What AI actually does
There’s a version of the AI conversation that stays permanently abstract – all potential, no proof. The leaders we spoke to for this report are thinking about it in concrete terms, however, and shared what’s already in production, what it’s already changing, and what it’s freeing their people up to do.
"I think about it in two ways: accelerating and augmentation. Of course the fear is that it's going to augment and do away with careers. But I don't think that's the way to think about it. I think about it as far as accelerating the work that we do today. We could all work 24 hours a day, around the clock, and still not be done with our jobs."
That reframe showed up across nearly every conversation we had. But acceleration means different things depending on where you sit. For some, it’s automating the administrative layer of hub operations. For others, it’s the speed at which a prior authorization gets resolved.
At Optum Rx, it’s both.
“Prior auth is ultimately a point of friction – improving turnaround times is key. The clinician is making the care decisions, but supporting that clinician to ultimately move through that in a more expedited, consistent fashion is super important to get patients to therapy faster. We've seen a 20% improvement, and we're excited about the opportunities ahead there.”
Numbers like that matter because they’re not projections. They’re production results from one of the largest pharmacy benefit managers in the country. And they point toward something the rest of the industry is actively working to replicate.
Today, the benefit verification workflow is one of the clearest illustrations of what that shift looks like in practice – and it’s something Infinitus has been automating for almost eight years.
"The use of AI allows us to use an agent or a digital connection with that payor – it still relies on a human, but that work can be done in a much more timely fashion while that person in the patient support program can be talking with the patient, talking with the provider, or driving other workflow mechanisms ahead."
This is the pattern that emerged again and again: AI handles the transaction so the person involved can handle the relationship. And as more of the transactional layer gets automated, the vision of what’s possible downstream gets bigger.
“Today most organizations are still using AI in relatively narrow ways. Summarizing meetings, automating workflows, generating drafts or supporting analytics. … From a healthcare standpoint, I would see a much broader range of AI utilization from front end work like scheduling and prior auths all the way through supply chain management, as well as probably some financial billing and claims process.”
For Novartis, the opportunity centers on what automation makes possible for the people closest to the patient.
"The most exciting opportunities for AI in patient support are to be able to augment our agents' capabilities and focus more specifically on improving our overall customer experience – versus focusing on those back-end tasks that oftentimes inundate their daily activities."
At Infinitus, we’d add one more principle to this list – one that George Van Antwerp of Prime Therapeutics put better than most: before you automate something, ask whether it needs to exist at all. Eliminate first. Then look to automation.
“Don't just take AI as a hammer and hit every nail. Think about how you can eliminate things, automate things, and then systematically apply AI to the right things.”
The leaders who are seeing the most meaningful results aren’t just layering AI onto existing workflows. They’re using it as an opportunity to rethink them entirely.
The next section explores what all of this means for the people at the center of the patient journey.
