In this episode

AI is here — in healthcare and in our everyday lives. Ankit Jain, founder and company lead at Infinitus, discussed the impact of AI on healthcare today and in the near future. He covered the following themes:

• AI can help healthcare data get to the right people, at the right time.
• AI can help get patients answers faster.
• The future of an AI-powered healthcare system will require “operating systems” to overlay workflows.

“I think we’re going to hear real stories of how the patient experience is getting better. We’re going to hear about how patients feel like they’re more heard, that the system has more time for them. The optimist in me believes that AI can give patients the gift of time.”

– Ankit Jain

Key takeaways

AI can help healthcare data get to the right people, at the right time.

“Before starting Infinitus, I’d started Google’s AI venture fund, Gradient Ventures, and I was investing in a number of different AI technology companies,” said Jain. “I showed my wife some technology to make restaurant and spa reservations, so you could say, ‘Hey Google, make me a reservation at my favorite restaurant.’ And she said, ‘This sounds like incredible technology, but is restaurant and spa reservations the best place where you could apply it?’” The question started what Jain called Infinitus’ “journey of understanding how data exchanges hands in healthcare.”

He argued that “the way processes get initiated, that data exchanges hands over phone calls and faxes…that’s the fundamental problem. How do we make data proactively, transparently, and instantaneously available to everybody who needs it?”

Jain covered the possibility of AI to help circumvent challenges of data structure and format that make interoperability more difficult. “The biggest problem in healthcare is that everyone’s data exists…in ways that are slightly different,” he said. [The previous generation of technology] said, ‘I can’t submit prior auth information from the provider instantaneously because the payer expects it in a slightly different format’…now, large language models came out of Google’s translation group, and the whole tenet of it was,  ‘how do you translate tokens in Language A to tokens in Language B?'”

“I think there’s a lot of excitement around that ability to transform data and and connect systems,” said Jain.

AI can help get patients answers faster.

Jain covered the ability of AI to provide a “knowledge base, when you want it, how you want it,” as an area where he’s optimistic about the potential to improve healthcare. Jain covered a personal example where more information, sooner, would have been hugely beneficial.

“My daughter was born with a very rare disease, and when we were told that we broke down,” said Jain. “We didn’t have the capacity to ask all the questions, and the questions we did have came at 10PM after the kids were asleep,” he said. Eventually, Jain’s family learned that different forms of the disease had different treatments, and one form was curable. “She was cured, but it was a six-week ordeal where Dr. Google was our resource,” he said.

Jain imagined the possibility of a better experience for patients and families by using AI to surface existing information anytime. “We have the opportunity to take those centers of excellence for each disease, whether rare or common, and…make it available to patients how they want it, when they want it. So at 10PM, when you have a question,” you as a patient could call or text to get answers, he said.

“There are so many parts of healthcare delivery and healthcare policy that I think will be affected by the ability to translate from what one group has published to what the other group needs to ingest, so we can connect those dots seamlessly,” he said.

The future of an AI-powered healthcare system will require “operating systems” to overlay workflows.

Jain noted that AI for administrative workflows, in particular, is already in use and will continue to grow. Looking toward the future, he sees coordination between AI and human workflows as the next big challenge.

“[Over the next few years,] hundreds if not thousands of companies are going to come up with point solutions of AI agents. The first five, ten people will excitedly go and implement, but then they’ll realize, ‘Well, humans were connecting step 1 in the workflow to step 2.’ Or step 1 might be done by an AI agent, step 2 by a human agent, step 3 by an AI agent,” said Jain.

He said that “what we’re going to need as an ecosystem is a coordination layer, an operating system, that brings together the workforce of the future, which is human and AI agents working side by side.” Jain is of the opinion that in the next few years, “we’re not going to be using the term ‘AI.’ We’re going to be focusing on the specific problems [the AI technology is solving]. It’s not going to be sexy anymore…but I think we’re going to start hearing real stories of how the patient experience is getting better.”

In the meantime, Jain recommended that for health systems evaluating AI, “do the same [test] task ten times in a row. It might work perfectly the first time. But [what happens] when you do it ten times?” And for community health systems in particular, Jain said that now is a great time to try AI because “you can find very hungry partners to go try things out with very little risk.”