Call2Well

Inspiration

In the United States, 27 million people are uninsured.

In early 2026, federal Medicaid rollbacks left 4 million more people without coverage overnight.

But the problem isn’t that care doesn’t exist.

There are over 14,000 federally qualified health centers (FQHCs) across the country. They are funded to treat patients regardless of ability to pay.

Most people who need them don’t know they exist.

If you’re uninsured and sick, you’re expected to:

know what an FQHC is
search for clinics online
compare services, hours, and costs
and figure it out yourself

All while dealing with pain, stress, or language barriers.

This isn’t a supply problem.

It’s an access problem.

Call2Well was built to close that gap.

What it does

Call2Well lets anyone call a phone number, describe their situation in plain language, and get connected to the nearest free or low-cost clinic in under 90 seconds.

No app.
No forms.
No insurance required.

Using Claude by Anthropic, the system:

identifies urgency
checks eligibility for FQHCs and Medicaid
selects the most relevant clinic
explains the recommendation in the caller’s language

If symptoms sound life-threatening, it immediately tells the caller to call 911.

Twilio handles the call itself, including real-time speech-to-text and text-to-speech.

How we built it

Twilio ConversationRelay for real-time voice input/output over WebSocket
Claude (Anthropic) for multi-turn conversation, multilingual support, and decision-making
HRSA data for verified clinic locations and services
Supabase (PostgreSQL) to store and rank clinics by distance, services, language, and cost
FastAPI to manage orchestration and call state
Next.js dashboard to visualize the pipeline during demos

Challenges we ran into

The main challenge was getting real-time voice interaction and AI reasoning to work reliably together.

We had to:

maintain conversation state across WebSocket messages
keep latency low while coordinating multiple systems
constrain the model to avoid unsafe outputs (e.g., diagnoses, guarantees, missed emergencies)