Inspiration
This project started very personally for me.
As a doctor, I’ve sat with patients who were already scared or sick — then watched them get even more overwhelmed when a medical bill arrived. I’ve seen people choose not to follow up on care because they were unsure what insurance would cover. And I’ve seen families confused, embarrassed, or even in tears over paperwork that should have been simple.
As an engineer, it bothered me that healthcare billing still feels stuck in another era — slow, manual, and difficult to navigate. There was no modern, intelligent assistant that could talk to patients the way a real human would and help them understand their bills without judgment.
So I decided to build one.
What I Built
Tiba is my attempt to blend both sides of who I am. It’s an AI voice billing assistant that can sit between the patient and the healthcare system — calmly explaining bills, checking insurance, and helping people understand what they owe and why.
It can:
- Listen to patients through natural conversation
- Explain insurance terms in simple language
- Track patient details like coverage, deductibles, and prior claims
- Connect to FHIR/EHR systems to pull relevant billing information
- Assist clinics with coding and billing steps
- Reduce the pressure on front-desk staff while giving patients clarity
- I wanted to give people the kind of explanation I would give them myself — but available anytime, without waiting on hold or feeling rushed.
How I Built It
- This project pushed me to merge my medical training with my technical skills:
- I built a voice agent using ElevenLabs that can understand the way real people talk — accents, emotions, and all.
- I connected it to structured function-calling workflows so the agent can remember insurance details and patient information.
- I learned how to integrate FHIR APIs so the agent can access data safely and meaningfully.
- I designed the logic for billing explanations, insurance checks, and coding support, with a human in the loop for accuracy.
- It wasn’t just writing code; it was designing a tool that respects the emotional weight of healthcare bills.
Challenges
- There were plenty of moments where this felt impossible: 2.Teaching the agent to understand medical and insurance language without confusing patients
- Handling voice issues — accents, pauses, anxiety, background noise
- Making sure the assistant stayed safe, accurate, and clinically grounded
- Translating real-world billing rules into something an AI can reason about
- Figuring out how to build trust into a system that deals with people’s money and health
- Every challenge reminded me why this work matters so much.
What I Learned
This project taught me more than just engineering:
- How to build a voice agent that feels natural, patient, and helpful
- How to connect clinical data with billing workflows in a meaningful way
- How to translate complex systems — coding, deductibles, prior auths — into human language And most importantly, how technology can restore dignity to people who are already going through enough
Built With
- anthropic
- azure
- elevenlabs
Log in or sign up for Devpost to join the conversation.