It is hard to believe that in 2025, we are still dialing to schedule doctor appointments, get referrals, refill prescriptions, confirm office hours and addresses, and handle many other healthcare tasks. In fact, I created Zocdoc nearly 20 years ago to help patients avoid the dysfunctional phone experience and schedule appointments online. But I must confess that I have to pick up the phone sometimes, too—and I dread it.
I am not alone. According to a recent survey my company conducted, most Americans say they dread calling their doctor about as much as they dread getting a shot. At best, it is an inconvenience. At worst, the phone is a barrier to care and a wildly inefficient and costly channel for providers. Given that billions of healthcare interactions occur over the phone each year, this is a calamitous liability.
Relying on technology that was revolutionary in 1876 is no way to manage America’s healthcare administration. Especially because 150 years later, a new technology has emerged that will transform how healthcare operates: artificial intelligence.
While AI incites a range of feelings when it comes to its clinical implications and possibilities, I am much more focused on leveraging AI to solve healthcare’s administrative challenges. Considering these burdens cost $600 billion to $1 trillion annually, AI offers a remarkable opportunity to increase efficiency, reduce waste, and deliver first-class experiences for everyone.
As I see it, the effort to leverage AI to improve healthcare administration will unfold in three stages: Assistive, Autonomous, and Augmentative AI.
Assistive AI: Supervised Intern
In the Assistive stage, I view AI as a capable but dependent intern. It supports healthcare workers—but requires ample supervision.
Take AI scribes, for example. Although the technology takes records and transcribes patient encounters, physicians must still review and revise the notes to ensure they are correct. However, because AI scribes cannot yet work autonomously, documentation burdens remain and there is not a pure efficiency gain.
And they do not even begin to clear the waste and friction burdening office staff and patients.
Autonomous AI: Productive Peer
AI that works independently defines the Autonomous stage. Instead of an intern, think of the technology as a peer. This is where efficiency gains begin to accrue, and it is where we are now. To understand Autonomous AI’s potential, let’s return to the phone.
In healthcare, the phone is bad for business. Up to 20 percent of calls go unanswered, with each missed call costing provider organizations $200 to $300 in lost revenue. Half of Americans say they’re likely to switch providers if they cannot get through to their doctor’s office.
The phone is also bad for health outcomes. More than half of patients admit to delaying care when they cannot reach their doctor’s office, while a third admit to giving up on scheduling a visit entirely.
Autonomous AI is perfectly suited to remove these challenges. Appointment management is high volume, with billions of calls a year. It is highly volatile, with fluctuating peak times that are hard to efficiently staff. And it is highly rote, with most scheduling tasks being simple and repetitive. This makes appointment management a prime area for AI, freeing staff to focus on more valuable, complex responsibilities. In fact, the best AI phone assistants can successfully and independently handle more than 70 percent of a practice’s inbound scheduling calls.
The AI phone assistant’s ability to scale efficiency, improve the patient experience, and help practices counter macroeconomic headwinds and bolster their bottom lines is compelling.
Augmentative AI: Superhuman Colleague
When AI surpasses—and then scales—anything humans can do, we enter the Augmentative stage.
Soon, polyglot AI phone assistants will detect and then offer to converse in a patient’s preferred language. AI models will recognize patients’ appointment preferences, from cadence of visits to time of day and day of the week. AI will even predict the likelihood of a patient attending an appointment and then create custom engagement plans to increase those odds.
Augmenting AI will also excel at anticipating patients’ needs. It may contact a patient before a prescription runs out and offer to refill it with their pharmacy, or it may proactively schedule a checkup at their preferred appointment time.
This might sound futuristic, but given AI’s rapid advances, this stage is fast approaching. My prediction is this is likely only 12 to 18 months away.
AI or Obsolescence
It is tempting to point AI toward healthcare’s moonshots, but our biggest and most immediate opportunities for transformation lie in fixing the basics.
With AI, we finally have the technology to bring America’s healthcare experience out of the 1800s and into the modern age. In doing so, we can remove the friction, barriers, and waste that have disempowered patients and saddled providers for decades.
Change will happen suddenly, and then all at once. Because AI improves by the minute, healthcare organizations that wait to adopt this technology lose more ground every day. In no time, they will find themselves as relevant as a rotary phone.