For thirty years, software promised to fix the paperwork crushing American medicine. It never could, because it could never do the actual work. Artificial Intelligence can, and it is already repaying a debt the country has carried for a generation.
Walk into an American hospital at nine o’clock at night and you’ll find someone still working who has no business being there. Not in the operating room. Not in the ED. At a keyboard. A physician finishing the notes she couldn’t get to between patients. A biller three weeks deep in denied claims. An administrator re-keying the same patient into a fourth system.
None of it is care. It is the cost of transacting care, and in the United States it has become one of the largest, least examined burdens in the entire economy.
The $250 billion work tax
There’s a name for it: the work tax, the toll the American healthcare system pays just to move a patient’s visit from the exam room to a paid claim. U.S. healthcare burns roughly a trillion dollars a year on administration. The revenue cycle alone, the coding, claims, denials, appeals, and collections that turn care delivered into care paid for, accounts for an estimated $250 billion of it every year, more than the federal government spends on most cabinet departments, and almost none of it touches a patient.
Every other American industry already fixed this
The strange part is that the rest of the economy already solved this. Tap a card at a coffee shop and money moves between your bank, a card network, and the merchant in seconds, at an all-in cost of about 2 to 3%. Those are the rails that let a corner store and a multinational run on the same plumbing.
Healthcare has none. The cost to transact a single unit of care in the U.S. has historically run 10 to 15%1, an order many times worse than what Americans take for granted everywhere else. We built the most advanced medicine on earth and made it run on antiquated systems.
It isn’t for lack of trying. For three decades the country was promised software would fix it, and hundreds of billions of dollars later the work tax hasn’t moved. The reason is the whole story: software digitized the paperwork, but it could never do the work. It gave a nurse a better form; it couldn’t make the call to the payer or write the appeal. The labor still landed on a human being, usually after hours, usually exhausted.
AI is now capable of doing the work
Artificial Intelligence and LLMs are the first technology in the industry’s history that can do the work itself. Not assist with it. Do it, end to end, at scale. Modern AI agents sit inside a health system’s existing workflows, read the clinical context, navigate each payer’s rules, code the encounter, submit the claim, catch the denial, and file the appeal. The same systems listen to a visit and produce the documentation before the patient reaches the parking lot.
It’s the bet Commure made almost a decade ago. Its platform now runs inside more than 500 healthcare organizations across 3,000-plus sites and 60-plus EHRs, including over 130 of the nation’s largest health systems, HCA Healthcare and Tenet Healthcare among them. It processes tens of billions of dollars in annual claims and completes more than 85% without a human having to do more than review the output.
The results are already in
The numbers from organizations running on Commure’s AI platform make the abstraction real. An NYC-based nonprofit teaching hospital was leaving nearly $2M on the table every month, until Commure’s Charge Note Reconciliation caught the services being documented but never billed, recovering $1.9M in monthly charges within one month of go-live. Hillside Foot & Ankle, a Pennsylvania podiatry group, grew its business by $3.6M a year. Vitality SarcoHealth, an Arizona neuromuscular practice, lifted patient collections 436% and drove denials to 1.2%, against a 5-to-10% industry norm.
Lattimore Physical Therapy cut denials in half and pushed first-pass claim rates to 98 to 99%, clean enough to scale across 35 locations. OneAccord Physical Therapy improved its reimbursement rate per visit by 115%, driven by real-time billing visibility, dedicated denial recovery, and accurate capture of previously unbilled services. “This is a fifty-year-old problem getting disrupted by some of the best minds I’ve ever encountered,” says founder and CEO Sean Flannagan.
Putting humanity back into care
The work tax is paid in dollars, and extracted from people. When WellNow Urgent Care deployed Ambient AI documentation across more than 600 providers, adoption hit 90% and, within four months, after-hours charting went to zero. Beyond time saved, they also saw a 10% increase in E/M acuity levels compared to the same period the prior year, translating to millions of dollars in revenue recovered for care that was always being delivered, just never documented.
Dr. Neel Palakurthy of Dignity Health now saves up to three hours of charting a day and four more each weekend, time he spends with patients instead of a screen. For some, that time savings makes the difference between continuing to practice medicine or not: “I was getting to the point where I was considering changing my career because I had no life… I’m now incredibly happy,” shared Andrea Hazleton, FNP-C at Pioneers Medical Center.
The effect reaches patients, too. At Yale New Haven Health, automated bilingual outreach for breast cancer cut no-shows and same-day cancellations by 54%, reducing the costs of missed appointments and downstream impact of delays in patient care. AI Agents are expanding patient access with low-cost, 24/7 call center coverage – enabling a multi-state behavioral health network to increase their patient call-to-admission conversion rate by nearly 8 times compared to manual workflows, translating to hundreds of new patients getting care. When the work tax falls, clinicians and teams look up from the keyboard, and the human part of medicine comes back.
Proof at the largest scale: HCA Healthcare
Nowhere is the shift more visible than at HCA Healthcare, the largest health system in the country with over 190 hospitals, 2,500 ambulatory sites, and 320,000 employees in the U.S. and U.K. In its 2025 community impact report, HCA detailed how it deployed Commure’s ambient documentation technology across its network, drafting accurate medical notes directly from the conversation between provider and patient, with the patient’s consent.
The results span every care setting across nearly 50 specialties. In the inpatient environment, 58% of hospitalists now generate notes using Ambient AI and nearly 90% of history-and-physical notes are completed within 24 hours. Adoption in the Emergency Department has reached 80%. In ambulatory care, HCA reports more than an hour of time savings per provider each day, along with higher satisfaction and more time for direct patient care.
The future of American competitiveness
No nation spends more to run its healthcare system than the U.S., yet our health outcomes and patient experience consistently fall short of what hardworking clinicians and taxpaying communities deserve. A quarter-trillion dollars a year sits idle inside the work tax: research that goes unfunded, capacity that goes unbuilt, ground America cedes while the rest of the world races to deploy the same technology. Reclaiming it is a matter of national importance.
It’s also the kind of problem a generation of American builders should run toward. The engineers doing this work aren’t writing another co-pilot; they’re sitting beside clinicians, taking apart a fifty-year-old machine and rebuilding it around the people it was meant to serve. The only question left is how fast America decides to get back into the driving seat and embrace applied Artificial Intelligence.
Commure is doing this work every day, inside the country’s largest health systems and its smallest practices alike. To help build what comes next, visit commure.com.
