software-mansion-labs/private-mind

What it solves

Private Mind provides a completely offline, private AI experience on mobile devices. It eliminates the need for cloud-based AI services, removing concerns about data privacy, subscriptions, and internet dependency.

How it works

The application runs AI models locally on your device. While models are initially downloaded from Hugging Face, all subsequent processing—including chat, document retrieval, and speech-to-text—happens on-device. Users can choose from supported models or add their own, and can customize AI behavior using system prompt presets.

Who it’s for

Privacy-conscious users who want advanced AI capabilities like chatting with documents or vision-capable models on their smartphones without sending data to the cloud.

Highlights

• On-device RAG: Chat with PDF, TXT, Markdown, HTML, or CSV files with local embeddings and retrieval.
• Multimodal capabilities: Supports vision-capable models for image analysis and on-device speech input.
• Conversation branching: Ability to fork a chat from any message to explore different paths.
• Hardware benchmarks: Built-in tools to measure model performance, memory usage, and speed on your specific device.
• Fully offline: Works without an internet connection once models are downloaded.