Walk into any healthcare conference these days and you'll hear the same pitch over and over. AI receptionists that answer your phones. AI scribes that document your visits. AI billing systems that handle your claims. AI for prior authorization. AI nurses. AI doctors.
br Everyone's building AI to do your job or the jobs of the people you work with.
Here's what nobody's saying out loud: they're betting you'll stay a buyer, not a builder. They're betting you'll keep writing checks for software that solves their definition of your problem, not yours. And honestly? That bet makes sense if you look at the last 20 years.
The Real Problem We Should Be Talking About
Doctors currently spend about 45% of their time working in electronic health records, with nearly half of that time on documentation alone PubMed Central. Over 80% of physicians report that the time and effort they spend documenting patient care is inappropriate. But here's the thing. This isn't a labor problem. It's a technical problem that we've been solving with labor.
Your EHR can't do what you need it to do, so you hire more medical assistants. Your scheduling system doesn't integrate with your intake forms, so you hire a front desk coordinator. Your prior authorization process is broken, so you bring on administrative staff to handle the back and forth.
EHRs were developed to support insurance-driven payment models, not primary care delivery PubMed Central. They were built by engineers for billing departments, not by doctors for patient care. So naturally, they don't work the way you work.
And the solution for the past two decades has been the same: throw people at the problem. Add more staff. Work longer hours. Take the charts home. Do your documentation in "pajama time" after your kids go to bed. That's expensive. That's exhausting. And it's exactly why investors are now flooding into healthcare AI
Two Bets on Your Future (Neither Includes You)
Right now, there are two plays happening in healthcare AI:
Play One: Replace Your Staff
The pitch is simple. AI agents will automate away your administrative team. No more salaries. No more benefits. No more management headaches. Just software subscriptions that handle scheduling, billing, documentation, and everything else your team currently does.
Play Two: Replace You
This one's even bolder. Why pay doctors $300,000 a year when AI can diagnose patients, suggest treatments, and manage care for a fraction of the cost? Train the models on enough data and eventually they'll be good enough to handle routine cases. Both plays assume the same thing: you're too expensive, and AI is the cheaper alternative.
Here's what's wrong with both of these approaches.
Staff Reduction Misses the Point
Cutting administrative staff doesn't automatically improve patient care. It just shifts the burden. If your EHR is still terrible and your workflows are still broken, automating them doesn't make them better. It makes them efficiently terrible. And trying to train AI to replace doctors? Unlike physicians, AI cannot draw on common sense or clinical intuition, and current AI systems resemble signal translators rather than reasoning clinicians. That's going to take 10 times longer and cost 10 times more than the alternative.
There's a third option nobody's really pursuing: what if doctors became the builders?
The Case for Digital Workers, Not Replacement Workers
Here's a different approach. Instead of replacing staff or replacing doctors, what if we gave doctors the ability to create digital workers that fill the gaps? Not AI that replaces your medical assistant. Digital tools that help your medical assistant do their job better.
Not AI that replaces your clinical judgment. Custom applications that handle the repetitive parts of your workflow so you can focus on the actual medicine. This isn't about cutting headcount. It's about adding technical capacity to practices that have been intelligence constrained for 20 years because of these three barriers:
- Lack of capital (custom software development costs hundreds of thousands of dollars)
- Lack of technical expertise (you studied medicine, not computer science)
- EHR barriers (your system is locked down and can't integrate with anything UNLESS YOU'RE AN ENGINEER)
When you remove these barriers, something interesting happens. Doctors who understand their workflows intimately can build exactly what they need. A psychiatrist creates an intake form that captures the specific information relevant to their practice. A family doctor builds a patient triage tool that routes questions to the right team member. A pediatrician designs a vaccine scheduler that actually makes sense.
These aren't off-the-shelf solutions that sort of work for everyone. They're custom tools that work exactly right for you.
Why This Approach Wins on Trust
Here's the part that matters most: when you build the tool yourself, you know exactly how it works. Trust remains one of the biggest barriers to AI adoption in healthcare, with doctors expressing concerns about the lack of explainability regarding what features drive algorithm outputs The opacity of algorithms may inhibit clinicians from relying on their outputs in clinical settings due to ambiguity about whether recommendations can be trusted.
But if you're the one who built it? If you can see exactly what it does and adjust it when something doesn't work right? That changes everything. This is what we mean by observability and surveillance of clinical AI. Not surveillance in the creepy sense. Surveillance in the sense that you're watching it work, you understand its limitations, and you can intervene when needed.
More observability leads to more trust. More trust leads to higher adoption. Higher adoption means you actually use the tools instead of letting them gather dust while you go back to your old workflows
What This Looks Like in Practice
Instead of buying an AI scribe that transcribes everything but still requires you to edit the notes for 20 minutes, you build a custom documentation assistant that knows your templates, understands your specific patient population, and outputs notes in exactly the format you need.
Instead of subscribing to a generic patient portal that nobody uses, you create a simple intake system where patients can upload relevant information before their appointment in a way that actually saves you AND them time.
Instead of fighting with your EHR's scheduling system, you build a lightweight tool that handles the specific edge cases your practice deals with (weekend appointments, telehealth slots, follow-up reminders that actually work). These aren't massive enterprise systems. They're small, focused tools that solve real problems you face every single day.
The Bottom Line
Healthcare companies are making a bet. They're betting that doctors will remain buyers of technology, not builders of technology. They're betting you'll keep paying for solutions that kind of work, sort of fit your workflow, and require you to adapt to them rather than them adapting to you.
We think that's the wrong bet.
Doctors have been intelligence constrained for too long. You've had to compensate for bad technology with more labor, longer hours, and increasing frustration. You've been forced to work around systems that were never designed with your actual needs in mind.
What if instead of replacing you or your team, AI helped you become the builder? What if the future of clinical AI isn't about automation that removes humans from the equation, but augmentation that gives clinicians the technical tools they've been missing?
That's the bet Cline is making. Not on replacing doctors. On empowering them. Because the person who best understands your workflow isn't a software engineer in Silicon Valley. It's you. And maybe it's time the tools reflected that.