PegaWorld iNspire 2023: enGEN: Pega-powered Simplified Experiences for Healthcare

Consumers expect a seamless experience when it comes to their health. They want to receive care when they need it, free from confusing information, delays, or barriers. Far too often, that’s not the case – creating frustration not only for them, but for payer and provider employees as well. Join this session to see how enGEN is streamlining that experience by engaging the member/patient across all channels, with easy-to-understand information that breaks down the silos of operations to accelerate access to healthcare – all by working with Pega’s healthcare applications. Today, this platform is available not only for Highmark, but for a growing community of payers, providers, and member/patients.

Transcript:

- Good afternoon everyone. I'm Gloria Romeo. I'm the Senior Vice President of Products and Development at enGen, which is a subsidiary of Highmark Health. And with me is Taya Irizarry, who's our director over listening systems and advanced analytics at Highmark. I'm really excited to be here today to talk to you about what we're doing to make healthcare easier for our members and our clinicians and our employees. I've been working with Pega a long time, know Susan from a prior life, but before we talk about what we're doing, wanna talk a little bit about enGen. So as mentioned, we are a subsidiary of Highmark. We are headquartered in Pittsburgh, and we also have an office in Camp Hill, Pennsylvania. With the pandemic, we started working from anywhere. So we now have employees pretty much scattered all over the country, which has been great. We can attract talent without requiring people to move. We have about 6,000 employees across the US and we have two offices in India. We have a captive thrive there. A lot of our teams are extended. We've got scrum teams over in India, and we also do a lot of BPO work over there. We are the technology arm for Highmark. So when you think about the payer life cycle from quoting, enrollment, billing claims, provider and clinical, that is the scope of our world. But we also support a number of other blues plans across the country with the full platform. And then with other capabilities like print solutions. We process about 225 million claims a year, and we service about five million calls for both members and providers. So anyone that lives in the US knows how complicated healthcare is in this country, right? It's highly regulated, it's just very complex, more so than it should be and that it really needs to be. And so we saw this opportunity to really try and simplify engagement with our members, right? Anyone that's been sick and everyone's been sick at some time in their life, right? You wanna see your doctor, you wanna get healthy, you just wanna know what you have to do. You don't wanna have to think about your insurance. And you know, even worse, you don't wanna go to the doctor and get a surprise from your insurance that something's not covered or it's been denied, right? So we're really working to enable through technology, to remove that friction from the members. So our solution in this area is this multi-product approach. I'm gonna talk about Predictal, which is our new clinical platform. It covers CMUM, DM and wellness, and then Taya's gonna talk about how we're leveraging CDH to power the member engagement hub that we affectionately call MeHUB, and how we're tying it to our member listening system, which is a product that Taya and her team of data scientists have created so that we're getting the next best action to our members and to our clinicians so that people can engage better in their health. So a little bit about Predictal. I've been with enGen about four years, and I remember we started on this journey to create this clinical platform. We looked at some of the solutions in the market, but we have multiple payers that we support. Highmark also has a hospital system, so it's not just a payer, we like to call it pay-vider, payer and provider. So we had to scale, we had our nuances because of how we are structured, and we really wanted to tackle interoperability, right? Get away from files and really get to true interoperability with fire and streaming. And so we decided to build our own solution and we leveraged the Pega CM, Pega Healthcare to do so. I remember towards the end of 2019, we had our clients come in from the multiple plans that we support. We had boards up, we had the sticky notes, and we worked with them to talk about what was the most important to them, what was their North star? How did they wanna prioritize? And we started the design. We used human-centered design, we got a lot of people trained. We started working in agile fashion. We're still going through that as an organization, but the clinical team was the first to really get into that model. We're delivering code, sorry, twice a month. Every two weeks, we have 1500 automated test cases. So we've really changed how we're engaging and we're using that as a model across the organization. Much like I heard yesterday, regardless of industry, I think everyone, and it is saddled with old technology, bad business processes. And so we were working to change the way we were engaging there. So we started, we had our clients come in, COVID happened, but we continued to engage with them to make sure that this wasn't a technology driven product, that we were really listening to our customers, especially the clinicians. And so we have communities of practices, we had all sorts of meetings and continue to work with them to refine the product. In the end of 2021 we launched our MVP and we started with utilization management, and then we added our auto auth hub, which has been really great to automate