November 30, 2022 was a key date in the history of artificial intelligence (AI). The debut of ChatGPT was the first time most people had heard of generative AI, and just two months after its debut, the chatbot gained 100 million users, making it the fastest-growing consumer software application in history.

Since the debut of ChatGPT, companies of all sizes have been in a race to integrate AI into their operations. But for Regence, ChatGPT was just one chapter in a story that started years before.

On this episode of HealthChangers, host Ashley Bach has the first of two conversations with Laurent Rotival, executive vice president and chief information officer for Regence. For this first part, Rotival talks about how Regence first started its work in AI and how that work has evolved.

For our second conversation, featured in an upcoming episode, we’ll explore how companies can think big and use AI to fundamentally change their business models.

Listen to the full podcast episode on the player above. Below are some highlights, which have been edited for length and clarity. 

AB: Regence first started working in AI a decade ago. What was the company's north star then, and has it changed?

LR: One of the great things about Regence is that we’ve never changed our north star. We’ve been single-mindedly focused on the individual. I've just crossed a decade at the company and I've got to be honest, that's one of the key reasons I'm here. I'm as passionate today as I was on my first day that our mission is not only to serve members, but it's actually to serve the individual, in their identity, whether they’re someone in a community, a parent of a family, an employee of a company, member of a [health] plan, and obviously, when they’re a patient, right as they pursue their best and healthiest life. So, for us, whether we make that possible through artificial intelligence, we do so through digital, or other technologies, ultimately, it’s always about serving the individual and serving the individual truly as a person, not just a population or a statistic.

AB: At Regence, we took inspiration and guidance from other industries outside of health care for how we've approached AI and customer experience. What have been our takeaways from these lessons?

LR: Ultimately, the game changer is being deeply focused on the humanity of the experience, right? And that's less about technology, and it's more about understanding the people you serve, understanding the context and where they live, understand the challenges they have, and ensure that as we design solutions, whether they're digital, whether they're AI-powered, whether they're human from a customer service standpoint or a clinical case management standpoint, that we always put that human interface and that human relationship and that human rapport first.

And so then technology becomes an enabler of making it easier, making it less burdensome, less bureaucratic, less time-consuming, so that [our employees], these fabulous humans who are inspired by our purpose and our cause to make health care more personal, can actually spend time doing that right, rather than filling out paperwork or clicking on buttons or using systems. And I think this is where not only the technology we've been working on for decades helps, but where AI now is a tremendous transformer of being able to shift from point-and-click and knowing these very rigid workflows to actually being able to speak and chat naturally with a technology system or a database or a knowledge base. And that's something that I'm incredibly excited about.

AB: There have been some challenges in integrating AI into health care. For instance, we have these legacy systems, and payers and providers may have different systems. It’s a challenge to get to a place where we are like an Expedia, where you can just book travel and have it instantly. Can you speak to those challenges?

LR: It's one of the tragedies of our industry, and one of the biggest differences with health care as a whole, is that in the book business or the travel business or the financial services business, these industries have a customer, they have a consumer, and they build all their systems to serve those individuals. In health care, if you look at health systems, hospitals and the electronic medical record systems that they’ve implemented, those are less about the patients and they're more about admitting a person into the hospital, registering them, coding them, transcribing the notes of a clinician or a physician, coding the orders and the procedures and the referrals they're asking them and then ultimately, billing. They're registration coding, billing and claim processing systems more than they’re really solutions that are built to make the patient have the best possible experience. And equally on the payer side or the health plan side, our traditional systems are about receiving claims, processing those claims, validating the eligibility and the appropriateness of the claim and then paying it.

The whole industry has invested hundreds, if not billions, of dollars to codify processes and workflows that at the end of the day, aren't particularly patient- or member-centric and certainly not person-centric. And that's one of the reasons why experiencing health care in the U.S. is so difficult because sadly, the systems aren't designed for them.

This is where Regence, over a decade ago, decided to make a major pivot, where, of course, we still want to ingest claims, process them and pay them. We do that very, very well, and over 94 percent of our claims are paid automatically, without any touch, and they go right through in almost real time. But what we've decided is that if we really want to serve our members at the deeply personal, individual-of- one level, we needed to have the same kind of platform, the same kind of technology, the same kind of design and experience philosophy that consumer companies had.

