The Billing Problem No One Wants to Solve

Nanda and I have been best friends since third grade. We grew up in South Jersey building things together: Minecraft mods, PCs, 3D printers. We always knew we wanted to build something of significance together.

In college, we went different directions, but after a year we both wanted to work together again. While still in college we joined fulltime at Refresh, a YC-backed company working on reinforcement learning environments for frontier AI labs. There, we became enamored by how modern AI systems improve every iteration with post-training.

When we came home, we started seeing broken systems as solvable.

Nanda's mom runs a six-physician practice, and my dad runs a solo practice with my mom as his office manager. They've been caring for their communities for decades.

However, they are losing revenue from insurance companies because the depth and complexity of their care is being misrepresented - and they don't even know it.

Coding is its own profession, and doctors focused on patient care don't have time to do it perfectly. As a result, claims get misrepresented, downcoded, and denied. Appeals take hours, so they often do not happen. Each mistake looks small on its own, but over time, it becomes tens of thousands of dollars a year.

At the same time, costs are rising. Small practices are becoming unviable, and my parents may be forced into retiring even though they still have to pay for my little brother's college.

Small practices are cash hogs compared to hospitals, which are cash cows.

Small practices generate less revenue per customer, but the work does not shrink with them. Every claim still needs to be coded, checked, submitted, and followed up on. From an economic standpoint, it makes no sense for a big billing company to serve them, and so they don't.

Regional medical billing companies fill this gap, but since they are small they often rely on manual labor and don't have the capital or technical expertise to automate like the big billing companies.

Healthcare administration infrastructure is killing small practices.

Well, that was the case before AI.

We're building Taiga, and we've been backed by Y Combinator to build an AI-native end-to-end billing company specifically for small independent practices.

We handle everything from the moment a patient's SOAP notes are written to the moment the practice gets paid: coding the claim, verifying it before submission, submitting it, appealing denials.

At Refresh, we worked on improving models for specific tasks through post-training. We're applying that to medical coding by training a system on realistic clinical data so it learns how to read notes, capture diagnoses and comorbidities that are often missed, and produce claims that hold up under review.

We make sure each claim is supported by the documentation before it goes out, and when claims are denied, we handle the appeals.

Doctors should just be doctors - they shouldn't have to worry about anything else.

We are onboarding our first pilot practices. We are working with real clinical data, measuring denial rates and revenue with and without our services, and proving that AI-powered billing can outperform the status quo on accuracy and cost.

If you run a small practice and you're tired of losing revenue to coding errors or want to help these small practices, we want to talk to you.