something that was a big pain point and fully, totally manual before that. And then last December we rolled on care management as well. So we've got the whole product out, it's up and running, we've got pharmacy left and we continue to refine. We've got a backlog. We're really working in an agile fashion with our business partners and tweaking the system as we see things and opportunities to continue to improve. So aside from the technology and the technology need to come up with a new solution, the way care management works has evolved over the last 20 years as well. There's more measuring of ROI, we've got virtual solutions. It's a whole different mindset, right? You're trying to really, from the payer side, get outcomes for your members. And in addition to that, we're also engaging with our providers differently, right? There's a lot of focus on value-based outcomes. That's been a hot topic for a number of years now. And so if you're gonna request that your providers get outcomes, you have to make life easier for them as well. And everyone talks about, and we talk about it, right? Better engagement for our members, but we also needed to focus on better engagement for our providers. So we had to make it easier for them to engage with us and for them to engage with their member patients, right? The auto auth process, very manual, very time consuming. Again, right? You don't wanna have that auth denied when you really need care. So thinking about the provider and the employee engagement, the users of the system and the members really created this holistic approach on how we would approach this system. So again, it's very outcome driven, business outcome driven. And I'll get into some of the statistics on where we've really seen those outcomes come to life at Highmark, this is just a snapshot of the dashboard. We built a full dashboard, business analytics on top of the product and our counterpart at Highmark, Dr. Endleshame talks about how every morning he sits in front of this reports, and I've heard him say with his cup of coffee, and he can see where there's opportunities for his team to do something differently. They can engage and make sure offs are processing properly, that people are focused on the right members. And by leveraging MLS, right, we can watch the clinicians and making sure that they're engaged, that they're doing the right things with the members, that they've got the right information. So we're really focused on the outcomes and tweaking based on these measurements, making sure that the system is really tuned, the business processes are really tuned so that we're getting those best outcomes. So what are some of the outcomes that we've seen, we've seen over 1% reduction in unnecessary emergency department usage, right? So this is when someone really shouldn't be going the emergency department, we're making sure they're not going back, they're not going back repeatedly. We've got an average net promoter score of 80 for our members, right? That's, I'll put a quote, Dr. always talks about that's the same as Disney. So we've gotten really good in a year, really good engagement and satisfaction. We've seen a big medical cost savings of 17 million annual reduction. And that's just looking at our fully insured business. It's not even looking at ASO. And that comes to about a dollar 72 PM PM. So per member, per month savings. And if you're in the payer space, that everything is measured by PMPM and we've gotten really good administrative cost savings, right? And there's a lot of pressure on all the plans to really watch your admin savings. And so this has been a really good way to tackle that. But in addition to getting the admin savings at the same time, while we brought down savings and we reduced some of the contracted staff that was working care management, we're actually getting 2.5 members more through the system in the period of time that we measured, right? So we've cut down costs, but we're also taking care of more patient members, right? And so what that results in, if more people are getting the care they need, right? We're getting better health outcomes, we're bringing down medical costs. So it all ties together, on the UM side, we've also had really good success. We've got over 500 users on the UM side and our business counterparts over in UM, they said they would measure success by seeing if within six months of migrating onto the new platform, that their team was just as productive as they were with the old system. We actually saw that not only did they get to parity, they actually got a 20% improvement over where they were on our old system. So again, we're getting efficiencies and we didn't increase staff, right? So we're getting throughput there. Coincidentally, as we were in the middle of the migration, CMS decided to audit us, wasn't part of our project plan, it just happened, the way Predictal is built, we can track every keystroke, everything is saved in the backend. And we were able to provide the data to CMS and the result of, I have heard, we are completely compliant in this migration. And we actually added some reporting that we were lacking in the old system. So along with our migration, where we were fully compliant, which is fantastic as well, we wanted to check throughput. And what we've seen, one of the really great things that we've done was at the auto auth hub, anyone that's had to wait for a prior, for an authorization before you can get care, it's frustrating. Sometimes they get denied. And that whole process is a really, it's a big abrasive point for providers. It's even being regulated now because it's so painful. We were able to put some NLP in front of it. We've completely automated our faxes and every, this was a hundred percent manual before we launched Predictal. And we're seeing a turnaround time increase by 50% in this space. And overall we leverage Pega surveys and as so the, we measure the feedback from the clinicians. I think it's a five question survey that they use. And we've seen the improvement. They are happy with the system. We've got 95% of the users giving us positive results, right? It's been relatively a short amount of time. It's very hard to get 100% of your population satisfied. 