As we enter 2026, healthcare is preparing for another major transition. This time, it is not about digitising the old system. It is about rebuilding the infrastructure entirely.
Last night I pieced together a likely roadmap for how companies like OpenAI and Anthropic may reshape healthcare. Whether this ends up reading as a prediction or instead serves as a system map for those building in this space, a system-level view is the clearest lens for what is now beginning to unfold.
When you connect the incentives, infrastructure, and recent moves, a clearer model starts to appear. I’ve been using “Healthcare 2.0” to frame this shift: a decentralised system where the patient becomes the centre of gravity. In that model, the patient is the operating system.
Early Signals and Predictions on OpenAI I Flagged Before They Went Mainstream
In May 2025 - seven months before any official announcements - I wrote on LinkedIn that OpenAI was going after healthcare providers in this post. Many pushed back, because it went against the mainstream consensus at the time.
Then, in October 2025, in the first edition of this HAIF newsletter, I wrote in “Healthcare AI’s Next Chapter Has Begun”:
“With OpenAI’s future hospital partnerships and expanding model capabilities, OpenAI may eventually connect both sides of care - patients and providers - through a single model infrastructure, positioning itself as a potential gateway to care. It will focus on becoming the middleware where data, reasoning, access, and infrastructure converge.”
Slowly, this is what is becoming evident as both OpenAI and Anthropic have announced going directly to healthcare organisations and patients simultaneously.
Healthcare needs restructuring, and decentralisation is emerging as the new paradigm. With that in mind, I started mapping what Healthcare 2.0 would look like - and what OpenAI and Anthropic are likely building toward.
Mapping the System
Last night, I found myself trying to visualise the next healthcare system end-to-end.
Where does the patient actually sit?
How do hospitals, GPs, pharmacies, labs, and life sciences connect?
What does healthcare look like when you rebuild it for outcomes - not legacy workflows?
How do LLM providers and companies enable such a system to exist?
I grabbed my iPad and Apple Pencil. No AI. No research papers. No published frameworks. Just me trying to make sense of the chaos by drawing it out.
What emerged was a system map of Healthcare 2.0, sketched in real time as the pieces finally clicked into place.
The Architecture of Healthcare 2.0
In this new ecosystem, healthcare is no longer a place you go to. It becomes a network that forms around you - connected by an infrastructure layer designed to decentralise care.
1. The “Ground Truth” Problem
For decades, the ground truth of your health data lived inside the hospital EHR. It was siloed, static, and owned by providers.
In Healthcare 2.0, the ground truth moves to the patient.
The Personal Health Record (PHR) becomes the master record - continuously fed by diagnostics, labs, and wearables.
But this shift introduces a critical risk.
If incorrect, fragmented, or poorly contextualised data enters the PHR, the ground truth collapses. A messy dataset interpreted incorrectly by a patient- or misreasoned by an LLM - turns “ground truth” into hallucination.
The patient-facing LLM becomes more than an interface. It becomes a companion and care navigator.
Grounded in the PHR and enriched by daily-life signals - wearables, behaviours, symptoms, psychological context - it helps the patient reason about their health over time.
It doesn’t only act as a 24/7 care companion, but also as a personal healthcare project manager:
coordinating care,
triaging concerns,
invoking specialised medical LLMs where appropriate,
making bounded clinical decisions (within protocol and risk thresholds),
autonomously requesting labs or prescriptions within defined boundaries
and escalating to clinicians when thresholds are met.
2. The Human Clinician-in-the-loop
In Healthcare 2.0, the patient-facing LLM becomes the routing layer. It monitors context, uncertainty, and risk, and then:
routes the patient to the right level of care (GP, specialist, urgent care, ED)
pre-charts the relevant history and signals from the PHR for rapid review
creates a clear audit trail of what was asked, what was recommended, and why
Clinicians operate as an on-demand oversight and decision layer. Their value concentrates where stakes are high: ambiguity, trade-offs, and accountable decisions.
