Watch the Fireside Chat

Transcript

John Lynn 00:00

Hey, everyone. I’m John Lynn, the founder and chief editor at Healthcare IT Today. We’re excited to bring you another in our series of interviews with top leaders in health IT. Our guests today are Ashish Desai, President of Benefits Lifecycle Solutions at Simplify Healthcare, and Jamie Hernandez, SVP of Product Engineering and Development at Simplify Healthcare. Welcome.

Ashish Desai 00:23

Thanks for having us, John.

Jamie Hernandez 00:25

Thanks, John. Great to be with you today.

John Lynn 00:28

Lots to talk about. You’re doing some really interesting work and I’m excited to learn more about it. But before we go there, Ashish, why don’t you kick us off? Tell us a little bit about yourself and Simplify Healthcare.

Ashish Desai 00:44

Sure, John. My name is Ashish Desai and I lead the benefits lifecycle solutions business as president at Simplify Healthcare.

At Simplify Healthcare, we have AI-powered platforms that automate business processes specifically for healthcare payers. We are 100% focused on the health insurance industry. That’s the only vertical we work in, so we have very focused, deep subject matter expertise. Within the payer world we focus on two key areas: benefits and provider lifecycle. We work with more than 70 payers across the country, including some of the large national plans as well as regional and smaller ones, across lines of business. We’ve been in business for more than 18 years, we have a thousand-plus people, and we’re headquartered in Chicago. We’ve had great growth over the last four to five years and we’ve established ourselves as leaders in our areas of focus.

As leader of the benefits lifecycle solutions business unit, I lead a portfolio of solutions along with Jamie and other leaders. That includes benefits plan management, automating customer service including benefit inquiries, and SimplifyDocs, which generates documents and collateral for customer communication and compliance.

I’ve been with Simplify for over a decade and have played different roles through the scaling of this business. I have more than 30 years in IT, services, and software products, predominantly in healthcare for over 15 years now. That’s Simplify, John.

John Lynn 02:16

Love that. Jamie, how about yourself?

Jamie Hernandez 02:18

Hello, everyone. I’m Jamie Hernandez, and I lead the development and delivery of the enterprise solutions Ashish mentioned. I work very closely with our teams and with our customers, making sure they’re implementing good operational models, that they see measurable outcomes, and that we listen to them and hear their voices when we look at our products in engineering and development.

John Lynn 02:42

Jamie, I’d love to start with you. How has the expansion of engagement channels changed the operational demands on health plans?

Jamie Hernandez 02:51

We’ve all seen it. There are countless ways to engage. How many different ways can you reach your members, providers, clinicians, back office, and so on? With that, you have to think that every channel has its own logic and its own owner, and not all of the sources are the same. So it’s really about maintaining consistency as you add channels, making sure you have a single source and that the information is being cascaded to those other channels.

Otherwise you have situations where members and providers call in to ask questions about benefits and get different answers. Members are frustrated, providers are frustrated, and then they don’t trust anybody. They want to call in and talk to everybody instead of using all of the other channels health plans are trying to innovate with and bring to market. There’s financial and legal risk too if you’re giving wrong or inconsistent answers.

John Lynn 03:52

That’s interesting. Ashish, what would you add to that? What Jamie just described was fragmentation. Where does that fragmentation create the most pressure across the experience and the operations for a health plan?

Ashish Desai 04:05

Typically, if you look at the IT landscape of any payer, we can broadly divide it into three areas: front office, middle office, and back office. Over the decades, payers have invested heavily in automating claims processing, but tons of peripheral processes are still either manual or semi-automated with point solutions or homegrown applications. You can still find processes being managed through Excel sheets. The data is scattered across the enterprise in a variety of different silos.

When we talk about benefits, the data could be sitting in various systems across the enterprise. Different teams in different departments are coding benefits in their respective systems, so the same benefit may be coded differently in different systems over a period of time. For a simple example, chiropractic may be coded as “chiropractic” in one system, “chiro” in another, and “spinal manipulation” in a third.

As Jamie mentioned, payers have been investing heavily in building the digital front door, adding more channels and giving members more ways to engage. But when a member or a prospect has questions about benefits, the data is scattered across the enterprise and coded in so many different ways, so providing an answer really becomes a challenge.

When a member calls a customer service agent, the customer service team now needs to aggregate information from multiple sources. Apart from the coded information, there might be documents, medical policies, DME lists, a knowledge repository. They have to refer to three or four different screens and several documents, pull all the information together, synthesize it, first understand for themselves what the appropriate response to the question is, and then explain it to the member in human-friendly language the member can understand. So it’s quite a complex job for customer service agents to provide the kind of response members are expecting.

