PegaWorld 2025: Navy Federal Credit Union: Unleashing the Power of Process Discovery and RPA Innovation

PegaWorld 2025: Navy Federal Credit Union: Unleashing the Power of Process Discovery and RPA Innovation

Hey, everybody. How's it going? Welcome. Welcome to our session. My name is Brian Kelleher. I'm the senior director of product for process and task mining at Pega, and I'm delighted to co-host this session with Becky Blackwell, the senior director of product for Robotic Process Automation.

Today, we're going to talk about how process discovery and automation work together at Navy Federal. And we're going to jump into how WFP is used there to identify opportunities for improvement, whether through automation or process improvement. We're going to talk about how robotics projects are ingested, evaluated and approved at Navy Federal. And we're going to take a little look into the future and see how AI is going to be changing the way that they they work every day.

We are glad to be joined by two esteemed speakers here. We have Kenneth Hearn, who is the VP of Digital Process Automation at Navy Federal. He's going to be talking about how WFP and RPA work together to drive productivity and automation improvements, and then Gina Aman, the enterprise process business process analyst at Navy Federal, who's going to talk about how they use WFP to crowdsource process improvement opportunities. So let's dive in and start start learning something about process discovery with robotics and in WFP.

Before we get into the details of WFP and robotic process automation, Kenneth, could you talk a little bit about what Navy Federal is and what you all are all about?

So we're a non-for-profit, not for profit credit union that was formed in 1933 by just seven members. We've since grown to 14 million members, all part of the military community. So our mission is to serve our members. Our members are the mission. So we're looking to be able to create financial wellness and opportunities for them and education. And then we look to automate our processes so that we can be able to provide them better products and experiences.

I love Navy Federal. You guys are a great organization. You're always member first, and I think that that's a really great thing. But Kenneth, as the EVP of Digital Process Automation, what are your primary responsibilities and how do they align with the organization's strategic goals?

So I get the opportunity to be able to lead a team that gets to provide digitization process, digitization and automation for our entire enterprise. So we were able to establish our center of excellence and set up the governance for all the development that happens on Pega Platform Pega Robotics. We've since added on the operations and support side of that, and then we have a large delivery arm as well.

That's great. So it sounds like you all have quite the quite the crew working on robotic process automation there. That's that's fantastic. So Gina, as the enterprise business analyst at, at Navy Federal, tell us a little bit about your roles and responsibilities and how they align to to Navy Federal strategy.

Yeah. So as a process improvement analyst, my job is to drive continuous improvement. So at the enterprise level I'm working with business units from contact center to the president's office, really looking at their current state processes to identify where their inefficiencies, opportunities to either automate, implement, process improvement or both.

Um, so another big part of what I do is actually Wi-Fi related. So with those process improvement efforts, we like to leverage Wi-Fi to give us some of that current state data and understand, um, not only what the process is currently taking, but what some of the impact of these automation and process improvement efforts might be. So I've served as the product owner for Wi-Fi for the last several years, and, um, really helped stand up, a program where we train analysts across the organization in how to leverage the tool on their efforts.

So how that ties back to Navy Federal mission is, um, you know, we have several enterprise strategic goals, one being employee powered. So we want our processes and our technologies to empower our employees. So my job, again, is to make those the most efficient as possible and leveraging technology like Wi-Fi to do that.

You're almost like a coach, right? Making sure your fellow employees are on the right track and making sure that the technology is working for you guys. That said, Kenneth, can you give us an overview of how you guys are using WFP and RPA together to improve productivity and automation?

So when we start our processes using WFP, we're able to identify manual tasks, repetitive tasks that users are performing, and those are the opportunities to be able to automate. It also gives us the data to be able to validate that it's a project that's worth taking on building automation for. From there, we can take that into robotics, using robots to help assist the users on their desktop to be able to perform those tasks tasks faster, or moving to unattended robots to be able to sit there and take that on as a service that we kind of set it up. So it's more like an API call. So the unattended robot can interact with the legacy application inside of a larger workflow.

From there, we're able to actually use Wi-Fi to then validate the adoption of the automations that we're building, making sure that they're actually using it, and that we're getting the time savings that we were expecting to begin with.

That's great. I love the fact that you all are using Wi-Fi to validate the automations afterwards, to make sure that they're effective, because I'm sure sometimes they may not work the way that you anticipated, and you all must have to work together a lot to to make this successful. How do you both collaborate in order to help make sure that Navy Federal achieves its goals?

Yeah, so pretty much everything that we do at Navy Federal is going to be collaborative, especially when it comes to process improvement and automation. We like to have, you know, process owners, system owners, the Pi team and the dev team work together so that we're delivering the best solution in the most efficient way possible.

