Waldo The Wellness Warrior

• 

  Sign In Page for Unauthenticated Users

• 

  Chat feature for AI Health Voice Agent

• 

  AI Health Voice Agent in Action when User uses Chat

• 

  Main Dashboard with Past Calls Summary in Light Mode

• 

  Main Dashboard with Past Calls Summary in Dark Mode

• 

  AI Health Voice Agent in Action when User uses Voice

Inspiration

Many hospitals offer video visits but charge high copays. Most of these visits are essentially just audio calls with no physical examination. Meanwhile, hospitals face huge backlogs for simple issues like cuts, coughs, or minor ailments. What if there was a way to handle these simple cases without overburdening healthcare professionals?

What it does

Waldo The Wellness Warrior is an AI health voice assistant that helps patients with everyday health questions. Waldo answers simple questions and guides people to the care they need, reducing the workload for doctors and nurses. This leads to shorter wait times, lower costs, and faster care for everyone. With Waldo, healthcare becomes easier, faster, and more efficient, letting human staff focus on the cases that truly need them.

How we built it

Frontend: Next.js Backend: Python using LiveKit SDK, Deepgram, Cartesia, and Gemini for voice assistant Database: Postgres with Prisma Authentication: Auth.js

Challenges we ran into

We couldn’t host a production link because we didn’t have the budget for Amazon ECS or equivalent services.

Accomplishments that we're proud of

• Successfully training the model to provide accurate responses using RAG (Retrieval-Augmented Generation)

• Getting the voice features working seamlessly with STT (speech-to-text), TTS (text-to-speech), and LLM for conversational flow

  What we learned

• Implementing a full-stack AI voice solution is challenging but feasible with current tools

• Integrating voice AI with real-time audio requires careful design for accuracy and latency

• RAG is powerful for improving AI reliability in specialized domains like healthcare

  What's next for Waldo The Wellness Warrior

• Expand sources for RAG to improve knowledge accuracy

• Connect with healthcare service providers and integrate into their apps

• Feed call summaries directly to healthcare teams to streamline follow-ups

Built With

• authjs
• cartesia
• deepgram
• gemini
• nextjs
• postgresql
• prisma
• python