95% is pretty darn good. And on top, in addition to that, we moved a lot of capabilities to self-service. So there were a lot of things that required a call to the IT team to configure in this, in the old system, we've now provided self-service. So if our clients have IT folks or folks that on the business side that wanna do the configuration, they can do it on their own, right? So that also helps reduce their costs. They're not calling the IT department and they're obviously getting it done a little quicker 'cause they don't have to make that call. So these are real results from a very short time with Predictal. So why do we think we have such differentiated experience? As I mentioned we started with our customers at the table with us. So it's a clinical tool designed by clinicians, for clinicians, and we continue to sit with them and refine it as we need to. I mentioned the self-service configurations. It's a big deal for them to be able to go in and set up the auth rules, set up the business rules, set up the workflow that they need. Taya's gonna talk a little bit about the member listening system. We use it for a lot of different functions at Highmark. For Predictal, it is used specifically for ID and stratification. We really think that our ID and Strat solution is the best there is in the market because we take in additional factors that some of the competing tools don't use. So we get that full whole person view and we can really get the right care to them. We've got integrations with some of our vendors, MCG CHC, and we are making sure we're leveraging the best of their capabilities rather than doing it ourselves. And we're secure, we're high trusts certified, we're HIPAA compliant. So this is just a little diagram to show you how the system works, right? We've got our members and our providers engaging. All of our channels are connected. We've got it's component based. So we've got the auto auth hub and CM, they're very tightly integrated from a business perspective as the driver of the development here, Mark in the audience likes to say, we are tightly coupled from a business standpoint, but we're loosely coupled from a technology standpoint. And what that means is, you see, we have integrated tools and so out of the box, we've got the integrations with MLS, with change healthcare for complex care management, the clinical guidelines, and then we're working with Healthwise for content, but we also have patterns. So if our other clients have other vendors that they wanna to use, we've got the patterns to make those integrations easy. We're set up for fire NCPDP, and you know, we still have some of those old flat files, can't get away from them 100% yet. We've also got some specialty partners in pharmacy and post-acute care. So next, I'll show you, this is a data flow, this is mostly for the CM side. We also have a similar diagram for UM, but decide to just focus on the CM. So how it works, Taya's team takes in all this member health data and there's a lot of data out there. So her team cleans it, makes sure, validates it, makes sure it's accurate. They get information for all sorts of purposes. Again, the data that comes into MLS, that comes into Predictal from MLS is strictly for our D and Strat. So they find the members that need the care, right? They develop those cohorts and they get the programs, the risk results, here's the task, the risk score, the beauty of this, and then we take that, we load it up in the cloud and we get it into Predictal. The beauty of having MLS behind the scenes, it gives our clinicians the next best member for them to go to. In the past, they would just have a pool of members that they had to reach out to and they had to determine on their own who needed that care. MLS tells them this person over here needs attention most and this is what they need to do. So that's really where the power with MLS coming in is. And then from the UM side, what you're not seeing here is that if someone is discharged from the hospital, we send that information back to MLS. So for the post care, right? So we've got the full life cycle here. We've got files going out, as you saw on the other side to the various channels. And we've also got streaming and Kafka in there. All right, so I'm gonna, in a minute, Taya's gonna come up and talk to you about how we're leveraging CDH to power the member engagement hub and how we're connecting it to MLS. And that's really when we're talking about the whole engagement for our members. I talked about Predictal from the clinical side, but the way I think about Predictal, it's just another channel for our members to engage. It might be indirect, right? Because our clinicians are using it, their providers are submitting the offs, but those are also channels too. And if you think about who's gonna have the most impact on the member, the member that needs the care, right? It's those clinicians, it's those providers. You don't wanna think about your insurance, you may not trust us as much. You're gonna trust your nurses, especially, you're gonna trust your doctors, right? So having the same consistent information through all the channels, Predictal being one of those channels, really makes that omnichannel experience come to life. And I'm gonna stop talking about it and leave it to Taya, keep going.

