Inspiration
I come from a small mining town where many people are older, busy, retired, or waiting a long time for care. I’ve personally been on a healthcare waitlist for years. When you finally get to a doctor, saying “I don’t feel good” is not enough. We built Helcon so people can bring their doctor the full story, not just a vague memory.
What it does
Helcon is a voice health companion you can call when you feel off. It asks simple questions, checks recent health signals, looks at life context, searches similar past episodes, and creates a doctor-ready report. It does not diagnose. It turns scattered signals into a clear timeline.
How we built it
We built a phone-call flow using Twilio, Grok Voice, a Node backend, Moss memory, and a Vercel-hosted dashboard connected through a Cloudflare tunnel. Grok Voice handles the live conversation and triggers backend tools for health data, calendar context, memory search, graph updates, and report generation. The dashboard streams core investigation steps live through SSE.
Challenges we ran into
The hardest part was connecting a real phone call to a live dashboard. The voice call and dashboard demo worked separately, but wiring the real Twilio call to appear live on the dashboard took careful event handling. We also had to keep the product safe by avoiding diagnosis and clearly separating facts, patterns, and possible contributing factors.
Accomplishments that we're proud of
We built more than a chatbot. Helcon can take a real voice call, connect to Grok Voice, trigger backend tools, search memory with Moss, generate a report, and stream core investigation events to the dashboard. We also used parallel tool calls so health signals, calendar context, and memory can be checked together.
What we learned
We learned that healthcare AI should not start by trying to diagnose. It should start by listening, organizing, and helping patients explain what changed. A vital sign without context is just a number. The real value is connecting symptoms, life events, past episodes, and health signals into one story.
What's next for Helcon
Next, we want to connect real wearable and calendar data, improve live-call dashboard syncing, and make the doctor report more useful for real appointments. Long term, Helcon can become a personal health memory layer that helps people show up prepared for care.
Built With
- cloudflare-tunnel
- dotenv
- express.js
- g.711-?-law
- grok
- html
- javascript
- json
- moss
- next.js
- node.js
- pcmu
- puppeteer
- react
- server-sent-events
- tsx
- twilio
- twilio-media-streams
- twiml
- typescript
- v0
- vercel
- websockets
- ws
- xai-grok-voice
Log in or sign up for Devpost to join the conversation.