Nowadays members are looking for an e-commerce experience like Amazon. I click a button and I get the information. I place an order and I get a delivery. Compare that with the experience they're getting with health insurance. The reason is all this fragmented data in the back end. Translating these disparate data sources into human-friendly language for a wide variety of inquiries is a huge challenge operationally as well as technically. That's the main pressure point, and that's why payers are struggling to provide the level of experience members are expecting from them through the shopping process, post-shopping when they really become a member, and through the whole lifecycle.

John Lynn 06:49

Interesting. How are they approaching the application and, in many cases, the interpretation of benefit and policy logic across channels? Is it this mix of Excel spreadsheets that you see across the plan, and what’s the better way to approach this?

Ashish Desai 07:06

The industry has approached this in a variety of different ways over the last few years. As the demand started increasing, and as government regulations put pressure on payers to become more transparent, significant investment is being made by payers in modernizing their infrastructure to provide a better member or consumer experience.

To do that, payers have invested significantly in building various channels over the last few years. But the back end is still a little messy. As I said, there are a variety of different data sources, and in bringing that together, payers have been driving a variety of different initiatives, many of them internally.

One area where we see a lot of investment in the last few years is contact center modernization. Many of the CCaaS vendors have come up with AI-powered offerings where they say they can help automate, but that’s still automating the front end. Then payers realize that while modernizing the front end, they still need to get clean data from the back end.

Just recently the LLMs came into the picture, and a lot of payers started initiatives to leverage these LLMs, saying they’ll aggregate the data, apply generative AI on top of it, and an answer will pop out. Yes, the answer will pop out, but the way these LLMs operate, it’s garbage in, garbage out. If you have messy data, you’re not going to get a clean answer.

A lot of POCs were conducted and a lot of prototypes were initiated in the last two years, but very few of them, close to none, have gone to production. The reason is that with generative AI you get inconsistent responses, risk of hallucination, and the possibility of giving a member the wrong benefit quote. There are compliance and financial risks.

So we took a really different approach, taking a holistic view and coming up with a solution that is end to end. Not solving the front-end problem, not the back end, but looking at the entire journey, starting from data. We have three platforms that, when brought together, provide a comprehensive solution.

First and foremost is a data activation platform we call Context1, which provides a payer ontology. It transforms fragmented data into clean, standardized, enriched information that’s contextual and helps make real-time decisions. That’s the data foundation we start with, leveraging our very deep subject matter expertise in this space.

Once the foundation is in place, you can apply meaningful AI on top of it. That’s Xperience1, the solution Jamie has been leading for the past couple of years. It’s an intelligent customer experience platform, and what it does is provide clarity for every benefits communication. We have a private LLM fine-tuned for benefits information, purpose-built for benefits.

We have intelligent content management, which allows you to smartly configure the content for a variety of different channels. For example, a customer service agent will give an answer on a call. For the same question, when the member goes to a portal, you may want to explain it in two paragraphs. On chat, it has to be in question-and-answer format. The same response, delivered through different modalities, has to be configured correctly through this digital content management.

Then we have advanced search, IntelliSearch, which is AI-powered search capability. With simple keywords, or for a provider calling in with a variety of codes, you can run a search and get the same consistent answer.

Very recently, we launched an agentic AI automation platform called Foundry1. If a payer is advancing its technology and looking to leverage an agent framework, we have that capability. The data foundation and Xperience1 intelligence put together help generate accurate, consistent, real-time responses for all the different channels, which can be powered through APIs, content streaming, or a variety of other ways.

That's the approach we've taken, and we've seen some phenomenal success with those capabilities. I have not come across any other company that has taken this comprehensive approach, starting with the messy stuff first and then going to the front end to provide clean, consistent responses. I hope that provides a picture.

John Lynn 11:55

No, for sure. It reminds me of a common tech principle: to make something easy takes a lot of work.

Ashish Desai 12:03

One thing I’ve learned is that people don’t want to take up the messy job of cleaning the data. That’s hard work, it takes time and a lot of effort. But we have AI tools and technologies to do it faster now, to solve this problem.

John Lynn 12:22

That’s awesome. Well, we’re in denial that our data is bad too, and we don’t want to admit that maybe we’ve made some mistakes before. There’s probably some social and behavioral things at play. Jamie, what would you add?

Jamie Hernandez 12:37

Just one thing to add. I think Ashish covered it really well. AI is the buzz, right? Let’s just apply AI. But people have to keep in mind that AI is only as good as the source it’s using. If you have inconsistency in your data, if you have competing information, if it’s in multiple places, you can only expect it to do so much. That goes back to having that core foundation and making sure you’re solid there. Starting with the messy stuff first, then building off of that. That’s really key.

Ashish Desai 13:14

Just one line to add there. When people talk about AI, we have to be very clear. All these LLMs will give you a probabilistic response. That means it's not necessarily going to be 100% accurate. The approach we've taken is to generate a deterministic response: the same answer every time you ask the same question. Unless you have that level of accuracy, no payer wants to put the solution in production.

