A GP I met with recently was shown, for the first time, a dashboard mapping care gaps across his patient population: patients overdue for chronic medication reviews, diabetic patients he had not seen in more than six months and benefits available to patients that had never been utilised.
He studied it quietly before saying: “You’ve just made visible what I’ve been sensing all along. I knew the care gaps were there, I didn’t have the tools to identify them or act on them.”
As generative AI evolves into agentic AI, it is increasingly positioned to address exactly this challenge.
Healthcare is entering a new era in which intelligent systems can automate entire practice workflows, synthesise patient data for diagnostics and streamline administrative tasks such as scheduling and insurance claims. Yet amid this technological transformation, doctors remain indispensable, not as passive participants but as care orchestrators.
Their role is to harmonise agentic AI systems with their clinical judgment, ensuring that efficiency gains translate into safe, ethical and patient‑centred outcomes.
In this evolved paradigm, doctors are the conductors of a complex digital ensemble, guiding technology to serve humanity rather than replace it.
The care orchestrator: A role doctors already know
Becoming increasingly prominent in global healthcare circles, the concept of the doctor as care orchestrator describes a clinician who directs intelligent, automated systems to co-ordinate and personalise care at scale, while retaining ultimate clinical authority and accountability for patient outcomes.
It sounds new. It is not. Any doctor who has led ward rounds instinctively understands care orchestration. You direct a team of medical students, registrars and fellows, each with varying levels of experience and reliability, towards the best outcome for a complex patient caseload. You do not treat a third-year student's clinical notes with the same confidence as a senior registrar's assessment. You calibrate trust, supervise accordingly and synthesise contributions into a coherent treatment plan.
Managing a suite of AI clinical agents works in the same way.
What shifts is not the physician's authority. It is where their highest value lies, away from personally executing every task, towards the judgment of knowing when and how to act on AI outputs, while never losing the human touch: the family conversation, the ethical call, the clinical intuition that no algorithm can replicate.
I do not doubt that better co-ordinated care delivers better outcomes, and the gains are significant:
- Increasing productivity through greater care team capacity, reduced human error and less operational waste.
- Improving patient experience by connecting physical and virtual care into a seamless, safe and personalised model.
- Driving sustainable revenue growth for the practice by keeping patients engaged, optimising reimbursable activity and proactively addressing care gaps.
The practical result, when done well, is not a marginal efficiency gain. It is a structural shift in what a medical practice can achieve with agentic AI.
What this looks like in practice
A GP instructs their agentic AI system: "Find my diabetic patients who haven't had a consultation in the last six months and reach out to them to schedule an appointment." As a collaborative partner, the AI agent interrogates the patient database, cross-references medical aid scheme requirements, checks the practice calendar and sends personalised messages under the doctor's close watch. It reports back, closes the loop and repeats monthly without being asked again.
This is not pie in the sky. The technology exists. What has been missing is the integration layer that makes it usable within a South African practice context, connecting medical aid scheme data, patient records, clinical workflows and communication channels into a single orchestrated system that the doctor can direct by exception rather than by exhaustive manual effort.
That integration layer, however, requires a foundation that many South African practices have yet to adopt. AI can only deliver real value when information is available, connected and usable. Where workflows remain manual and data sits in disconnected silos, the opportunity is constrained before it begins. Moving away from fragmented systems is not just a technology upgrade; it is the prerequisite for everything that follows.
What’s at stake
For decades, physicians have been unable to optimise patient care, not because they lacked the knowledge, but because of the operational realities of moving patients through each day.
When AI absorbs the administrative weight, documentation, care gap monitoring, patient outreach and funder co-ordination, it does not simply reduce workload. It reveals the next frontier: the white space that doctors always knew existed but never had the bandwidth to reach. Complex care co-ordination. Prevention-oriented medicine. The kind of continuous patient engagement that genuinely shifts health outcomes over time.
That is also precisely what the NHI model, and the schemes adapting to and introducing value-based care models ahead of it, will require and reward.
South Africa is not short of healthcare ambition. What it lacks is delivery capacity. The limiting factor is whether our clinical workforce has the tools, operational infrastructure and support required to personalise care at scale, without burning out in the process.
But the opportunity comes with responsibility. AI must be accurate, unbiased, secure and properly governed. Poorly designed systems can reinforce existing inequities or subtly skew clinical decision-making in ways that are difficult to detect. Weak data governance undermines patient privacy and erodes the trust on which any healthcare relationship depends.
For South African healthcare organisations operating across fragmented systems and populations with complex needs, these risks are not theoretical; they are practical design considerations that must be built in from the start, not bolted on later. AI is not a silver bullet. It is only as useful as the system in which it operates.
The doctors who will lead over the next decade will not be those who resist this shift, but those who embrace it: learning to orchestrate and direct AI clinical agents with the same calibrated judgment they apply to multidisciplinary care teams, while reclaiming the time and cognitive capacity to focus on what matters most.
The challenge ahead is adoption: for doctors to partner with health tech companies to build the capabilities, workflows and governance required to sustain and scale its impact.
Those who do will be best positioned to translate technological progress into meaningful and lasting clinical outcomes.
The time for doctors to lead as care orchestrators is now.
Share
Editorial contacts