In this model, “clinician involvement” is not only a safety design. It is a designed network effect - enabled by the infrastructure that connects patients, providers, and workflows in real time.
3. Hospitals and the New Core Infrastructure
EHRs are highly complex, and that complexity creates friction for clinicians and digital health vendors - most notably through vendor lock-in and restrictive approaches to data access and sharing. Recent legal disputes highlight how sensitive and high-stakes this ecosystem has become.
Healthcare 2.0 requires a new infrastructure layer on top of the EHR, not another bolt-on tool.
Hospitals need:
control over which AI tools run in their environment,
visibility into performance and failure modes,
the ability to orchestrate technology and workflows across multiple systems,
and robust AI governance by design.
Build and deploy their own AI tools in-house using local data
This infrastructure becomes the intelligent orchestrator - the plumbing layer that connects tools, data, and oversight while giving hospitals back agency. But it also introduces a new dependency risk: as the orchestration layer becomes core infrastructure, it can recreate lock-in dynamics in a different form.
4. Interoperability Rebuilt Around the Patient
Interoperability has failed because it was built around institutions. In Healthcare 2.0, interoperability works because the patient is the common denominator.
Wherever the patient goes, the same longitudinal view follows them. Providers connect to the patient - not the other way around. Hospitals then interconnect through shared patient-centred standards. I went through this just over a year ago with the founder and CEO of the world’s largest personal health record platform - looking at where PHRs are heading and where to introduce AI to make them clinically useful in real care.
Healthcare 2.0 forces a shift toward value-based care and new economic primitives:
Subscription care (capitation): patients or insurers pay for outcomes, not only encounters.
RWE inside the value exchange: the transaction isn’t just “care for payment.” It becomes care + evidence - where payers, providers, and life sciences contract around measurable real-world performance. And patients retain autonomy over whether, how, and with whom their data is shared.
Who Wins in a Decentralised Model?
Healthcare organisations shift from being the system-of-record to becoming the system-of-control: accountable clinical oversight, workflow orchestration across tools, governance by design, and high-acuity execution when stakes are highest. In parallel, interoperability strengthens to enable true continuity of care across settings.
Life sciences gain access to high-fidelity RWD, PROMs, and precision trial recruitment, accelerating development and lowering R&D costs. By connecting to patient ground truth, pharma can accelerate both D2C engagement and D2C therapeutic pathways.
Patients become empowered operators of their health, supported by an AI companion that works continuously, not episodically.
Patient-Centred System Design
Healthcare 2.0 won’t come in an instant. It’s a transition that will take months and years.
If you place the patient at the centre and rebuild the system around them, you can see why OpenAI and Anthropic are moving the way they are: they are not choosing between providers, patients, or life sciences - they’re moving into all three at once, positioning themselves as the infrastructure layer that connects them. And for anyone building or deploying AI & technology in healthcare, keeping a clear picture of where the healthcare system is heading is how you avoid optimising for a system that’s already changing.
For legacy healthcare systems and EHRs, the implication is straightforward: the centre of gravity is shifting. EHRs may remain the system of record for some time, but the modern stack will increasingly be built around orchestration, interoperability, and patient-centred workflows beyond the EHR.
There’s more nuance - and more risk - than I can fully unpack in a single article. But the direction is clear: whoever shapes the infrastructure layer of Healthcare 2.0 will shape the next decade of healthcare.
Hi - I’m Dr. Youssef Aboufandi, a physician and digital health consultant. I support healthcare and technology organisations with AI adoption and digital transformation strategy, with a focus on market intelligence and design thinking. If you found this article useful, consider subscribing to the newsletter. If you’d like to work together, message me on LinkedIn.
Subscribe to this newsletter for weekly market intelligence on AI in healthcare.
This is one of the ad-hoc Foresight pieces I publish between editions.
I’ll see you next Tuesday with the weekly newsletter - covering what’s shaping AI in healthcare.
- Dr. Aboufandi
📩 If something here stood out, hit reply - I read every message.