- Excellent, thank you. Just a little introduction to me and my background. So Taya Irizazrry, nurse by training and then continued on to work in a PhD that was specifically focused on longitudinal data. So data over time and correlated data over time. And really specifically for disease management. So chronic disease management, things that go on for a long time. And when we know when things go on for a long time, there's a lot of data associated with it. There's a lot of correlated data, lots of different data points. And then within that chronic period, there's also acute instances that pop up and down. So I'll be starting by more of a conceptual model of what member listening system is from a strategic standpoint. And then moving on to understanding how that translates into a data model and what that data model can do sitting as an accelerator under tools like Customer Decision hub, and also for powering tools like Predictal. But first member listening system. We call it a member listening system because we've started focused on members of a health plan, but we are expanding into listening systems and we listen to all sorts of things. So we might listen to information about members, but we can also listen to information about providers or about customer service experiences. So what we're really listening to is patterns and signals that we find in all of the data. And when we say all of the data, it really is coming in from any system that Highmark Health as our first customer moving on through now Wyoming and North Dakota, the data we have available through all of their systems comes together to create meaning, to create that picture and that data model. So what we have today is really disparate systems that end up impacting members specifically from all different angles. And when we say vendors, often a health plan works with a number of different vendor solutions to also drive care such as diabetes management. And so today, in most systems, each and every one of those is its own silo, members can get up to when they leave the hospital, 37 phone calls in the first week from different groups that are really there desperately trying to support a member's care and transition from hospital to home. But it really isn't organized and it's definitely coming from everywhere. And oftentimes those channels don't even know about each other. So what we're looking to do, if you see the middle space between the clinical operations sales and then the as engagement sources. So those are the data systems that we would access. And then that member engagement hub sits in there. Member listening system is an engagement accelerator. So we are the ones that are really taking all of that data and making meaning so that the model can then be understood by CDH, by that member engagement hub. And then that member engagement hub can then help with outbound and inbound channel orchestration. So that it feels like my experience on my website, the experience that the clinician who's calling me for clinical services and the customer service rep that I might be calling into, that feels like one experience. And that the customer rep, the case manager also has the same information so that they can make that experience for the member patient actually feel connected because they have that connected information. And in today, with all of the way that the system is set up, it's often four different systems that a customer service rep would look at, and clinical services may not even have access to it or feel confident in what to do or what to say about it. So there's often what they call warm handoffs between different areas, which creates a real burden on the member experience as well as just real conflict between who's serving who and when. So as I had mentioned, listening system specifically for Predictal is about ID and strat, identification and stratification for members for clinical services outreach. But member listening system expands beyond that to also identify clinical relevance for populations that might need vendor solutions like diabetes management, specialty behavioral healthcare, telemedicine as an extension of their benefit package. So really expanding to anything a health plan member might be experiencing or a patient within a healthcare system where they're interacting with providers and health insurance. So this is really a me slide, but what we'll call our horizontal themes when we think about what is our strategy for designing a data model that can really power or accelerate something like CDH, the horizontals are defined here. So giving you a little time to see what are the things that we think matter most for our member population? What are those moments in time or those journeys where we think that we could be most helpful with giving the right information at the right time? So creating those MBAs, right? And that we're thinking about the data elements across all of the data in all systems that we would end up capturing in order to create or identify, listen to these moments in time. So across all of these horizontal themes, right? We have these vertical situations. So an example of where these intersect for unplanned care. For unplanned care, I may want to help the member engage in their virtual solutions instead of going to a brick and mortar emergency room. So I know that this member is going onto the member portal to understand a possible benefit, or maybe they use our symptom checker. And I wanna be able to listen to that moment and say, in this situation, given your benefit package and what we know about you from the past, it would be best for you not to go to the emergency room and for you to check this out first. And then if you think about longitudinal care, how many times has this person gone to the emergency room for unplanned care? Maybe actually this person needs coordination and follow up orchestration because they've been to the emergency room six times. So across all of these verticals and where they hit the horizontals is where we create those next best actions and in the data model at the base, and then placing what we believe are the interaction points to be designed within a system like My Hub or CDHs. So how it works, there's what we call membership data that would include, what are your benefits? What are you actually capable of having because of your health insurance? Then information that we can get from customer service of what your experiences are there, really the output of those interactions, as well as the clinical services platform. What is happening for you and what is the member status? Has the member targeted, are they engaged? Have they enrolled, have they declined? Should they be enrolled? And they haven't yet. So time in relationship to member status is something that we're following in creating that data model with the layers of time mixed in. So that really, that direct integration with the eh HS platform that gets us direct authorizations, direct claims. So we're understanding when those moments are happening and within what part of the cycle of an authorization or the cycle of a claim do we really care about? Are we listening in on, so from the MeHUB perspective, it's creating that data model so that the decisioning engine on top has all the bits and pieces across every single one of those interaction moments to then create MBAs that are actually orchestrated with one another. If a member is just really out of the hospital and we know that transition of care is a moment in time that's very sensitive, not only do we wanna send very specific things to them, but we also wanna suppress a lot of other things. And so sometimes we talk about next best action as the action we want the person to take. But the other piece that's even more powerful, at least in a system where there's so much going on, is really reducing the noise and increasing the signal on the things we know matter most, especially in those more vulnerable moments. And that's really what that decision engine lets us do once you have all the right pieces there to put together for the puzzle. And then the distribution engine is really about how do we really think that this information needs to be displayed for those members, given the channel's preferences they have or given where they are in really the health ecosystem if they're a portal user, right? We know that a lot of people are and then a lot of people aren't. So we wanna make sure for all the different types of users, the way that they get the information is the way that we're actually able to also give them information. So final slide is really something I think, Gloria, that you wanna speak to about how Member Engagement Hub really fits with all the applications in Pega.

