Will AI fix prior authorization—or make it worse?

Near term, this is less a broad healthcare selloff than a re-pricing of the probability that AI can be deployed as a pure cost-cutting weapon inside insurer operations. The market should differentiate between incumbents with existing claims data, compliance teams, and physician-facing workflows versus stand-alone automation vendors that are pitching autonomous decisioning; the latter face longer procurement cycles and a higher burden of proof. For ADLI/MDCE-type exposures, the key risk is not model performance but buyer hesitation after any visible complaint cycle, which can delay revenue conversion by quarters.

The second-order effect is that oversight friction may actually widen the moat for the largest payers and rev-cycle vendors with audit trails, because smaller rivals cannot absorb higher legal/compliance overhead as easily. If automation is constrained to “assistive” use cases, the economic benefit shifts from denial reduction to admin throughput, which is smaller and slower to show up in margins. That argues for modest winners in scaled platforms, but only after evidence that appeal overturn rates and error rates are stable.

Catalyst path: over the next 1-3 months, watch for AMA/CMS commentary, earnings call language on AI-assisted prior auth, and any disclosure of appeal rates or human override volumes. Over 6-18 months, the durable winner is likely the vendor that can prove explainability and defensibility, not the one with the highest automation rate. The contrarian view is that the market may be overestimating the regulatory clampdown; AI won’t disappear, but it will be forced into a narrower, more auditable workflow layer, which is favorable for incumbents and unfavorable for opaque point solutions.