By DAVID SHAYWITZ
“There are many years the place nothing occurs; and there are weeks the place many years occur,” mentioned Lenin, probably never. It’s additionally a remarkably apt characterization of the final 12 months in generative AI (genAI) — the final week particularly — which has seen the AI panorama shift so dramatically that even skeptics are actually updating their priors in a extra bullish course.
In September 2025, Anthropic, the AI firm behind Claude, released what it described as its most succesful mannequin but, and mentioned it might keep on complicated coding duties for about 30 hours repeatedly. Reported examples together with constructing an internet app from scratch, with some runs described as producing roughly 11,000 traces of code. In January 2026, two Wall Road Journal reporters who mentioned that they had no programming background used Claude Code to construct and publish a Journal undertaking, and described the aptitude as “a breakout second for Anthropic’s coding software” and for “vibe coding” — the concept of making software program just by describing it.
Across the identical time, OpenClaw went viral as an open-source assistant that runs domestically and works by way of on a regular basis apps like WhatsApp, Telegram, and Slack to execute multi-step duties. The deeper shift, although, is architectural: the ecosystem is converging on open requirements for AI integration. One such commonplace referred to as MCP — the “USB-C of AI” — is now being downloaded practically 100 million instances a month, suggesting that AI integration has moved from exploratory to operational.
Markets are watching the evolution of AI brokers into probably helpful financial actors and reacting accordingly. When Anthropic introduced plans to maneuver into high-revenue verticals — together with monetary providers, regulation, and life sciences — the Journal headline learn: “Risk of New AI Instruments Wipes $300B Off Software program and Information Shares.”
Economist Tyler Cowen observed that this second will “go down as some type of turning level.” Derek Thompson, lengthy involved about an AI bubble, said his worries “declined considerably” in latest weeks. Heeding Wharton’s Ethan Mollick — “keep in mind, right now’s AI is the worst AI you’ll ever use” — buyers and entrepreneurs are busily trying to find alternatives to trip this wave.
Some founders are taking their ambition to healthcare and life science, the place they see a slew of issues for which (they anticipate) genAI may be the answer, or at the least a part of it. The method one AI-driven startup is taking in the direction of main care provides a glimpse into what such a future would possibly maintain (or maybe what contemporary hell awaits us).
Two Visions of Main Care
There may be real disaster in main care. Absurdly overburdened and comically underpaid, main care physicians have fled the occupation in droves — some to concierge practices the place (they are saying) they’ll present the standard of care that initially attracted them to drugs, many out of medical observe solely. Recruiting new trainees grows more durable every year.
What’s being misplaced is captured with extraordinary energy by Dr. Lisa Rosenbaum in her NEJM podcast series on the subject.
In a companion essay, Rosenbaum paperwork the measurable penalties when sufferers lose a main care doctor: an increase in mortality, emergency room visits, and hospitalizations, all in proportion to the connection’s period — suggesting, as she writes, “that the connection itself conferred well being advantages.” Worse, greater than three quarters of sufferers by no means type a brand new PCP relationship after dropping one.
However Rosenbaum’s deepest concern isn’t statistical. It’s about what she calls the “good physician” phenotype — not a ability set however a method. She describes a doctor whose hallmark was assuming duty for the totality of his sufferers’ issues. When Rosenbaum was caring for one in all his hospitalized sufferers, the affected person insisted she replace the physician, explaining merely: “He’ll wish to know.” For Rosenbaum, having your sufferers intuit that you’d wish to know — way over any high quality metric — constitutes the essence of being physician. A “tradition with no imaginative and prescient of the great physician,” she warns, “is a occupation with no soul.”
Her darkest fear: the system might morph into “some artificial-intelligence-enhanced triage system devoid of a relational core.”
Which is nearly precisely what physician-entrepreneur Muthu Alagappan, co-founder of Counsel Well being, aspires to deliver — for the sake of sufferers. His place to begin: 100 million Individuals don’t have a relationship with a physician, good or in any other case. The relational splendid Rosenbaum celebrates is already inaccessible to huge swaths of the inhabitants.
At Counsel Well being — recently backed by a $25M Sequence A from GV and Andreessen Horowitz — AI handles the upfront info gathering and preliminary medical reasoning, functioning, as Alagappan places it, like “a particularly sensible medical resident that’s reasoning together with them, serving up the plan and permitting them to approve or deny in a single click on.” Medical doctors see 15 to 20-plus sufferers per hour. The imaginative and prescient: main care visits costing lower than a greenback.
As Alagappan sees it “It’s exhausting to fathom a cognitive facet of the observe of medication in main care {that a} expertise system is simply not higher suited to do than the human mind.”
He acknowledges that people should still be essential for pesky, hands-on duties like wrapping an ankle or administering a vaccine, however past these, he appears to consider, the longer term belongs to the machines. He anticipates “regulation will ease and enhance in order that the AI can do an increasing number of.”
In Utah, the method pursued by a startup referred to as Doctronic suggests such regulatory change could also be nearer than we predict. The corporate’s AI prescribes renewals with no doctor within the loop for 190 routine drugs, at $4 per script — with a malpractice insurance coverage coverage protecting the AI system itself, and escalation and oversight safeguards. Enlargement is already contemplated to states like Texas, Arizona and Missouri, with a nationwide roll-out into account as nicely.
