- US healthcare is notoriously inefficient and immune to technological innovations that seek to improve it. Each year, $1 trillion is spent just on administrative costs.
- That’s starting to shift. The rapid evolution of generative AI agents is poised to transform patient care — making it easier for patients to access services while lightening the load for providers and administrative staff, allowing them to focus on what really matters.
- A successful use case is AI virtual receptionists, which are already handling most inbound calls from patients at dozens of healthcare facilities.
KEY INSIGHTS:
We’ve all been there. Your symptoms flare up again and you call your doctor, hoping for a quick referral to a specialist. After several minutes of hold music, someone finally picks up — only to put you on hold again while they check your information. You get an appointment, but later discover you didn’t have the right authorization from your insurance provider or were sent to the wrong specialist — a frustratingly common scenario. In fact, one survey found that nearly 8 percent of all referrals in the US are “clinically inappropriate,” meaning almost 20 million patients each year are directed to providers who can’t actually help them.
These inefficiencies aren’t just irritating — they waste time and money, delay critical care, and lead to worse health outcomes. It’s no wonder that 60 percent of patients skip medical appointments due to scheduling hassles alone.
So, why does this keep happening? The root of the problem lies in overburdened patient access centers. Despite the rise of online portals and apps, most healthcare appointments (88 percent) are still booked over the phone because patients want the reassurance of speaking to a human. But since COVID, call center staff have shrunk, turnover has surged past 30 percent, and many of the employees taking patients’ calls are now less experienced. With a skeleton crew handling a flood of inquiries on fragmented scheduling systems, errors and miscommunication are inevitable. Demand surges — like Monday mornings or post-holiday rushes — only make things worse, driving call volumes up by as much as 250 percent.
Generative AI can succeed where many have failed
If you’re like us, you believe accessing healthcare shouldn’t be this hard. Yet, these inefficiencies are nothing new. The US healthcare system is notoriously bloated, with up to one-quarter of all healthcare spending (roughly $1 trillion) going to administration, including billions poured into call centers. Over the years, tech companies have tried — and failed — to fix these problems.
“A new wave of generative AI applications are beginning to actually transform healthcare.”
But today, we’re on the cusp of something different. A new wave of generative AI applications is beginning to transform healthcare. They’re not going to solve every problem, and nothing will happen overnight, but they will reduce a whole lot of the drudgery in the system, making it easier for providers to do their jobs and giving patients faster, more reliable access to care.
Why is this time different? The rapid evolution of AI. Healthcare generates 30 percent of the world’s data volume, and AI’s ability to process vast amounts of unstructured information has always made it a promising tool for the industry. But now, large language models and text-to-speech and speech-to-text technologies have reached a level that can be useful for patients and providers.
An always-on super-receptionist
At Assort Health, for instance, we’ve built an AI-powered voice agent designed to revolutionize patient call handling. This intelligent, tireless virtual receptionist can manage nearly all inbound calls to a clinic or doctor’s office — without human intervention. It is handling millions of calls a year.
Here’s how it works: When a patient calls, the AI assistant answers almost immediately. Armed with customer phone data, it greets the caller and listens to their request. Depending on the need, the agent might provide an answer (“Your lab results are expected on Thursday”) or ask follow-up questions (“Do you have a preferred day to see the doctor?” “How long has the pain been bothering you?”). While it’s talking to you, the virtual receptionist works seamlessly in the background to gather information from your electronic health record and complete a variety of tasks, such as:
- Triaging the appropriate specialist for your condition
- Booking or rescheduling appointments for new and established patients
- Creating a new patient entry in the electronic health record (EHR) or practice management system
- Refilling prescriptions
- Updating insurance information
- Adding notes to the EHR for the doctor to review
- Routing calls to the appropriate staff or an on-call nurse
This level of multitasking would be exhausting — if not impossible — for a human receptionist. But AI never needs to put a patient on hold, never gets overwhelmed, and never takes a break.
This isn’t just another voice-recognition system that responds with “I’m sorry, I didn’t get that” or hangs up on you (“That’s all the information we have. Goodbye.”). Far from it. The AI assistants we and others have developed are sophisticated and intelligent enough that callers often forget they’re speaking to a machine. Assort’s agents, for example, can converse in two dozen languages and understand speech even with heavy accents or background noise.
“The AI assistants are sophisticated and intelligent enough that callers often forget they’re speaking to a machine.”
And here’s something surprising: Patients often prefer AI interactions over human ones. A recent JAMA Internal Medicine study found that nearly 80 percent of patients preferred chatbot responses to those of physicians. Why? Because AI spends more time acknowledging and validating patients’ concerns, whereas people often respond by sharing a related experience from their own lives. In other words, AI makes the exchange about the person; humans make it more about themselves.
