R1 Bought the Company Selling Faster Yeses. Physicians Are Not Convinced

What R1 actually announced#

On 18 August 2026, R1 RCM said it would acquire Humata Health, folding its ai prior authorization technology into R1's revenue cycle platform. The deal's own press release, carried via GlobeNewswire, leads with four numbers: up to a 96% first-pass approval rate, an 83% cut in rescheduled appointments, a 30% reduction in write-offs, and a 45% drop in staff touches. R1 CEO Joe Flanagan called it "an important step forward in our journey to automate the revenue cycle."

Read those figures the way a fox reads a scent trail crossing open ground: worth following, not worth trusting blind. Every one of the four is R1 and Humata's own claim. No named customer is cited, no independent trial, no sample size, no stated methodology. The 96% headline carries an "up to" qualifier doing real work, the kind that turns a best-case outlier into a marketing number.

None of that makes the figures dishonest, only unfinished. A vendor claim without a denominator is a den without a way in: it looks solid from the outside and tells you nothing about what is actually sheltered underneath. The gap between a headline percentage and an audited one is exactly where a healthcare marketing team earns its keep, and exactly where most of this category currently fails to.

None of that makes the deal itself a bad one. Financial terms were not disclosed and the acquisition is expected to close by the end of Q3 2026. Humata founder Jeremy Friese, MD, called the underlying problem "urgent" in the same release. He is not wrong about the urgency. The question is whether ai prior authorization automation solves the problem physicians actually have, or a different, more comfortable one.

Why physicians do not trust it yet#

Set R1's claim against the people who live inside prior authorization every week. The 2025 AMA Prior Authorization Physician Survey, a nationwide sample of 1,000 practising physicians, found 95% report care delays caused by prior authorization, and more than one in four, 26%, report it has led to a serious adverse event for a patient in their care. Ninety-four percent say prior authorization somewhat or significantly increases burnout.

The sharpest figure for this specific deal is the 60%: six in ten physicians already say they are concerned that AI is increasing prior authorization denial rates, not reducing them. Only a third, 33%, agree that denials are actually reviewed by a licensed or qualified clinician at all. That is the exact trust gap R1's 96% first-pass figure has to cross, and the AMA survey suggests it has not crossed it yet.

A surgeon replying to the deal's own announcement thread put the physician objection more sharply than any survey figure could.

That line is the whole physician objection in one sentence: a revenue-cycle platform is built to optimise for the payer's clean-claim definition, not the patient's fastest path to care. Automated prior authorization can be genuinely faster and still be optimised for the wrong outcome, and nothing in R1's own release addresses which of the two Humata's engine was actually built to serve.

A second, older AMA release adds texture the 2025 survey alone does not. In an earlier framing piece, the AMA reports that 88% of physicians believe prior authorization increases overall utilisation rather than reducing it, the opposite of the cost-saving story payers use to justify the process. That is worth sitting with: the tool sold to control cost is, in most physicians' own daily experience, quietly inflating it.

Forty prior authorization requests land on the average physician's desk every week, consuming thirteen hours of physician and staff time, per the same AMA data. Automated prior authorization promises to shrink that number. Whether it shrinks the number of requests or simply the time spent handling each one is a genuinely different claim, and R1's release does not distinguish between the two.

The regulatory floor already underneath this#

R1 is not entering an unregulated field. The CMS Interoperability and Prior Authorization Final Rule already requires impacted payers to send decisions within 72 hours for expedited requests and seven calendar days for standard ones, operational since January 2026. From this year, payers must also give a specific reason for every denial, a requirement that sits directly opposite any prior authorization software for healthcare that treats a denial as a black box.

The FDA, meanwhile, draws a line R1's marketing will need to respect. Its Clinical Decision Support Software guidance, finalised January 2026, clarifies which AI-driven clinical functions count as a regulated medical device and which do not. Administrative prior authorization automation, the kind Humata sells, sits outside that specific device framework, which means there is currently no FDA marketing-claim regime built to catch a vendor overstating what ai prior authorization software actually does.

That gap matters commercially, not just ethically. A vendor selling into a hospital system's revenue cycle team can survive scrutiny it never expected from marketing. A vendor selling the same tool to a health plan's medical director will face precisely the FDA-boundary question above, sooner rather than later, once ai prior authorization tools start touching anything closer to a clinical determination than an administrative one.

The pattern sitting before this deal#

R1 and Humata are not the first to promise that automation improves prior authorization outcomes. A 2024 US Senate Permanent Subcommittee on Investigations report, covered by the American Hospital Association, found the opposite pattern at scale inside Medicare Advantage: UnitedHealth's post-acute care denial rate climbed from 8.7% to 22.7% between 2019 and 2022, the years it leaned hardest into predictive tooling. Humana's denial rate for long-term acute-care hospitals rose 54% over roughly the same window.

The Senate report also found CVS tested a predictive model specifically to cut spend on skilled nursing facility services, and its own internal savings estimate rose from $10 to $15 million to a later $77.3 million projection over three years, a trajectory that tracks the model getting better at denying, not at deciding correctly. Separately, a 2023 STAT News investigation reported that UnitedHealth's naviHealth unit set staff targets to keep rehab stays "within 1% of the days projected by the algorithm," with discipline for missing that target regardless of medical justification.

None of that is proof Humata's specific tool will repeat the pattern. It is proof that "can prior authorization be automated" is the wrong question to stop at. The right one is what the automation is tuned to optimise for, and R1's own release, focused on revenue cycle metrics rather than patient outcomes, does not yet answer it.

Every vendor in this pattern was, at launch, confident its own model was different. The lesson from the Senate report is not that predictive tooling is inherently unsafe. It is that a model trained against a cost target will find cost savings even where the clinical picture does not support them, quietly and at scale, unless something outside the model itself is watching for the drift. Automation does not remove that risk. It moves it downstream, into a system fewer people are positioned to question.

How to market ai prior authorization honestly#

R1 is not alone in this pattern of unaudited, headline-first vendor claims. Competitor Cohere Health markets its own prior authorization software for healthcare with similarly unaudited figures: a claimed 90% automation rate, 94% provider satisfaction, and more than 12 million requests handled annually across over 660,000 providers, per its own October 2025 press release. The pattern is industry-wide, not one company's overreach, which makes it a genuine opportunity for whichever vendor breaks from it first.

The $35 billion administrative cost the Health Affairs Scholar study puts on prior authorization is real, and automated prior authorization genuinely can shrink it. The honest version of that pitch names its own limits rather than hiding behind an "up to" qualifier, because the physicians reading it have already read the AMA survey, and increasingly, they have read the Senate report too.

There is a genuine first-mover reward waiting here, the kind that rewards patience over a fast prowl toward the same overclaimed number every competitor already uses. The first ai prior authorization vendor to publish an independently audited denial rate alongside its approval-speed claim will not just win a headline. It will win the one credibility fight this entire category has so far avoided having in public.

If your team is building the marketing case for ai prior authorization or any other clinical or administrative AI claim, that evidence discipline is what folkfox's healthcare marketing work is built around, alongside our wider content marketing services for regulated categories.

Frequently asked questions#