Android app
Hafıza Tutucum
For people who don't want their notes and reminders in the cloud: the AI features run entirely on the phone, without an internet connection.
Purpose and audience
A note and reminder assistant for people who want to take notes, ask questions about them and set reminders without their data leaving the phone. No server, no account, no cloud.
The problem
Most smart note apps push data to the cloud and depend on paid APIs for their AI features.
My role
I handled everything myself, from design to store release.
Key features
- Ask questions about your notes with an on-device language model
- Offline text recognition from images (OCR)
- Voice dictation
- Automatic reminder extraction from text
- Everything works offline; data stays on the phone
Technical choices
The AI features use a model that runs on the phone instead of a cloud API. Data never leaves the device, the app works offline and there is no dependency on, or usage fee for, an external API.
- llama.rn (llama.cpp / GGUF)
- Runs the language model on the phone, offline
- ML Kit
- Offline text recognition (OCR)
- Speech recognition
- Voice dictation
- Jest
- Unit tests
Related projects
Android app
FocusLife
For people who plan their focus time with the Pomodoro method: separate durations for categories such as study, work and sport, plus daily focus statistics.
Android app
Market Listem
For people who keep a shopping list together with others: lists are shared by link and edited together, and spending is recorded and compared.
Android app
Benim Kasam
A personal vault app for keeping assets such as gold and foreign currency, and their movements, in one encrypted place.