Why we built it
Getting healthcare in the U.S. is confusing. Which doctor should I see? Does my insurance cover it? What will it cost? What should I say when I call?
We built FirstVisit to help people answer these questions using their own situation—without having to upload their health history to a cloud AI service.
What it does
Tell FirstVisit what you need, in your own words. It helps you find care options, compare estimated costs, and prepare questions for a clinic.
You can also add a lab report, review the extracted results, and prepare questions to discuss with your doctor. Reports are optional, and you can save your journey and come back later.
For our demo, we used a real lab report with identifying information removed and the owner's permission.
FirstVisit helps you navigate care. It does not diagnose or prescribe.
How we built it
We used Python, JavaScript, and local models through Ollama. Health records are stored in an encrypted local vault.
External provider searches use location and care type—not your lab results or private health story. Optional browser voice input has a separate disclosure because the browser's speech service may process audio externally.
What was hard
Making it personal without making people fill out a huge form.
We also had to separate information useful for finding a clinic from information that should stay private. A clinic search doesn't need your entire medical history.
What we learned
People don't start with a perfect medical question. They start with “I'm worried,” “I got this report,” or “I don't know where to go.”
The product needs to help them from there.
What's next
Broader starting points, better document support, and testing with people who are new to the U.S. healthcare system.
Built With
- cryptography
- css
- html
- javascript
- ollama
- pypdf
- python
- qwen
