Waystar turns agentic AI loose on claims, denials, and documentation

Waystar has launched agents that interpret payer responses and automatically resubmit denied healthcare claims, which it calls the industry’s first autonomous resubmission capability. Germany caps how many hospital invoices may be audited each quarter and charges a flat fee when one is reduced.

An American company has built software that argues with insurers on its own. Waystar has launched agents that read a payer’s rejection, apply that payer’s rules, decide what to do next and resubmit the claim unaided.

The scale behind it is the selling point. Waystar processes more than 7.5 billion healthcare payment transactions a year and says it touches roughly 60% of US patients, which is what gives its agents the history to act on.

The economics of the problem are peculiar. Payers eventually pay about 70% of the claims they initially deny, but only after providers spend money resubmitting them, and that delay is the gap being automated.

Every performance number belongs to the vendor. A 75% cut in analysis time comes from early adopters and “may vary by organization,” the 25% reduction in documentation review is expected rather than measured, and “industry’s first” is the company describing its own launch, a conditional payoff in miniature.

The most useful disclosure sits in the risk language. Waystar flags variability in AI claims-matching accuracy across payers, and the possibility that payers change the appeals and recoupment processes the agents were built to navigate.

Europe automates health administration too, just not this part. NHS England handed Copilot to 505,000 staff this year, inside a system that generates almost no denied claims to chase.

Where Europe does run this fight, it is capped by law. Germany limits how many hospital invoices a sickness fund may send for medical audit each quarter, at 5%, 10% or 15%, depending on how many of that hospital’s bills survived the previous review.

It also puts a price on losing. Since December 2024 a hospital pays a flat €400 surcharge when an audit reduces one of its invoices, which is a ceiling on the argument rather than an invitation to keep having it.

The AI rulebook does not reach any of this. The AI Act treats risk assessment and pricing in health insurance as high risk, and says nothing about the machinery surrounding a bill after it is issued.

Europe’s health regulators already know the gap exists. WHO Europe warned in July that nearly two-thirds of countries in the region use AI diagnostics while only 8% have a health AI strategy.

So the contrast is not about who has better software. America is automating an argument at scale, and Europe wrote rules to make the argument shorter.