Can we ride the GenAI wave without being overwhelmed by it? – Healthcare Blog
David Shewitz
Lenin said: “There are decades when nothing happens; there are weeks when decades happen.” Maybe never. It’s also a very apt description of how the last year of generative artificial intelligence (genAI) – and especially the last week – has seen the AI landscape change so dramatically that even skeptics are now starting to update their prior knowledge in a more optimistic direction.
In September 2025, Anthropic, the artificial intelligence company behind Claude, released what it called its most powerful model yet and said it could perform complex coding tasks for about 30 hours straight. Reported examples include building web applications from scratch, with some runs described as generating approximately 11,000 lines of code. January 2026, two wall street journal A reporter who claimed to have no programming background used Claude Code to build and publish a Magazine project, describing the feature as a “breakthrough moment for Anthropic coding tools” and “vibe coding” – the idea of creating software simply by describing it.
Around the same time, OpenClaw quickly became popular as an open-source assistant that ran locally and performed multi-step tasks through everyday apps like WhatsApp, Telegram, and Slack. However, the deeper shift is architectural: the ecosystem is moving toward open standards for AI integration. One of the standards, called MCP (“USB-C for AI”), is now downloaded nearly 100 million times a month, indicating that AI integration has moved from exploratory to operational.
The market is watching and reacting accordingly as AI agents evolve into potentially useful economic actors. When Anthropic announced plans to expand into high-revenue verticals including financial services, legal and life sciences Magazine The headline reads: “Threats from new AI tools cost software and data stocks $300B.”
Economist Tyler Cowen observed that this moment will “be a turning point of sorts.” Derek Thompson, a long-time worry about the AI bubble, said his concerns have “diminished significantly” in recent weeks. Taking a cue from Wharton’s Ethan Mollick — “Remember, today’s AI is the worst AI you’ll ever use” — investors and entrepreneurs are scrambling to find opportunities to ride this wave.
Some founders are turning their ambitions toward healthcare and life sciences, and they see a lot of problems for which genAI could be the solution, or at least part of it. One AI-powered startup's approach to primary care offers a glimpse into what such a future might hold (or perhaps what new hell awaits us).
Two visions of primary care
There is a real crisis in primary care. Overburdened and ridiculously underpaid, primary care physicians are fleeing the profession—some turning to concierge clinics where (they say) they can provide the quality care that attracted them to medicine in the first place, and many disengaging from clinical practice altogether. Recruiting new students becomes more difficult every year.
Dr. Lisa Rosenbaum captures all that is lost with extraordinary power. new england journal of medicine A podcast series on this topic.
In a companion article, Rosenbaum documents the measurable consequences when patients lose their primary care physician: increases in mortality, emergency room visits, and hospitalizations, all of which are proportional to the duration of the relationship—suggesting, as she writes, “that the relationship itself provides health benefits.” Even worse, more than three-quarters of patients who lost a relationship with a PCP never formed a new PCP relationship.
But Rosenbaum’s deepest concern isn’t the statistics. This is what she calls the “good doctor” phenotype—not a skill, but a style. She described the characteristic of a doctor who takes responsibility for the patient's entire problem. When Rosenbaum was caring for a hospital patient, the patient insisted that she brief the doctor, explaining simply, “He's going to want to know.” For Rosenbaum, letting your patient intuit that you want to know—far more than any quality indicator—constitutes the essence of being a good doctor. She warned, “A culture without a vision for good doctors is a profession without a soul.”
Her biggest concern: The system could evolve into “some kind of AI-enhanced classification system that lacks a relational core.”
That’s pretty much what Council Health co-founder and physician-entrepreneur Muthu Alagappan aspires to achieve — for the benefit of patients. His starting point: 100 million Americans have nothing to do with their doctors, for better or for worse. The relationship ideals Rosenbaum espoused are already out of reach for most people.
Counsel Health, which was recently backed by a $25 million Series A round from GV and Andreessen Horowitz, uses AI to take care of upfront information gathering and initial clinical reasoning, acting like “a very smart resident, reasoning with them, providing a plan, and allowing them to approve or deny with just one click,” as Alagappan puts it. Doctors see more than 15 to 20 patients per hour. Vision: A primary care visit costs less than a dollar.
As Alagappan believes, “It is difficult to understand the cognitive aspects of primary care medical practice, and technical systems are no better suited to doing so than the human brain.”
He acknowledged that humans may still need to perform some annoying hands-on tasks, such as bandaging ankles or administering vaccines, but otherwise he seemed convinced that the future belongs to machines. He expects “regulation will loosen and improve so that AI can do more and more.”
