Talkative

feeling a little Talkative
the wellness dashboard
Memoriessss
just breath

Inspiration

The Spark That Started It All 🌟 It began with a simple observation: mental health support shouldn't be a luxury. As I watched friends struggle to find therapists, face months-long waitlists, and battle the stigma of seeking help, I realized technology could bridge this gap. The explosion of AI capabilities in 2024 presented an unprecedented opportunity to create something meaningful. What if we could build an AI companion that genuinely remembered you? Not just another chatbot that treats every conversation like the first, but a true companion that grows with you, learns your patterns, and provides personalized support when you need it most?

What it does

1. Memory Makes All the Difference The biggest revelation was discovering how transformative a memory system could be. Traditional chatbots feel hollow because they lack continuity. By implementing the Smart Memories System that automatically extracts and categorizes important information from conversations, the AI becomes a true companion rather than a digital stranger. 2. Holistic Wellness Requires Multiple Touchpoints Mental health isn't just about talking—it's breathing exercises when anxiety strikes, tracking mood patterns to identify triggers, and setting goals that provide purpose. The app needed to be a complete wellness ecosystem, not just a chat interface. 3. Privacy is Non-Negotiable Working with sensitive mental health data taught me that privacy isn't a feature—it's a fundamental requirement. Every architectural decision had to prioritize user trust and data security. 4. UI/UX Can Make or Break Adoption A mental health app needs to feel safe, familiar, and calming. The chat interface wasn't just aesthetic—it leveraged existing user mental models to reduce cognitive load during vulnerable moments.

How we built it

Phase 1: Authentication & User Management (Week 1) Phase 2: The Memory System (Week 2) Phase 3: Chat Interface & AI Integration (Week 3) Phase 4: Wellness Features (Week 4)

Challenges we ran into

1. JWT Token Expiration Hell The biggest technical challenge was handling JWT token expiration gracefully. Users would suddenly get logged out mid-conversation, destroying the therapeutic flow. Solution: Implemented a JWTRefreshManager class that proactively refreshes tokens before expiration. 2. Memory Extraction Accuracy Getting the AI to accurately identify and categorize important information was tricky. Too sensitive, and it saves everything. Too restrictive, and it misses crucial details. Found the sweet spot through extensive prompt engineering and importance scoring. 3. Real-time Sync Complexity Ensuring conversations, memories, and wellness data sync properly across devices while maintaining offline capability required careful state management and conflict resolution strategies. 4. Balancing Features with Simplicity Every feature added complexity. The challenge was creating a comprehensive wellness platform that didn't overwhelm users. Solution: Progressive disclosure—start simple, reveal features as users engage.

Accomplishments that we're proud of

Building a mental health app is uniquely challenging because you're dealing with people at their most vulnerable. Every error message, every loading spinner, every interaction has the potential to impact someone in crisis. I spent countless hours on details others might consider trivial:

The calming gradient backgrounds The reassuring error messages ("I'm experiencing some technical difficulties right now. Could you please try again?") The celebration animations when users complete goals The gentle reminders that the AI is here to support, not replace, human connection

What we learned

1. How this whole Full stack thing works As a student with mainly Cybersecurity knowledge, this whole experience is an eye opener. I never knew it took so many things all orchestrated beautifully together to get apps to work