When New York-based Zocdoc was founded back in 2007, the idea was to help patients get off the phone, founder and CEO Oliver Kharraz told Healthcare Brew.
The company created a website that helps patients find clinicians who fit their needs in their area and are under their insurance, and books appointments online.
But on May 1, Zocdoc launched a new product to get people back on the phone: an artificial intelligence (AI) voice agent called Zo. Zo helps people book doctor appointments 24/7—but instead of speaking with a person, patients speak with an AI voice that is trained to meet their needs.
“Until recently, we didn’t do the phone because the experience on the phone was just so miserable,” Kharraz said. “Now you can actually have a consistent experience, where the AI can pick up after the first ring an unlimited number of times concurrently [and] have a natural conversation with you.”
All things artificial intelligence (AI) are still capturing the most venture capital in healthcare, according to data from digital health strategy group Rock Health. In Q1 this year, $1.8 billion (60% of total digital health venture funding) went to companies centered around AI of some sort, including agentic AI and voice agents.
On top of Zocdoc’s offering, New York financial platform company Cedar also unveiled its own product, Kora, on April 29. Experts told us how these work and their hopes for the future of AI voices in healthcare.
Meet Zo
Zocdoc currently sells Zo to provider call centers, and is already working in real settings, including SINY Dermatology, Schweiger Dermatology Group, and Qualderm Partners. Kharraz said the product is creating value for providers who see spikes in utilization during consistently busy periods, like on Mondays or during lunch hours, and don’t want to hire more staff to reduce wait times.
“It’s a win-win-win for the patient, for the doctor, for us,” he said.
But it took two years to make sure Zo, which is trained to answer patients’ questions, “actually works,” he added. It also had to be able to understand “the cadence in which people speak with the specific idioms that they use.”
For example, Zo was originally trained on patients in New York, and when patients in Georgia used the virtual agent, it would interrupt them and not understand that “kindly” means “yes,” Kharraz said.
Zo also has a “working memory,” which helps patients not have to repeat themselves. If a caller says they can only book appointments on Wednesday, for example, the voice agent knows to not ask about availability on other days.
Navigate the healthcare industry
Healthcare Brew covers pharmaceutical developments, health startups, the latest tech, and how it impacts hospitals and providers to keep administrators and providers informed.
In the near future, Kharraz said, the company sees Zo filling prescriptions for patients who call in, completing referrals to other providers, and conducting outbound calls for appointment follow-ups.
Here’s Kora
Cedar offers a similar tool, but different from Zocdoc, is a medical bill payment platform. Its new AI voice agent, Kora, is designed to help manage call centers and answer patients’ questions about billing.
“Healthcare has many different demographics, many different age segments. And a large portion of patients, when they have a question, the first thing they do is they pick up the phone,” Dugan Winkie, head of commercial strategy at Cedar, told us.
For health systems and hospitals that buy Kora, the AI voice agent is trained to answer questions like why a bill is more expensive than expected (perhaps due to a high deductible, for example), Winkie said. It can also share payment options with patients and discuss financial assistance.
The company partnered with Atlanta-based national physician practice ApolloMD to test the product, which was made in collaboration with Twilio, a software development company based in San Francisco. It was trained on healthcare billing data and patient and health plan information, and is meant to respond naturally to patients.
“By providing fast, comprehensive responses without long wait times, Kora enables our teams to dedicate more time to complex patient needs,” Yogin Patel, CEO of ApolloMD, said in an April release.
Future of AI
Kharraz considers voice agents as part of phase two of AI implementation in healthcare.
While phase one included assistive tools, like ambient scribes that take notes for patients, this next phase is introducing automated tools such as voice agents that can be akin to a human “peer,” he said.
“This is, for now, limited to the administrative domain, but this is where Zo fits in. It can take the call front-to-end. There’s no…supervision by humans necessary,” Kharraz said.
Phase three, he said, will include augmentative AI to tackle tasks that are difficult for humans. For scheduling, this could mean tools that predict if a patient is more likely to miss an appointment and then coordinate transportation to help them get there.
“What makes us really so excited is not ‘How do you replace a human doctor?’…it’s about making the entire environment and ecosystem in which healthcare happens much more human-centric,” Kharraz said.