a small gut check for noticing when ai starts doing too much of your thinking.

what it is

a self-check for knowledge workers who suspect they have started thinking in the cadence of a model. five minutes. no account. no cloud mystery. the app gives you a 0–100 index, a tier, and the important bit: a calibration miss, which is the gap between how sharp you thought you were and what your unaided writing actually showed.

if the index is high and you want to do something about it, atrophy pro unlocks a small set of calibration drills: slop spotting, the verification reflex, confidence calibration, and explain-back prompts. no ai tutor, no streaks, no leaderboard, no forced hype. just short reps you can come back to until your first draft starts feeling like yours again.

what it is not

• not a chatbot. there is no conversational ai inside the app. we do not give you a chatbot. that is the whole point.
• not a diagnosis. the app uses assessment-style language, but the atrophy index is just an app score. we say that on every result screen.
• not a tracker. no analytics sdk. no remote server. no third-party requests. see the privacy policy. it is short on purpose.
• not free forever. the diagnostic is yours to keep. atrophy pro is $4.99/month or $49.99/year. terms.

the idea, briefly

the app is not measuring "ai use." plenty of heavy ai users are calibrated, careful, and still fully capable without it. the thing we care about is the drift between confidence and output quality: that weird moment when polished text starts feeling like proof you still have the skill, even though the checking has quietly gone missing. the longer that loop runs, the harder it gets to write a messy first draft on purpose.

read more

• privacy — what we store, which is very close to nothing.
• terms of use — subscriptions, refunds, and the boring-but-necessary parts.
• support — how to reach the person building this thing.