Um, a lot of what my team can do is help bridge the gap for some of those business units. So in the past we've worked with like Contact Center, for example, um, looking at the business membership application process and where they may be the process owners, the digital team is actually the system owners. And so we can help bridge some of those gaps and helping build some of those cases for automation with, again, Wi- Fi metrics showing duration, showing waste percentages, some of those more friction level, uh, tasks or activities like cut, copy paste, stuff like that.

Gina's team interacting with the business areas also gives them time to educate them on the art of what's possible. A lot of times, they don't know the technology that's available to be able to improve their processes.

That's great. Gene, can you talk more on how Navy Federal is leveraging task mining to federate process discovery?

Yeah, so we stood up our Wi-Fi program in probably around 2021. And as part of that, we really had some key things in mind around enabling the business unit, enterprise governance and scalability. And so with that, we train analysts, like I mentioned in both the user and the admin role, where our users are those analysts that are going out there doing that process analysis, they're doing that use case delivery. And then we have our admin level where they kind of become that tier one support. And then my team are the enterprise admins that are then that tier two support.

Another piece to it is we do have internal lean Six Sigma Green belt trainings for my PaaS out there. Um, and so in that training, we really do also like to train on how to leverage task mining on your process improvement efforts. Again, whether that's understanding what the current state is or maybe it's measuring the impact of the changes that we've made.

That's fantastic. And you all have been at this for quite a while. And I'm sure, you know, with getting this out to everybody at Navy Federal, I'm sure you've run into a lot of different challenges and and things throughout the year. So could you maybe tell the audience a little bit about some of the challenges that you've encountered during this time and how you've how you've overcome them?

Yeah. So I'll say, when it comes to new deployments specifically, sometimes there is that hesitancy of being watched, right? That kind of big brother aspect to it. And so that is why we do play such a heavy governance role in the tool, because we don't want it used for performance evaluation. We want it to be used to understand processes. So as part of that, we really encourage transparency from the beginning. We do let the various business unit leaders manage that communication how they want with their teams. But again, we're encouraging that you are speaking with the end users. We're not doing this behind the scenes. We're doing this for you. We want to leverage task mining to understand how to make your processes better. Make your work life better.

Another thing is also, again, with new business units starting out, sometimes it can be overwhelming. There's a whole lot of different data available. And so kind of shifting that business mindset from what can I find to a more targeted focus, kind of clearly defining a problem statement or a goal within Wi-Fi so that they, again, can kind of narrow in on specifics versus trying to, you know, search or search for what they don't know. So we're really challenging what's the or the business value that this is providing. What type of questions are you answering with this type of data.

That's really great. Okay, Kenneth, can you tell us about a successful RPA project or initiative that you've led and its impact?

So some of our robotic processes can be short lived by design. So oftentimes we're able to get the speed to market by building something using robotics that you can't through other technologies, but ultimately you'll end up having the long term solution come into place and you can retire the robots.

The example that I'm thinking of is part of our loan origination loan application system. So we had a large influx of applications coming in, and as some of them were getting withdrawn early on in the process, there was a disconnect between the origination system and the application system, and there was a manual process that the user was going to have to do and cancel out and close out that application. So we were able to quickly build a robot that was able to go in there and take on that manual task on a large volume of work for the business area, and then around six months later or so, when they finally got the integration working between the final systems, they were able to retire that robot. But during that time, that robot saved a massive amount of time for that business area.

That's great. Yeah, I love I love that you're focusing in on quick wins to get the savings right out of the gate and then kind of thinking about the long term solution with, with you all kind of spreading out the the process improvement out throughout the organization. You must get a lot of requests for for robotics and for for new automation opportunities and improvement opportunities. Could you tell us a little bit about the process on how you all take those in and evaluate them, and then then approve them afterwards?

So there are a couple of different ways to initiate a project with inside of Navy Federal, but they're all going to go through the similar steps of review. Make sure that you're looking at the business case, determining the correct technology to leverage on them, and make sure that you're going to get value out of the automation that you're building. So for the robotics ones, though, a lot of those are initiated by the business area. Business is looking to be able to bring that on through system development and bring on the project themselves. And so that's where our Center of excellence comes in, making sure that we're working with them to make sure it's the appropriate technology. And then given the guidance that they need to bringing it to bringing it through to production.

Cool. Yeah. I think it's great that your Coe is really focused on making the business successful and bringing the business, identifying the right use cases and making the business successful on that. What strategies did you guys employ to ensure successful adoption and integration of all of these technologies across the org?

So with the Center of Excellence, we're able to help drive adoption of the technology, showing the success that we're having with inside the projects. So for every project that we build with inside of Pega Platform or Pega Robotics, we're looking to track the ROI for that. A lot of that goes back to the time saved versus to perform the manual process. So we're tracking those numbers and actually have a dashboard to be able to show the savings that we get out of all of them. That tends to drive more customers coming back to our business area and to our program.

