Mind Screen
Mind Screen is a voice-guided self-screening app. Talk to an AI assistant, complete a standardized mental health questionnaire, and get a clear, private result in minutes.
Most people never complete a mental health questionnaire: forms feel cold and clinical. Mind Screen removes the form. You simply talk.
An AI assistant reads each question aloud, listens to your answer in your own words, and returns a score with a plain-language explanation in a few minutes. The app covers eleven standardized questionnaires, including the PHQ-9 for depression and the GAD-7 for anxiety. Four are free.
The wording of each question is never rephrased by the AI, and scoring follows fixed rules. Your history stays on your device. Mind Screen is a self-awareness aid, not a diagnostic tool. Available on Android for adults 18 and over.
How it is built: Flutter and Dart. Each questionnaire is a script shipped with the app; the app moves from one question to the next and computes the score with fixed rules, while a language model only maps the spoken answer to one of the standard options. Speech recognition uses the device's native engine, with a Whisper fallback, and a single text-to-speech voice is kept for the whole session. AI calls go through a Supabase Edge Function that verifies the user's Firebase token, so no provider key ships in the app. Subscriptions run on RevenueCat, history is stored locally in SQLite, and PDF reports are generated on the device.
An AI assistant reads each question aloud, listens to your answer in your own words, and returns a score with a plain-language explanation in a few minutes. The app covers eleven standardized questionnaires, including the PHQ-9 for depression and the GAD-7 for anxiety. Four are free.
The wording of each question is never rephrased by the AI, and scoring follows fixed rules. Your history stays on your device. Mind Screen is a self-awareness aid, not a diagnostic tool. Available on Android for adults 18 and over.
How it is built: Flutter and Dart. Each questionnaire is a script shipped with the app; the app moves from one question to the next and computes the score with fixed rules, while a language model only maps the spoken answer to one of the standard options. Speech recognition uses the device's native engine, with a Whisper fallback, and a single text-to-speech voice is kept for the whole session. AI calls go through a Supabase Edge Function that verifies the user's Firebase token, so no provider key ships in the app. Subscriptions run on RevenueCat, history is stored locally in SQLite, and PDF reports are generated on the device.
