Investing in Epsilon Health: A New Model for Radiology

One of the strange things about healthcare is that even though the technology keeps getting better, the experience hasn’t kept up. 2026 can feel like 1996. Where else are you still filling out forms on a clipboard, faxing records, and waiting days for results that take minutes to process?

That last part is especially true in medical imaging. You can get an X-ray in minutes, but getting someone to interpret it can take days. For anyone with a potentially serious medical issue, those days can feel like an eternity. 

Naturally, the industry has tried to solve the problem with AI. More than 750 AI tools have been built for radiology, and billions of dollars have been invested in the category. Even so, wait times are longer than a decade ago. Something clearly isn’t working. 

That’s why we’re thrilled to announce our investment in Epsilon, which has launched out of stealth with $27.6 million in funding led by AlleyCorp, with participation from Uncork, Renegade Partners, SemperVirens, and Jack Altman. 

Instead of building another piece of software that radiologists have to learn, Epsilon is building something fundamentally different: a new type of radiology practice built from the ground up around AI.

More tools, worse outcomes

The U.S. radiology industry is in a crisis. More radiologists are retiring than entering the field, while demand for imaging keeps rising. Interpretation times have tripled since 2014. And despite the hundreds of AI tools at radiologists' disposal, fewer than one in three are actually using them. 

It’s easy to see why. The average radiologist is reading 50 studies a day, and that’s alongside consultations, protocols, lunch breaks, and anything else. These doctors don’t need another tool added to their workflow. The system is broken, and you can’t fix it by selling more software.

What makes Epsilon different

At its core, Epsilon combines AI, physician radiologists, and the clinical workflow into a single practice designed to make each radiologist dramatically more capable.

Scans are now interpreted 2-3x faster than before, which effectively doubles or triples the capacity of a system that desperately needs it. Each radiologist is now far more productive, without changing the way they work. This is why I like to say Epsilon is building the Iron Man suit for radiologists.

Epsilon's compounding advantage

One of the things that got us most excited is that Epsilon’s AI gets better every time a radiologist uses it. Every correction becomes new clinical data for the models, so instead of using textbook data or a labeled dataset bought off the shelf, the data is based on real doctors correcting real cases in the course of doing their jobs. This creates a powerful flywheel.

Some of the largest imaging providers in the country have looked at what Epsilon built and decided it was easier to become a customer than try to recreate it themselves. That’s a pretty good sign that the advantage is real. 

Their scale is already starting to show. Epsilon has served more than a quarter million patients, is processing roughly 2,500 studies a day, and is on track to interpret 1% of all daily U.S. X-rays in its first year!

The team behind it all

Epsilon’s incredible traction comes down to one thing: the team. Epsilon’s founder and CEO, Rustin Rassoli, is one of the most ambitious founders I’ve met. The man is never satisfied! He grew up in the industry and saw firsthand the problems radiology faces, and genuinely believes that if he doesn’t build this, nobody will.

And the team around him is just as strong. Epsilon’s Chief Medical Officer, Dr. Roi Bittane, previously ran the largest radiology group in the country, and now uses an entire career in radiology to make the product he always wished existed. He still reads images almost every day.

Arjun Karpur leads Epsilon’s ML and AI team, bringing his past experience as a researcher at DeepMind to build the technology capable of serving hundreds of millions of people in short order. 

Their team is a rare combination of people who know the problem deeply and have the ambition and talent to go after it. 

Congratulations to Rustin and the entire Epsilon team!