithm detects the pattern, you already feel it—you know you’re doomscrolling. The block isn’t punishment; it’s permission to stop. Cooldown expires, counters reset. No residual guilt. The calibration system learns your edge cases and adjusts over time, silently.
How this was built
Kotlin + Jetpack Compose. Foreground service with watchdog auto-restart. Detection via UsageStatsManager polled constantly, piped through a pure risk-scoring pipeline.
Four independent signals (reopen, switch, session, scroll) feed into a weighted combiner. Each is tunable, test-able, disable-able. Modes only adjust thresholds. The overlay is dismissal-resistant by design. All data stays local.
The Minoai family
Pause.ai is part of Minoai: focused tools that solve one problem well, run offline, and collect nothing.
Zeriko
Language learning tools built around active recall and spaced repetition — not passive consumption.
Strivus AI
Fitness coaching that adapts to your schedule and equipment, not a generic plan you abandon in February.
Minko
A quiet note-taking tool focused on getting ideas out of your head and into structure, without the endless feature overhead.