John Lynn 13:43

Yeah, absolutely. It makes sense. And we were used to that, right? We're used to a system telling us it doesn't know the answer, whereas the LLMs now happily give you an answer whether they have one or not, which becomes problematic in healthcare in particular.

Jamie, what defines a well-governed approach across digital and contact center environments? Ashish described the platform. How about governance, and making sure it’s working effectively across digital and contact center environments?

Jamie Hernandez 14:14

I always think it really starts from an organizational standpoint. They have to have good governance in place, not only with their data but with their channels, their business units, their sub-business units. If you don’t have good governance across the board, you have silos and different departments doing everything independently, which brings you back to square one.

So I think it really comes down to making sure you have that centralized data. You have a repository that holds all of your different data feeds, types, and formats, and you’re creating a single source of truth for your organization, not for a department.

From there, I think about traceability. Where is this data going? How are people using this data? What information is being shared outside of the health plan to providers, clinicians, and members? For organizations to really understand how well the process is working, they have to be able to measure against something, and that is your output.

And then finally it all comes down to controlled content. It’s about making sure you have the right content in the right channels. If a member is speaking to a CSR they may get a paragraph, but if they use the portal they’ll have a subset of that. The information is the same. It’s just the version they’re getting.

So when I think about controlled content, it’s about having a content manager, a place where you can update your content one time and have it cascade across all of the different streams, business units, and channels. From your base content you can also define whether you want to follow a health literacy model for your members, or maybe for your mobile app you only want 200 characters instead of what you see in the portal. The bottom line is the same base is being used across all of those channels.

That really is an organizational movement. They have to come to the table and agree that yes, we need to have a centralized source, we need to have traceability, and then we need to be able to control our content to ensure that all the work we've done is now being cascaded across different areas of the organization.

John Lynn 16:54

Interesting. I really love what you said there about having a single source of truth across the organization, not just the department. That's powerful. Ashish, I'd love to hear from you. What distinguishes the organizations that execute effectively in this area, to Jamie's point?

Ashish Desai 17:12

John, that’s an important question. Payers have been trying to solve this for quite some time and have taken different approaches. I think one of the key challenges has been change management: investing in the right area, using the right tools, and taking the end-to-end view.

You'll see that initiatives taken by the business go one way. IT takes another approach. Now you also have digital innovation teams, and they want to take a different approach. Unless the whole organization, typically the enterprise architect or someone in that role, takes a holistic view of addressing this challenge end to end, there are continuous ongoing challenges. The initiative is taken up, millions of dollars are invested, we see progress, but in the end the solution is not ready for prime time. That's the biggest challenge.

So starting with an end-to-end view, coming up with a comprehensive solution approach, investing in cleaning the messy data, willingness to get your hands dirty, that is necessary. And implementing the right kind of data governance. It's not a one-time job, it's an ongoing process. Those are some of the important aspects.

We have seen some payers successfully deal with this challenge. With a couple of the customers we're working with, we started engaging almost three years ago. It took some time for us to get there, but the willingness to put in the effort and investment and to keep that end-to-end comprehensive approach made the difference.

I’m very proud to say that one of our customers, a large Blue plan on the East Coast that we’ve been working with for over three years now, is live with this kind of automated end-to-end solution on some of our platforms. Millions of members, thousands of customer service agents, eight to nine different source systems we’re dealing with. And we successfully cut down their average call handle time for benefit inquiries by more than 50%.

That’s a huge gain. Another big challenge for payer contact center operations is very high turnover and attrition in customer service teams, because the job is really frustrating. It takes a lot of time for agents to learn the lingo, understand the processes, and then deal with messy data to handle calls. Attrition rates in payer call centers run 30% to 50%. When you onboard a new agent, they take anywhere from three to four months to reach a level of proficiency where they can effectively handle complex inquiries. With the solutions we’ve implemented, we’re able to cut that three to four months of training time down to three to four weeks.

John Lynn 20:06

Wow, that’s a significant saving.

Ashish Desai 20:08

And life improves too, because with this solution you look at one screen, get the answer on one screen, and handle the inquiry. So there’s significant improvement there. We mentioned earlier the penalty and compliance implications, and with the high level of accuracy in the benefit quotes they generate, they have significant savings in those areas too. Very recently we’ve also enabled their digital channels, gradually.

It has been a phased approach. It took time, and over that period the solution has matured. Now, to onboard any new customer, phase one can be done in maybe three to four months and get the customer live, so speed to value has also increased significantly.

That Blue plan is more of a post-sale customer service inquiry success story. Another example is one of the top five payers in the country, for their Medicare Advantage business. We've helped them automate the shopping experience. A prospect goes to their portal and searches for benefits, and all the content and the ranking of the content is displayed on the screen through a lot of logic sitting in the back. The content gets populated by zip code and by the demographics of the person searching. What benefits do you want to present? What kind of plans do you want to present? What kind of benefit coverage do you want to showcase to them? Compare plans. That intelligence is coming from the back end through the Xperience1 system, which feeds into that portal.

