Claude for Healthcare: Automating Prior Authorizations, But Can It Replace Doctors?

Anthropic claims its Claude for Healthcare can cut clinicians’ paperwork time in half—but can an AI agent really outperform a doctor at administrative tasks? The company’s CMS connector tool, designed to automate prior authorization reviews, is now being tested by a mid-sized health insurance company.

Early trials focus on comparing the AI’s speed against manual reviews while monitoring risks like medical coding hallucinations.

Anthropic announced Claude for Healthcare, targeting providers, payers, and patients with tools like 'connectors' to access CMS, ICD-10, and PubMed databases.

The system aims to streamline tasks such as prior authorizations, which currently consume significant clinician time. 'Clinicians often report spending more time on documentation and paperwork than actually seeing patients,' said Anthropic CPO Mike Krieger.

OpenAI’s ChatGPT Health has 230 million weekly users discussing health topics, while Anthropic emphasizes automation of administrative workflows. Both companies caution against replacing medical professionals, though 230 million users already rely on large language models (LLMs) for health advice.

The mid-sized insurer’s test case will measure whether Claude’s automation can reduce review times without introducing errors in critical coding decisions.