Shemsa

In-app purchases
Content rating
Everyone
5K+
Downloads
Content rating
Everyone
Learn more
Screenshot image
Screenshot image
Screenshot image
Screenshot image
Screenshot image
Screenshot image
Screenshot image
Screenshot image
Screenshot image
Screenshot image
Screenshot image

About this app

Shemsa is a local-first AI assistant with offline chat, translation, memory, and smart home tools for everyday use.

What you can do:
- Chat with a local on-device AI model
- Translate text between supported languages
- Translate speech from supported audio input
- Extract and translate text from images with OCR
- Save reminders, birthdays, anniversaries, notes, and important personal information
- Use smart home controls for supported local Shelly Wi-Fi devices
- Control supported smart home devices by touch from the Shemsa widget
- Use Hey Shemsa voice commands from the widget for supported local actions
- Keep chat, translation, memory, learning, coding, and smart home tools in one clean interface

Why Shemsa:
- Local-first design focused on privacy
- Offline AI chat after model setup
- Offline translation for text, voice, and images where supported
- Local memory features for reminders and important dates
- Local smart home control without Shemsa cloud control
- Home-screen widget for quick smart home and Hey Shemsa access
- Simple switching between Chat, Translator, and Memory tools
- Optional support purchases help development but do not lock core features

Important:
- Runtime/model downloads are required during first setup
- Feature availability depends on device RAM, Android compatibility, permissions, and supported hardware
- Performance depends on RAM, CPU/GPU, available storage, and model size
- Some internet-based tools, such as weather, news, Wikipedia/knowledge, and model downloads, require network access
- Smart home features require supported local devices and correct local network setup
- On-device AI responses may be incomplete or incorrect
- Memory items such as reminders, birthdays, and anniversaries may not always be detected perfectly
- OCR translation depends on image quality, lighting, font, and layout
- Voice input and wake-word detection may vary by microphone quality, noise, accent, and device performance
- Voice output may sound unnatural for very short phrases in some languages
Updated on
Jul 16, 2026

Data safety

Safety starts with understanding how developers collect and share your data. Data privacy and security practices may vary based on your use, region, and age. The developer provided this information and may update it over time.
This app may share these data types with third parties
Location
This app may collect these data types
Location
Data is encrypted in transit
You can request that data be deleted

What’s new

Improved Piper voice handling with automatic cleanup of broken or partial voice files.
Added a retry prompt when a voice package fails, so users can download it again easily.
Improved the no-internet protection popup in Translator mode.
Voice downloads now stay on-demand and do not start automatically unless you press download.
General stability improvements and bug fixes.