With that solution in place, it's not only the portal. The system feeds 30 different systems that need the same information across the audience. The consumers are not only members. There are lots of internal consumers who need this accurate information. Another very good use case is care managers. At the point of care for any emergency care, if they want to quickly check the benefit coverage for a particular member and want to prescribe a procedure, having that available means they can make a real-time, immediate decision. In the absence of that, they have to go through a long circle of connections, connect the dots, and then make a decision. Care decisions can become much faster with this kind of comprehensive solution.

These platforms have been live with customers for a couple of years now, compared with some other initiatives that might be going on for years and are still not in production.

Jamie, if you'd like to add something.

Jamie Hernandez 22:51

I'm really glad you brought up pre- and post-enrollment, because I think that's something we should touch on. If you think about a member's journey and lifecycle with a health plan, it starts before they're even a member. They're a prospect shopping for a plan. Then they select the plan based on what's of interest to them, which could be their age, population, area, zip code, whatever that is. Maybe a gym membership is important, or better coverage, or lower costs.

They select that plan and then they’re still going through the system. Now they have the plan, and now they need to use their benefits. So when they go to a chatbot, a portal, or a mobile app, when they talk to a customer service representative, or when they go to the doctor and the doctor verifies their benefits, they’re expecting that what they bought is actually what they’re going to get.

Even if you continue through that journey, now the claim has been adjudicated and it's either paid or rejected. If it's rejected, you're expecting that the appeals and grievances department has the same benefit information that was given at the time about what their coverage is for that service, and also what was sold to them. So it really is about that full lifecycle. It's not just members calling to get information about benefits, or prospects trying to engage and become members. It's truly about their whole journey end to end, which, if you take a step back, involves a lot of people and departments.

John Lynn 24:34

I think that's the challenge too. The biggest challenge is getting all of those departments on the same page. We've had this problem for a long time with cloud, where any department could implement any solution because they could just go to the cloud and do it. Well, AI is making that even worse. Not only can they implement it in the cloud, but now they can use AI to do really sophisticated things, but within their silo. So that becomes a problem. It sounds like what you're espousing is coordinating the effort across enrollment versus member success and engagement as well. Fascinating how what's old is new, I guess.

Ashish Desai 25:18

It’s really surprising that we’ve come across a few payers who are trying to solve this problem through CRM systems, thinking they’re going to configure or customize Salesforce to handle this. Those systems are not designed to handle this kind of process. They’re designed for CRM, customer relationship management, more from the sales aspect. Benefit inquiries and those kinds of complex payer-specific processes require more tailored, purpose-built solutions.

That’s where realizing that data, management of data, and activation of data are so important led us to come out with this Context1 payer ontology platform, which I don’t think anybody in the industry currently does, to simplify the whole effort for payers to deal with data.

One of the customers I was talking about earlier, the Blue plan on the East Coast, is thinking about coming out with a mandate that providers should not be calling the contact center for routine inquiries and should self-serve instead. Think about a member visiting a provider and the provider needing to call a payer to get eligibility verification, prior authorization, and a variety of other things. It takes a longer time for them to provide care to the member.

John Lynn 26:45

Yeah. I think we’ve all sat in a waiting room waiting for that response to come. Your CRM example reminds me of one time I implemented an electronic medical record for a health center, and we tried to make it work for the counseling center too. It feels like a very similar thing. Did it work? Mostly. Did it work well? No. It's like you can make it work, but should you? And what are you missing out on when you don't use one that's customized to the needs of the organization? It sounds like a very similar example.

Ashish Desai 27:25

Absolutely. Another challenge when you talk about change management and messy data: many times through the implementation of these solutions, we’ve come across one of the big challenges. You go to the payer, pick a set of data, and ask who the owner is. There’s no single owner.

John Lynn 27:44

That’s a hard question.

Ashish Desai 27:46

For vendors like us, it takes time to get into the weeds, work closely with the payer stakeholders, and navigate this whole landscape.

John Lynn 27:55

Well, it’s interesting that your company is called Simplify Healthcare, because to simplify healthcare takes a lot of work to understand the governance. So I appreciate you both coming on to talk about some of the hard work you’re doing to simplify healthcare. Thanks so much to each of you for sharing.

And thanks, everyone, for watching and listening. If you want to find more great healthcare IT content like this, be sure to check it out at healthcareittoday.com or search for Healthcare IT Today on your favorite podcast application. Thanks, Ashish. Thanks, Jamie.

Ashish Desai 28:32

Thanks, John.

Jamie Hernandez 28:33

Thanks, John.