How Aetna's CTO Built an AI Strategy That Keeps Humans in Control

Healthcare payer Aetna is deploying AI aggressively across its operations, but CTO Nathan Frank has drawn a hard line: AI will not deny care or make clinical decisions on its own. The company's product-first AI philosophy, developed in partnership with parent company CVS Health, is emerging as a useful blueprint for enterprises trying to scale AI without losing governance control.

AI Approves 82% of Prior Authorizations in Real Time

Aetna uses AI to adjudicate electronic prior authorization requests, and the technology issues approvals for roughly 82% of cases immediately. When a request doesn't meet the threshold for AI approval, it passes to a human reviewer, where AI still plays a role -- pulling relevant information together -- but the final decision stays with a person.

"AI is not going to deny care," Frank told TechTarget. "It's not going to make clinical decisions on the behalf of our colleagues, and we've stayed true to that over time."

Key Takeaways

• Product-first governance: Every AI use case must solve a concrete business problem. Aetna explicitly rejects "AI for AI's sake" or competitive optics as reasons to deploy.
• Pod model for buy-in: AI agents are co-designed with the employees who will actually use them, combining subject matter experts with product managers, engineers, and designers in small "pods." This approach has measurably improved adoption.
• AI inside the governance process: Rather than let a backlog of AI approval requests slow innovation, Aetna built AI into its own governance review. Low-risk use cases that match existing patterns get faster automated approval, while novel or high-risk cases still go through full committee review.

Ambient AI for Care Managers

Aetna has also deployed ambient AI for its care managers, similar to how hospitals use the tools. The AI consolidates key member information before virtual visits and generates notes afterward, letting care managers focus on the member interaction rather than documentation.

"Our goal is around matching up AI with humans, and we think we get the best results when we do that," Frank said. "We're trying to use AI to remove the friction."

Read the full article on TechTarget