- Yeah. So this is for all the technical folks in the audience. Just wanted to show you how we created this ecosystem at Highmark. And you can see the layer cake in play here, right? So we've got the Pega Foundation and then we added our own magic to the foundation for healthcare. We built an enterprise Pega layer that connects our customer service platform, which is also built on Pega. And we unfortunately didn't have time to really get into that today and Predictal. And so we are making sure we've got everything connected. And then with CDH in the middle of it, we've got our products, we've got, as you can see, our inbound and outbound channels. So our Predictal and customer service, as I mentioned earlier, I think of them as not only as products but also as channels for our members. Maybe a little indirectly, but they are channels where they're getting information. And then Taya talked about where we're distributing the information, meeting the members where they are and getting the information to them in the best way that they're gonna consume it. And I think that's it.

- Yep, we're open for questions.

- [Speaker] Thank you so much, that's a very nice presentation. My name is Aku from Health, I have few questions. Number one is on the previous slide, I think you have the suppression so that the member will not get too many pounding with email or text. So I'm curious on how do you manage their profile, like the subscription and how do you manage so that they will not get repetitive for too many interactions. So that's number one. And number two, what kind of product you use for the outbound channel for sending the email, text and other things. Thank you.

- Right, so the first one about suppression.

- [Speaker] Yes.

- And that can really be set, within CDH tool, there is like how many interactions do we really think this member needs per channel? So those things can be manually adjusted as the first level of design of how many things I actually think are supposed to go out. And then using a weighting with a propensity scoring and understanding what business impact is, then you create more of the little rules about how many things should go out and which ones matter most for an instance. So this context I gave was transitions of care and I know when this member is going from hospital or home or from facility to facility, these are the rules that are gonna matter. And those rules are then set within the larger scheme of everything else and they play out when that member has hit an authorization for a hospital stay. So my rule says when a member has a hospital stay, then this rule set happens and I know this is the number of things that the member is gonna receive and these are the top out of, I want them to receive five things this week and these are gonna be the top five.

- Okay, so it makes sense. So now you have the potential like the campaign or email or whatever to be sent. Then do you use pick a decision hub to also send the communication interaction to that customers or you use a different products for the outbound outbound channels?

- So you can partner customer decision hub with any kind of other vendor that does the actual sending. So if I'm the designer in the tool, I can say yes, deploy these things right by channel email. And then there is an integration with Relay is the company that Highmark uses. So today if I wanted to send a text message, that integration would be between CDH and Relay to get that information to that channel, right? Or you can connect it to other CRM tools that can then do the channel extension, right? But the CDH piece can be like the brain and then those integration points are there so that it can push information into those other systems for actions.

- Okay, makes sense, yeah, thank you so much.

- [Carla] So I haven't used a microphone in a long time. Alright, I'm Carla Daly from Blue Shield of California. And couple questions, just really high level, this is awesome. How long did it take you to put this together and create this, I guess ecosystem is what you were calling it, of product offering? And then more tactical question. Can you explain a little bit more on the closing the loops of the next best actions? Like how do you know when someone's actually taken that action and then brought it back in so you don't offer it again or things like that?

- I'll tackle the first one.

- Yeah, you go first.