In Utah, the approach taken by a startup called Doctronic suggests that such regulatory change may be closer than we think. The company's AI can write refill prescriptions for 190 routine drugs without a doctor's involvement, for $4 per prescription, and comes with a malpractice insurance policy that covers the AI system itself, as well as upgrade and oversight safeguards. Expansion to states such as Texas, Arizona and Missouri has been considered, and a nationwide rollout is under consideration.
Who is responsible?
With the rapid compounding of AI capabilities, it is easy to apply them where they are most suitable. Without intention, this approach can quietly redefine the discipline in terms of tasks that technology performs well. Because AI can effectively process symptoms, match regimens, and update prescriptions, we may start to define medicine as these specific tasks—just as we can measure steps, sleep scores, and VO2 max, we are tempted to define health as the optimization of dashboard metrics. As Kate Crawford astutely warns, we must not allow “the affordances of tools to become the horizon of truth.”
This tension also extends to biopharmaceutical R&D. Here, efforts to leverage AI succeed in limited domains with dense data and established benchmarks, but struggle where critical data is scarce, highly conditionalized, or both—as Andreas Bender discusses particularly eloquently.
We can't help but wonder where the light is. Although it can be difficult to focus on what really matters rather than what technology can most easily provide, it can be done.
A company built on what matters
I have been arguing in this space, at KindWellHealth, and elsewhere for some time that truly promoting human flourishing requires attention to three broad dimensions: physiology (exercise, nutrition, recovery, preventive screening), agency (belief in your ability to shape a better future), and connection (the value of meaningful relationships and purposeful pursuits).
The latest news that caught my attention is that someone independently built a business around exactly this framework. Unbound is a UK-based preventive health company operating out of a just-opened location in Shoreditch, London, which describes itself as being “founded on the belief that physical, mental and social health are inextricably linked”.
Several design choices differentiate Unbound from optimizing cultural norms. They measure connectivity alongside biomarkers—in effect, assessing social connectedness as a clinical input. Their medical director, Dr. Elliott Roy-Highley, defines health as “not simply the result of internal cellular mechanics but an emergent attribute of social integration, purpose, and public regulation.” Coffee shops replaced waiting rooms; community circles, running clubs and art exhibitions were not facades of health but structural commitments – the social environment was seen as a meaningful part of the intervention.
Perhaps most distinctive is the post-assessment “Future Self” exercise—an evidence-based positive psychology intervention that asks participants to envision their best future self and identify personal barriers to realizing that vision. By strengthening the mental connection between your present and future selves, this exercise can increase goal clarity, self-efficacy, and motivation for behavioral change. This process occurs through the mechanism of narrative—imagining, evaluating, and orienting toward personally meaningful goals—translating assessment insights into actionable health strategies.
Crucially, Unbound does not reject measurement and technology. They offer a companion app for extending connection and tracking recommendations beyond the clinic; their assessments combine blood tests and physical performance tests with emotional and social components. As Unbound puts it: “Yes, we use tools like clinical tests, but not as a way to measure your worth or push you toward perfection. We use them to guide and support a larger goal: to help you live the life you want with clarity and confidence.” Intent: Use science and technology with purpose, pointing them where they should be aimed, rather than where they are most inclined to go.
Of course, there's a big gap between a compelling concept and improved health. Unbound may prove to be savvy health marketing aimed at proactive, affluent city dwellers. People walking into the trendy Shoreditch wellness studio are already relatively motivated and may have been attracted to purposeful engagement. Evidence that the program actually improves health, while theoretical, remains to be seen.
But the interest Unbound is attracting shows there's a huge appetite for going beyond relentless metric optimization—and there doesn't seem to be anything particularly proprietary about their approach. The same basic principles—deepening connections, growing agency, paying attention (with compassion) to physiology—all can be applied at scale by existing businesses and digital platforms. Peloton, for example, had community infrastructure and user engagement; what it lacked was a framework that went beyond leaderboards and performance dashboards to help users not only perform, but thrive.
bottom line
GenAI is developing at a pace that would have seemed unimaginable just a year ago. Developments over the past few weeks have forced even seasoned skeptics to recalibrate. There is huge motivation and good reason to leverage this wave of technology to capture compelling opportunities like the primary care crisis. But as these capabilities compound, the core challenge will be to ensure that the technology meets the actual needs of patients and people, rather than having those needs defined by what the technology can most easily deliver. As many technology-driven companies in healthcare, biotech, and fitness have demonstrated, the risk of intrinsically degrading fitness to the point where it can be optimized through technology is real. But it’s also possible to use technology to achieve a more complete, less reductive vision—one that focuses on physiology, agency, and genuine human connection—as Unbound suggests and, hopefully, many others are pursuing.
Dr. David Shaywitz is a physician-scientist, a lecturer at Harvard Medical School, an adjunct fellow at the American Enterprise Institute, and the founder of KindWellHealth, an initiative dedicated to advancing health through agency science. This article was previously published on The Timmerman Report
