Former ARPA-H director launches startup to solve AI's basic challenges in health care

A former director of the U.S. Advanced Research Projects Agency for Health has founded a new company focused on fixing what he calls the "dumb problems" of artificial intelligence in medicine. The venture aims to improve the reliability of AI tools that are currently prone to simple errors.

The startup targets issues such as the accuracy of AI-powered medical scribes, data entry mistakes, and other low‑level failures that can undermine clinical workflows. By building better validation and correction mechanisms, the team hopes to make AI outputs more trustworthy for doctors and patients.

In a broader sense, the effort reflects growing scrutiny of AI applications in health care, where even minor glitches can have outsized consequences. If successful, the company's solutions could help accelerate adoption of AI across the industry while reducing the risk of avoidable mistakes.