- So like I said, I have been at enGen since early 2019. I think the work got started in earnest probably in 2020. So we're in 23, so two and a half years. And as I mentioned, we did it iteratively, we rolled out UM and the auto auth hub and CM went up in December of last year. So we really worked closely with the business to put it out in a way that made sense. We paused a little to make some corrections on some pieces. That's why pharmacy's coming out now. But yeah, not super long in terms of IT projects.

- So the question, remind me again, sorry.

- The closed loop.

- Oh, the closed loop, yeah, okay. So when I talk about a data model, it's really models within models. So the first part is about who is eligible for a thing and within that eligibility, there might also be exclusions for specific reasons like I'm eligible for these diabetes solutions, but if I have cancer of X, Y, and Z types, I don't really want that diabetes solution to touch me because I'm actually being managed by my oncology specialist. So within each of those groups, we care about who's eligible or why they might not be. So that's one kind of set of data model. And then the partner to that data model is what we call our response model. So for every action there is a reaction and those responses have to be listened to as well. So we listen through all of the claims authorization, member profile information, anything about the experiences about these humans to decide who's eligible or not. And those responses are a piece of that model. So for any trigger we often call it for a member, now we know because this member has been triggered for this thing, now I'm listening for the response. And once I get that response, what is the next action that actually matters or doesn't? And what is most tricky, I mean there's so many things that are tricky about healthcare is that it's not just, I expect this person to take this banking offer for a credit card, bam, they did it or they didn't. There's a lot of things where this member's eligibly eligible and enrolled and now I need to actually watch pattern of is the member doing the thing? And this is where that chronic care management piece comes in because there's often waves of, am I actually engaged in this or am I not? And so listening to how well is a person actually engaging with the thing is a type of member status. If the member falls off, maybe they're supposed to be doing something and then they stop doing it, that becomes a moment in time with a member status that we can then react to. So within the Predictal system, we are listening to those member statuses of who's open, who's triggered, who's open, who's enrolled in near real time, so that we can use that information to decide what's gonna happen with all the other kinds of often called offerings in other marketing spaces, but all the other interventions that may or may not actually be impacted because that member is actively enrolled or declined.

- [Speaker 2] I guys, you talked a lot, a lot of this is around the care of the members. Have we explored, I mean, we've heard other presentations where sales and service, you have all these different kind of, you have all these different competing priorities that CDH can weigh and that determines what that next best action would be. Are you guys just using it for care? Is there thoughts of expansion around? And then how many actions do you actually have, like the action library and how often are you able to create new actions within CDH? So a couple questions if that's all right?

- Yeah, so right now the focus has been mostly on clinical and we're in the design space and kind of building out what that looks like to connect it to customer service. And the first wave of that is to make sure that customer service understands what's actually happening in the clinical space. And the reason we're starting there is because members are getting outbound phone calls from strangers, they don't know who it is. I mean we get so many different outbound phone calls, how are we able to trust it? So members will call into customer service asking about the calls that they're getting from providers. We wanna make that visible for customer service so that they can easily create that trusted relationship about what's happening in clinical space, in the customer service space and vice versa. Our clinicians often don't know about the seven different offerings that are happening through marketing campaigns. And we wanna make sure that if a member is getting contacted for a clinical reason, that they can also speak to all of those other kinds of offerings that are being addressed either in customer service or in the marketing space. So yes, it's on the roadmap, but really stabilizing what we're doing with Predictal in the clinical space with step one.

- And then I know we're almost done, but like how fast are you able, now that you've got this library, how fast are you able to create a new action or treatment and add to that library, would you say?

- Right, so again, we're very focused on the clinical space. If you were to say, I have 175 different, we're talking about the horizontals and the verticals, I might have 75, actually I do have 125 different points of those interactions. But then across those interaction points, when you add in member status and what happens at each member status, then you see how it's expanded exponentially, right? So I can say I have 175, but then it ends up expanding into the if and then statements across the way. Once you create the base library, which is what we have now, we have, those are considered standard features and then you can create those on the fly. They're really that simple. And you can actually do it without having to wait for a complete production cycle to actually put them into production. So then that happens, I don't have to worry about is my production cycle three weeks or six weeks or whatever. There's that level of fluidity, but then there's custom ones and we don't have that built into the data model. And so that's on a six week cycle of this is new, there's maybe a new data source, maybe we're getting a new health risk assessment from a different location, which is happening now and it will take us six weeks to map. Where does that data exist? How do we bring it into the current model? Is it an enhancement of what we're already doing or is it something completely new? And thinking about how we integrate that with everything else.