Patients aren’t just experimenting with AI for health advice anymore – they’re relying on it.
A recent survey found that more than 57% of participants had a predominantly favorable view of AI in healthcare. The result? Around the globe, more than 40 million people turn to ChatGPT daily for health information.
The Centers for Medicare and Medicaid Services (CMS) has put a thumb on the scale as well, prioritizing conversational AI as a critical use case for early adopters of their new interoperability framework.
OpenAI recently made a major move in this space with the launch of ChatGPT Health, which allows consumers to upload their health records and get advice. Anthropic then had its own announcement, unveiling Claude for Healthcare for providers, payers, and consumers.
These and other moves represent a tremendous turning point for the healthcare industry, which for the past decade has repeatedly tried to engage consumers with their health data – with varying levels of success. But as exciting as these breakthroughs seem, they still don’t solve a critical problem:
What if advice isn’t enough?
Until AI can reach through the screen to perform a physical exam or prescribe a medication, the default advice will remain “go see your doctor.” That sends the patient back onto the same exhausting treadmill of hunting for appointments, navigating provider directories, and sitting on hold.
And when the patient finally arrives at the precious visit, informed by a source not vetted by the physician, more friction ensues. The patient spends time rehashing what AI said before the provider can begin diagnosis and treatment.
Beyond Navigation: AI Triage?
Looking to stop that treadmill, one evolution we might see is embedded care navigation. This may come in the form of…
- AI agents navigating analog front doors and taking on some of the the legwork of finding a provider for patients by perusing provider directories and calling front desks.
- Organizations with care capacity proactively seeking placement where patients are already trying to find health advice.
Neither of these reflect current reality, but they’re not hard to imagine with hints of ChatGPT ads and sponsored content being rolled out in 2026. Combine this with a robust existing industry on patient care navigation, and you don’t have to look too hard to see this in the future.
But there’s even more potential in the form of provider organizations bringing this AI triage in-house. Rather than navigating after-hours recorded messages or making the drive to an urgent care outlet with an unknown wait time, patients and caregivers could be greeted by a friendly medical AI.
Once historical data and symptoms have been gathered, the AI triage nurse could provide at-home advice tailored to the patient’s health literacy level (sometimes an Ibuprofen and a good night’s sleep really does the trick), guide the patient to an emergency setting, or help schedule a follow-up appointment with the practice.
The key here is the seamless transition to the provider. Since the service would be offered by the practice, physicians could review transcripts, make follow-up plans, and perhaps even provide guardrails for how patients are supported. The experience helps maintain the patient relationship and ensures the appropriate urgency for treatment.
Some organizations are already experimenting with this type of model, combining consumer-facing medical AI with a direct path to telemedicine appointments. It’s a good bet that this paradigm will expand to a broader array of care settings, helping to provide a more consistent patient experience.
After all, a patient receives the same level of service from an AI triage nurse, even if they’re reaching out in the middle of the night or during a short-staffed shift.
For both new or existing patients, AI triage has the potential to make lives easier and help alleviate the barrage of inbound calls providers face – all while maintaining the sanctity and continuity of the patient-provider relationship.
How Does AI Fit In?
As with any new innovation, the healthcare industry will be forced to grapple with the boundaries of how current governance, infrastructure, and reimbursement paradigms handle these novel patterns, confronting questions such as:
- How much health advice is appropriate for the AI to offer?
- Does that line change if the chatbot is offered by the provider group – as opposed to independently found by a patient?
- How would this technology be viewed in the interoperability ecosystem, particularly when acquiring external data is involved?
This last question in particular is a near and dear one. We in the interoperability world have spent the last several years fiercely debating what constitutes treatment. An AI agent playing this “triage” role doesn’t neatly fall into individual access (IAS) or today’s definition of treatment under TEFCA.
That’s why it’s important that we prepare for this future and even encourage innovation. When considering the use case, we should look to who is implementing the technology. In cases where providers are making the calls, an evolved definition of treatment should apply.
Let’s Get Ready for AI Triage
AI triage has the potential to transform healthcare operations by enhancing efficiency, improving outcomes, and supporting professionals in high-pressure environments. All of which leads to better patient care.
And while we’re still very much in the early days of this type of model, the advent of healthcare-specific AI models, CMS’ call for conversational AI use cases, and aggressive investment in clinician-facing AI suggests it won’t be long before it becomes more prevalent.
But regulations and restrictions on patient data – including keeping an AI triage workflow ambiguous in the realm of treatment versus IAS – remain a very real challenge to this vision.
It will take cooperation among the industry’s leading innovators to tackle these challenges, to promote the game-changing possibilities of the technology, and to help providers move to where AI is heading – rather than simply settling for where it’s been.
About Ada Glover
Ada Glover is co-founder and chief product officer at Zus Health, a healthcare data platform that enables value-based care by giving providers a real-time, comprehensive view of each patient’s care history in the form of a common patient record. Before Zus, Ada led teams at athenahealth and Zearn.
