On Device AI hack Project (LMStudio x Microsoft x Qualcomm) @ NYU
Mindful is an AI-powered on-device companion that prioritizes user privacy and wellbeing. It acts as a complementary support tool, working alongside mental health professionals to provide personalized, empathetic assistance.
Mindful leverages on-device AI with the following key principles:
- Personalized Memory: Enhanced empathy through customized user interactions.
- Ethical AI: Built-in safety thresholds for responsible assistance.
- Privacy First: Zero cloud dependency ensures user thoughts stay private.
- Professional Guidance: Assists users in connecting with mental health professionals.
- Local Large Language Model (LLM) integration optimized for NPU.
- Basic conversational interface with strong privacy guarantees.
- Core memory creation from user interactions.
- Safety thresholds for professional referrals.
- Emotion tracking and response system.
- AI-driven personalization with mood detection and proactive queries.
- Evidence-based support with factual citations for coping mechanisms.
- Adaptive conversation engine for dynamic emotional responses.
- Comprehensive analytics:
- AI-powered journaling.
- Emotional journey visualization.
- Progress tracking.
- Smart predictive support system:
- Early warning detection.
- Customizable wellness milestones.
- Multi-modal interactions:
- Voice analysis.
- Guided exercises.
- Personal mental wellness companion for daily reflection.
- Private journaling with empathetic responses.
- Secure space for exploring thoughts and ideas.
- Personalized guidance based on accumulated user context.
- Snapdragon X Elite processor with dedicated NPU optimized for on-device AI processing.
- LLaMA 3.2 3B QNN: Mobile NPU-optimized.
- Local database: Persistent memory storage.
- Flask + React architecture: Responsive and user-friendly UI.
Thank you for your interest in Mindful! If you have any questions or feedback, feel free to reach out.