Even as the world is going digital, medical practice staff still spend a lot of time communicating over the phone. Front office staff handle a large volume of calls and texts to patients for appointment scheduling, referral follow-ups and prescription refills.
It's time consuming and inefficient for practices while also contributing to administrative overhead.
Startup Hello Patient built generative AI call agents to do the work of traditional support staff, freeing them up for more valuable, patient-facing work.
Hello Patient co-founder and CEO Alex Cohen spent about four years at Carbon Health running consumer and growth product teams and wanted to build better patient engagement experiences for other healthcare practices.
"I looked at the things that we had built at Carbon over that time, I rebuilt all of our call center infrastructure and I managed all the patient communications. It turns out the majority of those still happen over the phone and texting," Cohen told Fierce Healthcare in an exclusive interview about the launch of Hello Patient.
"We basically ended up building generative AI phone agents for healthcare practices. We essentially operate as a managed service where we go to a medical group and say, 'Hey, you're spending so much time calling patients that are referred to you, outsource that work to us and we will integrate with your entire tech stack, your EHR, and take that work off your plate'."
Cohen brought over a team from Carbon Health to start the new healthcare AI startup that has been working the last seven months to automate patient-facing communication work. The AI support agents are used for clinically administrative, non-medical workflows and the technology is HIPAA-compliant.
Hello Patient is now deploying the AI agents with a few medical groups, Cohen said, with a focus on working with medium and large healthcare practices.
"We're getting a lot of interest from virtual care, actually, because they don't have big phone administrative teams. Medical practices rely a lot on their medical receptionist and their back office teams to do the work. And if you're virtual, you don't have that luxury. You have to go stand up somewhat of a separate team and we become that arm for them, with the ability to stand it up without hiring an entire team," Cohen said. "We've had a surprisingly high number of conversations with health systems who have said, 'This is a top priority for us in 2025 and how do we get pilots off the ground?'"
The startup's approach is part of a paradigm shift where AI companies are offering service as software.
"I'm a big believer that the right way to do this is through a managed services business. We're not building SaaS. I believe we can do this as a service-as-software model where what was traditionally a services business now has software margins because I don't have to go employ a whole bunch of people to do the services under the hood," Cohen noted.
Hello Patient has picked up $6.3 million in seed funding backed by 8VC, Bling Capital and Max Ventures, as well as angel investors Ellen Dasilva, founder of Summer Health, Russell Fradin, previously vice chairman of Carbon Health, Shrav Mehta, founder of Secureframe and Swapnil Jain, founder of Observe.AI, among others.
“The Hello Patient team combines exactly what's needed to transform healthcare communications - technical virtuosity and deep operational experience from scaling Carbon Health," said Sebastian Caliri, partner at 8VC. "Credible voice agents need to exhibit higher-order logic, address patient concerns, and handle payments. Hello Patient enables each by capturing both structured data and the institutional 'dark matter' of healthcare practices: crucial information previously confined to post-its, whiteboards, and staffers’ heads.”
The startup will use the funding to invest in its engineering team and go-to-market strategy.
"I think people underestimate the amount of work that goes into building a production-ready AI agent that can go talk to patients. We've seen a lot of companies go after prior auth and talking to other practices. When you're talking to patients, it's a completely different expectation of the experience and they're holding us, I think, to a higher standard," Cohen said.
During a demo of the service, the AI agent on the call had a human-like voice and cadence, responded promptly and then efficiently scheduled an appointment.
Generative AI is taking off in healthcare and companies are using the technology to power medical scribe tools and handle other administrative tasks.
Hello Patient is focused on using the technology to modernize healthcare communications without eliminating phone calls and texts, which is still many patients' preferred way to connect with providers, Cohen noted.
"We saw this at Carbon. No matter how many self-serve tools I would build for consumers to get access to their charts or make appointments, they just picked up the phone and they called the practice, or they called the 1-800 number. It's just human behavior in healthcare, and frankly, the way I view it is we're building the same-self service tools just over the phone in the modality of communication that a patient wants to use," he said.
And, generative AI technology has evolved to the point that it can be used for more complex conversations to tackle tasks like appointment scheduling and answering patient concerns.
"The latency is there. It sounds realistic. It can actually handle crazy complex edge cases and unstructured inputs from the customer or from the patient, where this never existed," he noted.
The company does not build foundational AI or voice AI software, but uses existing software and then fine tunes its model specifically for healthcare practices and front-end administrative work.
"The place where we come in is all the infrastructure to actually make the agent work in real-time with a patient, for the context of the practice and the context of the standard operating procedure that it has to follow at that point in time," Cohen noted. So, for example, the AI agent has to understand the context of the conversation with the patient to reschedule an appointment if the patient requests it.
"We have all this internal tooling that we've built where we can essentially get the entire context of a practice loaded up into the system, similar to how you would train a support rep to do that, and then programmatically insert context at different points in the conversation depending on what the patient is asking or the reason for that call," he said.
There are other companies building AI agents, but within healthcare, there is a high bar to ensure AI operates within legal, ethical and practical boundaries, provides accurate information and protects patient privacy.
"There's a lot of core foundational infrastructure work that had to be built alongside of it. How do you actually get it to talk to an EHR? How do you ensure all the guardrails and safety where it's not giving medical advice and it's not going off track. We built our own like simulation framework. We can run simulations against the agents to make sure that they're not hallucinating and they're following all the right standard operating procedures. We evaluate every single call, and we programmatically tag them if we think that they're incorrect, and then we adjust all the rules based on that," Cohen said.
And, he's confident that Hello Patient AI support agents can handle these administrative tasks better than a human.
"All these workflows, they're expensive for the practice, and frankly, they're frustrating for patients too. It's all inefficient and sub-optimal. I think we're going to provide a 10x better patient experience and bring the cost down of doing this in the practice; it's better for everyone," he added.