Healthcare Is Drowning in Paperwork While Doctors Burn Out. A Computer Use AI Agent Can Fix This.
American physicians spend roughly 13 hours every single week on prior authorization requests. Not treating patients. Not doing research. Filling out forms, calling insurance reps, and waiting on hold so a bureaucrat can decide whether a patient deserves the medication their doctor already prescribed. That's nearly a third of a full-time job, gone. Multiply that across the 700,000+ active physicians in the US and you're looking at hundreds of millions of hours vaporized every year by administrative theater. And that's just prior auth. We haven't even talked about the rest of it: the billing codes, the EHR data entry, the insurance eligibility checks, the referral paperwork, the discharge summaries. Healthcare in 2026 is one of the most technologically advanced industries on the planet, and its most expensive professionals spend their days doing work that a well-trained computer use agent could handle before lunch.
The Numbers Are Embarrassing. Like, Actually Embarrassing.
Let's put some real figures on this because vague hand-wringing doesn't capture how bad it actually is. Billing and insurance-related administrative functions alone cost the US healthcare system roughly $400 billion per year. Not outcomes. Not treatments. Not drugs. Paper-pushing. The Commonwealth Fund estimates that administrative costs eat up about 30% of all US health spending. For context, that's more than most countries spend on their entire healthcare systems per capita. A 2025 analysis from BlueBrix Health found that manual data entry processes are costing the system $12 billion specifically in clinician time. Clinician time. The most expensive, hardest-to-replace resource in medicine, burned on copy-pasting data between systems that should have talked to each other a decade ago. And the AMA's own survey data shows that 39 prior authorization requests per physician per week is now the average. Not the outlier. The average. This isn't a niche inefficiency. This is the operating model of American healthcare.
Doctors Are Quitting Because of Software. Not Patients.
Here's the part that should make everyone furious. Physician burnout is at crisis levels, and study after study keeps pointing at the same villain: the EHR. Not difficult cases. Not long hours. The software. A 2025 study published in JAMA Network Open found that ambient AI scribes at one health system saved over 15,000 hours of documentation time and measurably reduced burnout scores. That's one tool, at one system. The AMA's own data shows that EHR tasks follow doctors home. They're doing what the industry grimly calls 'pajama time', charting after their kids go to bed because the system doesn't give them time during actual work hours. A 2025 PMC review of EHR usability challenges confirmed that documentation burden is one of the top drivers of professional dissatisfaction across physicians, nurses, and allied health staff. So when people ask why there's a doctor shortage, part of the answer is that we built a system where the job is 40% data entry and then acted surprised when talented people decided they'd rather do something else.
Physicians complete 39 prior authorization requests per week on average, burning 13 hours of staff time per physician. That's not a workflow problem. That's a $400 billion administrative catastrophe that AI automation could start fixing today.
Why Old-School RPA Is Not the Answer (And Never Was)
The healthcare industry didn't ignore automation. It tried. UiPath, Blue Prism, and a dozen other RPA vendors sold hospitals on the promise of bots that would handle the repetitive stuff. And to be fair, some of it worked, for a while, for specific narrow tasks. But traditional RPA is brittle. It works by following rigid scripts tied to exact screen layouts. The moment an EHR vendor pushes an update, or an insurance portal changes its interface, or a form adds a new field, the bot breaks. And in healthcare, interfaces change constantly. Epic alone pushes updates that can invalidate entire automation workflows overnight. So hospitals end up with IT teams spending half their time babysitting bots instead of building new things. The maintenance cost eats the efficiency gains. Menlo Ventures' 2025 State of AI in Healthcare report noted that prior authorization solutions were 'largely digital forms that still needed manual completion' because the underlying automation couldn't handle the unstructured, variable nature of real clinical data. RPA was a band-aid on a broken leg. It looked like progress. It wasn't.
What AI Computer Use Actually Changes
Here's what's different about modern computer use agents versus the RPA era. A real computer use agent doesn't follow a rigid script. It sees the screen the way a human does, reads what's there, decides what to do, and acts. It can navigate an insurance portal it's never seen before. It can fill out a prior auth form that changed last Tuesday. It can open an EHR, pull the relevant patient data, cross-reference it with payer requirements, and submit the request, without anyone writing a new rule or updating a workflow template. This is the actual unlock for healthcare. The tasks that burn out doctors and staff aren't complex. They're tedious, variable, and endless. Prior auth, eligibility verification, referral processing, claims follow-up, appointment scheduling, discharge documentation. Every single one of these is something a capable computer-using AI can handle end to end. The 2025 LessWrong AI forecasting analysis noted that leading computer use agents were already hitting 65-70% on OSWorld, the standard benchmark for real-world computer tasks, and the trajectory is steep. We're not talking about narrow demos anymore. We're talking about agents that can actually sit down at a desktop and work.
Why Coasty Is the One Worth Paying Attention To
I'm not going to pretend every computer use agent is equal, because they're not. Coasty scores 82% on OSWorld. That's not a marketing number, it's a benchmark result, and it's higher than every other computer use agent on the market right now. Human performance on OSWorld sits around 72%. Let that sink in. For healthcare specifically, that gap matters enormously. A computer use agent handling prior authorization can't afford to fail on 40% of tasks. When the margin for error is patient care, you need the most reliable tool available. Coasty controls real desktops, real browsers, and real terminals, not sandboxed API calls that only work in controlled environments. It can operate inside your existing EHR setup, navigate insurance portals, handle multi-step workflows across different systems, and run as agent swarms for parallel execution when the volume gets heavy. It's not a chatbot bolted onto your workflow. It's an agent that actually does the work. There's a free tier to start with, BYOK support if you want to bring your own model, and a desktop app that doesn't require a six-month IT implementation project. For healthcare organizations that are tired of watching their staff drown in admin work, that's not a small thing.
Here's my actual opinion on this: the healthcare industry has been so busy surviving its administrative burden that it hasn't had bandwidth to fix it. That's the trap. Doctors are too burned out from paperwork to advocate loudly for the tools that would eliminate the paperwork. Administrators are too deep in legacy RPA maintenance contracts to rethink the approach. And patients keep waiting longer for approvals that should take minutes. In 2026, the technology to fix this exists. It's not experimental. It's not a pilot program. A computer use AI agent running at 82% on the world's hardest computer-task benchmark can handle your prior auth queue, your eligibility checks, and your claims follow-up starting this week. The question isn't whether AI automation works in healthcare anymore. The question is why you're still paying humans to do it. Stop waiting for your EHR vendor to build this into a future release that's always 18 months away. Go to coasty.ai, start a free trial, and let an agent handle the stack of forms that's been stealing hours from your best people. Your doctors will thank you. Your staff will thank you. Honestly, your patients will too.