To be clear, AI isn’t meant to replace genuine human connection and empathy, especially in healthcare. Assort’s agents identify themselves as such and don’t attempt to mimic emotions. They won’t say “I’m sorry” for a patient’s discomfort and they don’t offer medical advice.
Driving impact
Early results have been incredibly promising. Clinics using AI-powered agents have seen dropped calls and hang-ups decrease nearly threefold, simply because patients aren’t left waiting or forgotten. Surveys show that most people will only tolerate being on hold for a couple of minutes — many hang up after just 60 seconds. Even more impressively, the majority of patient inquiries are fully resolved by AI without requiring human intervention. This allows front-desk teams to focus on higher-value tasks, like assisting in-person patients.
For healthcare practices, the reduction in operational complexity and overhead in patient access centers lowers costs. At the same time, an increase in the number of patients seen and amount of appointments booked boosts revenue. Just a small difference in performance can mean millions of dollars in reimbursements for practices every year.
“Clinics using AI-powered agents have seen dropped calls and hang-ups decrease nearly threefold, simply because patients aren’t left waiting or forgotten.”
In the long run, we believe this technology can make healthcare far more accessible — especially for the 60 percent of patients who skip care due to the hassle and the millions living in rural areas who are underserved by an insufficient number of overworked doctors.
That said, it’s far too early for a victory lap. Transforming healthcare demands significant effort, resources, and continuous refinement. Through our journey, we’ve learned that two key qualities are non-negotiable for AI agents in healthcare:
Accuracy. AI is only as good as its training data. To function effectively, models must be continuously trained to handle thousands of different scenarios, ensuring they take the right action every time. Ongoing testing is critical for catching and correcting errors before they impact patients, including hallucinations, where an AI model might generate a plausible but incorrect response.
Seamless integration with specialty workflows. To truly assist both patients and providers, AI agents must have access to comprehensive healthcare data. This means deep integration not just with major EHR platforms like Epic and Athenahealth, but also scheduling systems, insurance databases, and administrative rules. Assort’s inbound agent, for instance, can handle complex inquiries because it is able to analyze a patient’s entire medical chart, insurance details, and appointment history — allowing it to tailor interactions to individual needs. It can also automate scheduling with precision, knowing, for example, which doctor handles knee injuries versus shoulder injuries, and how long a diabetic check-up should be.
“Today, AI might book your appointment. Tomorrow, it could be a virtual assistant explaining your lab results or checking in on your chronic condition.”
Beyond the call center
We’re rapidly entering an era where AI-driven, multimodal agents — capable of voice, text, image and video generation — will be seamlessly integrated into nearly every aspect of healthcare. Today, AI might book your appointment. Tomorrow, it could be a virtual assistant explaining your lab results or checking in on your chronic condition. Essentially, AI agents will help facilitate or automate any interaction between you as a patient and a provider or practice.
The momentum behind this shift is undeniable. A recent analysis found that 145 healthcare AI companies have collectively raised over $20 billion in funding. Here are several other ways these companies are transforming healthcare:
Personalized patient outreach and care coordination. Instead of waiting for patients to call, AI can proactively engage them, with the goal of improving healthcare access and reducing missed appointments. These efforts can range from simple appointment reminders to post-surgical check-ins or steering high-risk patients toward preventive care to avoid costly emergencies. Many serious health issues escalate simply because patients forget to take medications or skip routine screenings. AI-powered reminders and follow-ups can provide the gentle nudges that make all the difference.
Pre-visit preparation and post-visit follow-through. Ever spent the first 15 minutes at the doctor’s office filling out tedious forms? AI can eliminate that hassle. Through interactive voice or text sessions, AI agents can guide patients through pre-visit intake, update insurance information, and handle consent forms — all before you step into the office. After a visit, AI can deliver discharge instructions in plain language, explain test results, and ensure follow-ups are scheduled. Acting as a personalized health concierge, AI keeps patients engaged so they don’t fall through the cracks.
Workflow automation for administrative staff. Beyond patient care, AI is also revolutionizing medical office operations, helping overworked staff complete in seconds what used to take hours of phone tag, paperwork, and document scanning. The Infinitus AI agents, for instance, call insurers to verify benefits and obtain pre-authorizations. Notable’s AI tools scan massive volumes of patient data, flagging high-priority cases so care coordinators can focus on what matters most (more here).
AI support for providers. AI won’t replace doctors or nurses, but it will make their jobs easier. Many doctors already use AI-powered note-taking assistants, allowing them to focus on patients instead of screens. In the future, AI will also help physicians quickly review patient charts and medical histories, retrieving critical insights in seconds rather than forcing them to sift through pages of records.
Healthcare providers can no longer afford to ignore the inefficiencies that frustrate their patients — or the technological solutions that can eliminate them. Embracing AI-driven patient engagement isn’t about replacing human touch; it’s about removing the barriers between patients and that human touch. The future of healthcare is calling — thanks to AI, we just might get an answer on the first ring.
