SHELIA

Inspiration

We realized that in moments of crisis, people need more than a panic button—they need a proactive, intelligent companion that sees danger, speaks comfort, remembers everything, and alerts others. That vision sparked the idea for SHELIA: an AI safety guardian combining real-time detection, blockchain logging, emotional support, and personal memory.

What it does

• Detects distress cues and panic words.

• Starts live video, logs metadata (time, location, emotion), and stores it immutably on Algorand blockchain.

• Uses a Tavus-powered avatar to speak comfort in real time.

• Sends live alerts and location updates to trusted contacts.

• Learns safe zones, logs past incidents, and personalizes support via Supabase.

How we built it

• Bolt.new for frontend & core logic.

• Tavus for avatar emotion and speech.

• Supabase for incident logs, user preferences, and memory.

• Algorand blockchain for tamper-proof metadata storage.

• Netlify to deploy and host our demo.

🎓 What We Learned

• Balancing speed & reliability: We prioritized core MVP features first, then iterated to refine authentication and blockchain integration—mirroring best practices for rapid but stable development

• Tech synergy: Integrating Tavus, Algorand, and Supabase taught us how diverse platforms can converge into a unified user flow.

🧗‍♀️ Challenges We Faced

• Real‑time avatar latency: Syncing Tavus speech and UI under async delays was tricky.

• Blockchain costs: We had to efficiently package metadata to avoid excessive on‑chain fees.

• UX under stress: Designing an intuitive, soothing interface for high‑anxiety situations required careful iteration.

Built With

• agent-tavus
• ai
• algorand
• algorand-sdk
• auth-supabase
• blockchain-algorand
• bolt.new
• database-supabase
• domain-ionos-entri
• hosting-netlify
• javascript
• supabase
• supabase-rest-api
• tavus
• tavus-cvi-api