The Center for Data Innovation spoke with Ankit Jain, CEO and co-founder of Infinitus, a California-based company that develops AI-powered solutions to automate healthcare communications. Jain discussed how the company’s platform leverages voice AI agents and machine learning to streamline administrative tasks, reduce provider burnout, and improve patient access to critical healthcare information.
Martin Makaryan: What was the inspiration behind Infinitus and what led you to found the company?
Ankit Jain: In 2018, I was working for Google when the company demoed Duplex, a technology that could make phone calls to book spa appointments or restaurant reservations on a person’s behalf using natural language processing and speech recognition. When I told my wife Shailvi, she said, “You mean to tell me that you have this incredible technology that can make phone calls for people and you’re using it to book spa treatments?!”
Shailvi has spent years in healthcare, where administrative burdens often take time away from patient care. When I asked how she would use the technology, she immediately saw its potential to automate back-office tasks, reducing time spent on hold and eliminating the need for repetitive phone calls. These inefficiencies force healthcare professionals to focus on logistics instead of patients.
That conversation sparked an idea. I called my longtime friend and former collaborator, Shyam Rajagopalan, to discuss the problem and a potential solution. Soon after, Infinitus was born.
Makaryan: What is Infinitus’ mission?
Jain: The World Health Organization (WHO) projects a global shortfall of 10 million health workers by 2030. Our mission is to help address these staffing shortages while also improving patient outcomes at scale. We do this through our suite of AI agents and copilots designed to automate and augment administrative and clinical tasks.
For example, our voice AI agents fully automate phone calls for tasks like verifying benefits with payors or educating patients about their medications and procedures. Our AI copilots enhance human productivity by taking on time-consuming elements of calls, such as navigating automated phone menus or waiting on hold. Currently, human healthcare workers, who are experiencing record levels of burnout and leaving the field, handle most of this work.
Makaryan: What are some of the challenges of automating healthcare calls?
Jain: From a technical perspective, both administrative and clinical healthcare phone calls involve multiple phases that AI must be able to navigate—automated menus, hold times, transfers, escalations, varied ways of phrasing questions and responses, and long conversations requiring extended context retention. In 2024, the longest call we processed lasted 3 hours and 15 minutes, with 306 conversational turns—a significant challenge for automation. Ensuring data accuracy is also critical. For example, when speaking with an insurance representative, our AI must detect and challenge incorrect information when necessary—something that happens more often than you’d think.
When automating patient calls, the AI must not only have the right clinical knowledge (without hallucinations) but also express the appropriate level of empathy. Consider a patient recently diagnosed with Crohn’s disease receiving a call from an AI agent welcoming them to a support program. If that agent lacks a thoughtful and reassuring tone, the patient is unlikely to trust it—let alone engage with the call.
Finally, the healthcare industry remains highly fragmented. While we hope for a future where administrative calls are minimized and data flows seamlessly via APIs, today’s reality is that information is scattered across multiple systems. Anyone who has called their insurer about coverage has likely experienced this first hand—sometimes even human agents must rely on reasoning and judgment to interpret elements of a specific plan.
Makaryan: What is the technology underpinning Infintius’ offering?
Jain: Infinitus AI agents and copilots run on a multi-model, multimodal AI system supported by an in-house knowledge graph. This graph continuously compiles up-to-date intelligence on commercial and government insurance rules and guidelines, drawing from millions of automated phone calls, electronic transactions between healthcare providers and insurers (such as eligibility checks and claims processing), direct integrations with insurers’ systems, and official policy documents outlining coverage and reimbursement rules.
We use large language models (LLMs) safely and effectively through a custom-built coordination layer that sits on top of LLMs. This layer ensures AI conversations stay within approved topics while still allowing for natural, context-aware interactions—a level of control that even the most advanced LLMs don’t natively provide. This approach enables reliable AI deployment in healthcare, eliminating risks like hallucinations or misinformation.
Our AI system integrates both third-party LLMs and proprietary models trained on our exclusive dataset. We started by fine-tuning Google’s BERT and T5 and have recently optimized Gemini and OpenAI’s GPT-4o to enhance their performance for our specific use cases.
Makaryan: How has Inifinitus helped healthcare companies and patients?
Jain: Today, Infinitus supports 44 percent of the Fortune 50, impacting the workflows of over 125,000 providers. Our automation alleviates staffing challenges while ensuring patients receive timely access to necessary medications.
One success story is Cencora (No. 10 on the Fortune 500), where our voice AI agents offset the workload of more than 100 full-time employees—positions that would have been difficult to hire and retain. Another example is a large biotech company, where our AI copilot helped reimbursement specialists verify patient benefits faster. One specialist even told us that using the copilot “maintains [their] sanity.”
Beyond our direct customers, Infinitus is streamlining healthcare interactions for providers, patients, and insurers. For example, when our voice AI agents call one of the largest insurers in the country, those calls are over one-third shorter than comparable human calls, reducing wait times and helping the insurer manage high call volumes more efficiently. On the patient side, those who interact with our AI agents gain faster access to critical information—and they can ask questions 24/7, rather than waiting for support programs to open during business hours.
