NeuroTwin

Inspiration

Mental health is deeply personal, yet most AI-powered therapy apps depend on the cloud—uploading emotions, conversations, and vulnerabilities to remote servers. Many users avoid seeking help due to privacy concerns, and internet access is often unavailable during emotional crises. We wanted to build mental health support that is instant, ethical, and completely private.

What It Does

NeuroTwin is an offline AI mental health companion that supports users during anxiety, panic attacks, and emotional distress. It listens through voice or text, recognizes emotional patterns, and responds calmly with grounding techniques, reflective dialogue, and personalized coping strategies—all without using the internet.

How We Built It

NeuroTwin uses Whisper (quantized) for on-device speech-to-text and emotional cue detection. DeepSeek R1 (distilled) powers empathetic reasoning and therapeutic-style responses. The RunAnywhere SDK orchestrates these models efficiently on-device and enables secure, encrypted local memory for long-term emotional pattern learning.

Challenges We Ran Into

Designing a system that feels supportive rather than clinical was challenging. We also had to optimize reasoning models to respond instantly on mobile hardware while maintaining privacy, emotional accuracy, and zero cloud dependency.

Accomplishments That We're Proud Of

Built a fully offline, personalized mental health AI
Enabled real-time emotional support with zero latency
Ensured sensitive emotional data never leaves the device
Demonstrated ethical AI design for mental health use cases

What We Learned

We learned that emotional trust is impossible without privacy. Edge AI is uniquely suited for mental health because it removes latency, surveillance, and data risk—making support available exactly when users need it most.