A Fresh Look at Governance and Technology in Healthcare
By Matt Phillion
Healthcare is facing changing technologies moving at a pace that the old playbook for governance simply isn’t working anymore. But if the old rules for keeping pace with technology no longer work, what needs to rise up to replace those rules?
“I think we got hit in the face with a big bat,” says Fawad Butt, CEO and co-founder of Penguin AI. “All of this is going to be great in the long term, but right now it hurts. We’re seeing the market try to make some sense of the confusion as the things we’re documenting keep changing every five minutes, whether it’s the model we use, then privacy, then internal government, and so on.”
This level of confusion is normal for the adoption cycle of new technology.
“The difference we’re seeing is that on the provider side, the hospitals, home health, academic centers, they are, for some reason, learning to adopt AI and get through that confusion, while the payers are still deliberating,” says Butt. “Recent reports have found that providers are buying into technology at a faster rate, while payers are buying at 25% slower than they have traditionally done.”
While there isn’t any one single driver for this, Butt notes, two factors do stand out: trust and experience.
“There’s a lack of trust across the board, and a lack of experience—and that lack of experience feeds into that lack of trust,” says Butt.
More rapid adoption on the provider side makes a lot of sense taking into account what those professionals are experiencing, he adds.
“It speaks a bit to the frustration that exists within the healthcare community, especially on the provider side. They’re drowning. The doctors are overworked and the nurses are leaving the profession like we’ve never seen before. Pharmacists are already starting down that journey,” says Butt. “There are less than a million active physicians in the country now and there used to be 1.3 million. Burnout is a big factor.”
The “killer apps” that got providers invested, he adds, were for scribing. It solved a specific, limited task that removed an administrative burden.
“There’s still venture capital firms and others pontificating that AI is the future of personalized medicine,” he says. I’m shaking my head. There’s no personalized healthcare coming soon. What the industry needs is clean, organizational data structures. Things like faster prior authorization are achievable.”
Different pressures, different needs
A major reason we’re seeing adoption spiking on one side of the healthcare continuum and not so much the other comes from the pressures and pain points each are under.
“Payers don’t have the same pressures. For them it’s not personal, it’s financial, whereas on the provider side it’s personal and human,” says Butt. “Meanwhile, the payers are also paying the penance now for immense amounts of tech debt. In the past, they’ve solved these problems by throwing people at it, but when you add more people, you add more structure and when you add more structure, you add more governance, so everything has to pass a really significant threshold for anything to get deployed.”
To get to the next level of technology adoption, healthcare needs to rethink risk, Butt explains.
“We’re at this inflection point with technology,” he says. “I think if you use the old risk models, which are very absolutist, you take the worst-case scenario and mitigate for it. Companies have to be better at assumption of risk with new technology, and I don’t think leadership is set up for that right now. I don’t think there are processes set up for that, and I don’t think people have the training or the mindset, so there’s a bit of a freeze.”
It all adds up to a perfect storm, Butt explains.
“The tech is outpacing us, but processes aren’t set up to support fast-paced tech, and the good old days of money pouring in are behind them. There’s capital constraints, there’s bloat, there’s structure changes in terms of regulation and government programs,” says Butt. “If this was 2017 and AI happened in healthcare I predict the processes would have been the problem, but we would have thrown more people at the problem by hiring the right people, the top talent. But the money’ not there and that talent is being acquired by organizations like OpenAI or Meta. Healthcare doesn’t have the capital to compete for the talent.”
The path through this is having a grand vision but executing smaller, says Butt.
“The paralysis comes in because we’re looking at all the problems at once, all the risks at once. We need to do more timed experiments, sparking many little fires instead of one big bonfire,” he says. “Pick a couple of processes where you have a need and the ROI you believe AI can provide, but also where it intersects with the availability of data in that subject area. Where you find a place where you’ve got ROI, data availability, administrative function, and an existing talent pool, those are the experiments you should be running.”
Butt sees 2026 as a shift forward in AI.
“[In] 2024, the cycle was AI is real. 2025 is the hyper-hyping of AI. 2026 is where you’ll see real use cases delivering real value to real companies. That’s where the market pressure is going to kick in,” he says. “You’ll have players in your space doing something better, cheaper, faster, and more intelligently, and that drives more adoption, change, and innovation.”
But at the moment, there haven’t been enough success stories published and pushed out from early adopters. Butt predicts we will see a number of clinics or small practices taking the lead as early adopters because the transition to new technology is a less of a lift than in a massive organization, but on the payer side, he sees the value first being extracted the largest players.
“They have more capital and more capacity to hire and bring in the talent,” he says.
Educating your workforce in the face of a new technology is a steep climb, Butt notes, but will come in time. What organizations need right now, however, is expertise.
“You can have out-of-the-box capabilities that are the Ferrari of AI models. I’ve given that Ferrari to a couple of costumers and they’ve crashed it into a tree or burned out the clutch on the first try, because they don’t yet know what to do with it,” he says. “The model that’s working better is deploying engineers into the environment to drive that Ferrari. You have someone who knows how to use the platform and we can then come together to solve specific problems.”
Preparing for the future
With those use cases right around the corner, Butt notes, it is important for organizations to have more of a plan to deploy technology in a safe environment in a safe use case themselves.
“Tech is evolving quickly and the models will get smarter. I think in 2026 a lot of the financial constraints and paralysis is going to become a bottom-line issue for a lot of organizations and is going to be a basis for action,” he says.
It’s hard to know how the AI cycle will evolve in 2027, but Butt envisions that the year could see a shift back into profitability mode.
“Providers, or at least the ones who have deployed this technology, will start seeing the lessening of bleeding on their cost side and hopefully have invested in anti-burnout capabilities,” he says. “I just see full acceleration: 2028 is going to be a year you see companies will become known the way Epic is today.”
Butt also sees technology opening the door for companies to develop technology and tools for a fraction of the cost previous generations could.
“A billion dollars couldn’t give you a language model four years ago, and now I can rent one for $100,000,” he says. “I think it will democratize the processing of healthcare information. Right now, we’re at the mercy of the very few and their technology prowess to improve our outcomes. In a year or two, these smaller companies will have as much capability, and it’s going to be a game changer.”
Matt Phillion is a freelance writer covering healthcare, cybersecurity, and more. He can be reached at matthew.phillion@gmail.com.