And so we built a full cloud-based solution that was digital-first, mobile-first, always on, always available, with the added requirement, of course, of the deep sensitivity, the deep privacy, the deep security that is unique to health care and deeply important to us, and obviously deeply important to the people we serve and the institutions we partner with. And that’s a huge difference. That's super exciting to have both those capabilities so that we can serve our members in a way that they expect and want, and we can play our role in full compliance at the highest performance levels with payers, providers and other third parties in the industry.

AB: How has our work in AI evolved over the years, while maintaining our north star of serving our members as individuals?

LR: To serve an individual, you need to understand who they are, so you need the data. You need to understand where they are, what's important to them, what challenges they face and then you need to be able to predict. We need to predict whether they’re at risk of certain types of disease or certain types of physical risk, or whatever it may be. And so that's why we started investing in artificial intelligence, machine learning, deep learning, data sciences, natural language processing, a little over a decade ago. And one of the great things about living in this part of the world [the Pacific Northwest], is that we were able to recruit, and thanks to our cause and our purpose, retain fabulous engineers and AI professionals who chose to stay here and really embrace this innovation journey and this transformational journey that we've been on.

We’ve been able to develop some great solutions for [predicting risk] so that we can help individuals be matched to the right services, the right products, the right capabilities in our products, and in our company; we've streamlined the way we process claims. And that's been very, very powerful. And we've had about a decade of experience enabling and supporting our clinical case managers with AI solutions that help them understand who they're serving and how they can proactively reach out to individuals who are at risk.

In the last [ChatGPT] inventory I read, there are now about 1,500 gen AI, large language models released in the product. They have specialty for images and graphic design and all kinds of other things, right? So really, just an explosion. But this is where a lot of companies are a little bit paralyzed by the speed at which this is moving. It's daunting if you don't have the experience to understand this technology, the experience and the talent to be able to try it, to experiment with it, to build proof of concepts, to kick the tires on it, the experience to understand well, how do you test it for reliability? How do you test it for trust? How do you test it for responsibility, right? How do you test it for privacy and security?

But as all of this world exploded around us, Regence had about 20 deeply professional AI experts on the team who not only had the technology expertise, but were deeply inspired by the same purpose, the same mission that every Regence employee has, which is to serve members. And it's allowed us to really lean in and develop what we think are very exciting innovations that we're already putting in the hands of our customer service and case management teams, and that we're contemplating putting directly in the hands of our members.

AB: What is an example of a gen AI solution that Regence is currently putting into the market?

LR: Over the last year, we've released a customer service assistant that uses gen AI to help our customer service professionals address the needs of our members. In the past, [our customer service professionals] would have had to find the PDF of the member’s benefit booklet, and then they had to use clunky PDF search capabilities, knowing exactly the right word. And it’d take minutes for this customer service professional to find the information that related to the member's question, and then be able to interpret what was written in this static PDF document, and then speak it back in a way that was context-relevant to the question that the member was asking.

Of course, our customer service professionals want to do the best job they can, and they want to do so in a personal, empathetic way, and they want to give good information. And sometimes the longer it takes to answer a question, the less confidence the member might have. What's amazing about this solution we've given them, and we've been running for almost a year, is just like you do with ChatGPT for non-health care purposes. Our customer service professionals can just type a question in exactly the way the member asked it, and our AI services will go straight into the documents, analyze them and bring back a response that’s phrased in a way that’s relevant to the member's question.. And if the customer service professional and the member want more information, they can just click, and it gives them the next level of detail with the exact page, the exact line, the exact area in the benefit document.

Now you can have a situation where our customer service professionals are far more confident; they have the information immediately available. The member doesn't need to know exactly the jargon or the technical terms or the specialized words that would allow them to get the answer they want. Our gen AI-powered customer service assistant will do that for them.  

We’ll have our customer service professionals able to serve half of our members with those solutions by the end of this year. And we’re piloting direct-to-member services for the same technology with our Regence employees between now and year-end. If that works out well, and we feel that it passes the levels of trust and responsibility that we believe are acceptable, then we’ll release it to our customers.

AB: That's really incredible, because I think that's one of the complaints people have about the insurance industry is those benefit booklets are just so hard to navigate.

LR: It's shockingly complex, unfortunately, but it's the nature of the industry. This is one of the things that makes us so excited at Regence. We've gone a long way in automating, digitizing, making it easier to manage good information, reliable information; but to your point, it was still hard to interact with that information right, or to find it. And what generative AI is doing, in particular the services that we've configured and built, is that it's allowing you now to speak naturally, like a normal person, to that data, to those expert documents, to those knowledge bases. And that's actually pretty extraordinary,