What Self-Driving Cars Taught This CEO About Safe AI in Healthcare

Most healthcare AI conversations focus on clinical care. Nitesh Shroff, co-founder and CEO of Arintra, is focused on what happens after the doctor leaves the room — the medical coding, prior authorization, and billing infrastructure that determines whether a hospital actually gets paid for the care it provided.

Nitesh came to this problem from an unexpected background. He was an early AI engineer at Zoox, the autonomous vehicle company now owned by Amazon, building perception models that had to detect pedestrians, interpret stop signs, and predict movement in real time.

The safety-first mindset he developed there, where misreading the environment has immediate physical consequences, is exactly what he brought into healthcare AI. At Arintra, every medical code is explainable, every output is tied back to the source documentation, and the system is designed to be more accurate than human coders. "You cannot afford to hallucinate here," he says.

Five years in, Arintra's agentic network reads physician notes after sign-off and converts them into the codes that drive insurance reimbursement. Hospitals that have adopted the platform are now doing something Nitesh didn't originally design for using Arintra's code explanations as evidence in denial appeal letters to payers. It's one of the more unexpected signals that the system is doing real work.

Watch the full episode here

‍