Who’s in cost?
As AI capabilities compound quickly, there may be super temptation to use them wherever they match most naturally. With out intentionality, this method dangers quietly redefining disciplines by the duties the expertise performs nicely. As a result of AI can effectively course of signs, match protocols, and renew prescriptions, we would begin to outline drugs as these particular duties — in a lot the identical manner that as a result of we will measure steps, sleep scores, and VO2 max, we’re tempted to outline well being because the optimization of dashboard metrics. As Kate Crawford astutely warned, we should not let the “affordances of the instruments turn into the horizon of fact.”
This stress extends to biopharma R&D as nicely. Right here, efforts to leverage AI have succeeded in restricted domains with dense knowledge and established benchmarks, however have struggled the place the vital knowledge are scarce, extremely conditional, or each — as Andreas Bender, particularly, has eloquently discussed.
We’re all the time tempted to look the place the sunshine is. However tough as it may be to take care of concentrate on what really issues, reasonably than what expertise most readily delivers, it may be finished.
A Firm Constructed on What Issues
For a while now, I’ve argued — on this house, at KindWellHealth, and elsewhere — that genuinely enhancing human flourishing requires consideration to a few broad dimensions: physiology (motion, diet, restoration, preventive screening), company (your perception in your skill to form a greater future), and connection (the worth of significant relationships and purposeful pursuits).
The news that caught my consideration not too long ago was that somebody independently constructed a enterprise round precisely this framework. Unbound, a UK-based preventive well being firm working from a single just-opened location in London’s Shoreditch, describes itself as “constructed on the idea that bodily, psychological and social well being are inseparable.”
A number of design selections distinguish Unbound from the optimization-culture norm. They measure connectedness alongside biomarkers — actually assessing social connection as a medical enter. Their medical director, Dr. Elliott Roy-Highley, frames well being as “not merely the results of inner mobile mechanics, however an emergent property of social integration, objective, and communal regulation.” A espresso store replaces the ready room; neighborhood circles, run golf equipment, and artwork exhibitions aren’t wellness window-dressing however structural commitments – the social surroundings is handled as significant a part of the intervention.
Maybe most distinctive is a post-assessment “future self” train — an evidence-backed constructive psychology intervention that asks individuals to check their optimum future self and establish private limitations to attaining that imaginative and prescient. By strengthening the psychological connection between current and future selves, the train enhances purpose readability, self-efficacy, and motivation for conduct change. This course of works by way of narrative mechanisms — imagining, evaluating, and orienting towards personally significant objectives –that translate evaluation insights into actionable well being methods.
Crucially, Unbound doesn’t reject measurement and expertise. They provide a companion app for extending connection and monitoring suggestions past the clinic; their assessments combine blood work and bodily efficiency testing alongside the emotional and social elements. As Unbound places it: “Sure, we use instruments like medical testing — however not as a solution to measure your value or push you to chase perfection. We use them to information and help a a lot larger purpose: serving to you reside the life you need, with readability and confidence.” The intent: leverage science and expertise with intentionality, pointing them the place they need to be aimed, reasonably than the place they’re most inclined to go.
After all, there’s a big hole between a compelling idea and improved well being. It’s doable Unbound will show to be savvy wellness advertising and marketing geared toward motivated, prosperous urbanites. The individuals who stroll into a classy Shoreditch well being studio are already comparatively motivated and sure already drawn to purposeful engagement. The proof that this system really improves well being, whereas theoretically grounded, stays to be seen.
However the curiosity Unbound has attracted reveals a considerable urge for food for one thing past relentless metric optimization — and there’s little of their method that appears particularly proprietary. The identical foundational ideas — deepen connection, develop company, attend (with compassion) to physiology — all might be utilized at scale by incumbents and digital platforms. Peloton, as an example, has the neighborhood infrastructure and the person engagement; what it lacks is a framework that extends past leaderboards and efficiency dashboards towards one thing which may assist customers not simply carry out however flourish.
Backside Line
GenAI is advancing at a tempo that might have appeared fantastical even a 12 months in the past; the developments of the previous few weeks have pressured even seasoned skeptics to recalibrate. There may be super incentive — and good motive — to trip this expertise wave towards compelling alternatives just like the disaster in main care. However as these capabilities compound, the central problem might be guaranteeing the expertise serves what sufferers and folks really need, reasonably than permitting these must be outlined by what the expertise most readily delivers. The danger of basically decreasing well being to what may be optimized by expertise is actual, as so many tech-powered corporations in healthcare, biotech, and health display. However additionally it is doable to leverage expertise in service of a extra full and fewer reductive imaginative and prescient — attending to physiology, company, and real human connection — as Unbound suggests, and hopefully, many others pursue.
Dr. David Shaywitz, a physician-scientist, is a lecturer at Harvard Medical College, an adjunct fellow on the American Enterprise Institute, and founding father of KindWellHealth, an initiative centered on advancing well being by way of the science of company. This piece was beforehand published on the Timmerman Report