Very cool. Being able to track that is pretty, pretty impressive. So, Gina, could you tell us a little bit about some of the successful projects that you've had with Task Mining with Wi-Fi?

Yeah, so I'm going to go back to my business membership example because we had a couple of different automation efforts on that. So high level, if you want to be a business member at Navy Federal, you have to go through a separate application process. And that used to start with the member going online, completing a pre-application and then calling into the contact center. And so for the msrs the people taking these phone calls, they would have to then manually move data between these systems. And so we wanted to be able to automate that for these msrs, especially when we first started this analysis, it was right after the onset of Covid P-p-p loans were crazy. The volume was just, you know, very hard to keep up with.

And so we leveraged Wi-Fi. We found workflows to understand how long that process was taking, and leverage those metrics to build a business case for the digital team to go in and deliver an automated solution. So with that one, again, we captured current state. We were able to get an automated solution delivered. And then again leverage Wi-Fi to understand the impact of that. That was a great success story. We did see where they were able to replace that entire manual process. They dropped that average duration from like 12.5 minutes down to five minutes. The waste percentage decrease where they were doing less Non-value added tasks during that process, and they had an annualized savings of just over 1.4 million. So great success story.

Same process. We looked at a separate section of the business membership application process, where they were moving manual data, again from a system generated word document into the system for basically to finalize the business profile creation. And with that one, they built a robotic solution because no APIs existed to send the information between the systems. And what we found was once that was implemented again, use Wi-Fi and the average duration had actually increased. And we were seeing a split in the data where half of our users were adopting the new process and leveraging the automation, and the other half were still doing it the manual way. So we had that problem.

The other issue was that with the users that had adopted the robotic solution. We were seeing significant latency, and we were able to quantify that with WFP and take it back to the dev team and say, hey, what's going on? We're seeing a bunch of latency with this solution. Thankfully, it ended up being a quick fix. It was an issue with the VDI and the browser settings, so we could just go into the Pega adapter, make that change, and then again be able to see where that latency had completely gone down. And then over time, with change management, how the users continued to adopt that process.

So one thing with WFP is that sometimes it's going to give you that that happy story of here's the savings. That's exactly what we expected. And other times it's going to tell us where we missed the mark and what we can probably do to make it better. It's still good in the end that you can figure that out, right? It's really a continuous improvement cycle that you're describing, right? You use WFP, find the problem, use the RPA bot, fix the problem, use W5, measure the fix. It's all really cool.

Um, okay. At this point, what I want to share with you guys is a video of some up and coming capabilities we have. It's some new ways that we have to generate robotic automations. So if you could take a look at this.

[Video Demonstration Begins] I am excited to share a new Pega Robotics tool called the Task Modeler. Use the task modeler to better understand workflows and legacy applications, and when there isn't an API available, use it to create new robotic automations to do that workflow for you.

The task Modeler is a lightweight utility that records actions as you perform them in legacy apps. That's what's happening on the screen right now. And as you can see, there is no performance degradation to those applications. I'm recording actions in a windows desktop app called Banker Insight and You plus Ops, which is a web application running in Chrome. The use case is to collect a credit card account number out of Banker Insight, and use it to search for a Fico score in U+.

Now let's talk about what's going on under the hood. Good. The test modeler is looking at application events for everything happening on your computer. It identifies controls and applications and the actions that you are taking on those controls. All the actions that the task modeler sees are recorded as raw events, and some are meaningful events like typing into a text box or clicking a button, but some are not needed for automation, like a tooltip showing up or a hover over a menu item.

Once you are done recording the model, we use an AI agent to analyze the events and pick out the meaningful ones. This is the foundation for your robotic automation. The analysis shown here includes the suggested automation inputs of username, password and customer number, and Fico score as the suggested output for steps. It includes logging into both applications and navigating the web app to the customer details screen, then searching for a customer, finding their credit card account number, and finally using that in Uplus to extract the Fico score for further use.

From here, the model is uploaded into Robot Manager, where it is stored in the Task model catalog. Then robot Manager uses an AI agent to turn the analysis into automations that can be added to a new or existing automation project.

Another great way to create a task model is from Pega process and task mining. For example, here we are seeing an opportunity showing excessive manual data movement. When I drill into the timeline, I can view the analysis that was collected as a task model and see the copy of data from Banker Insight, and then the paste into U+. And directly from this screen I can submit this to the Task Model catalog along with the manual process we had just recorded.

Now we are in Robot Studio. It's where we create attended and unattended bots. This view of the task model catalog shows the two recordings we just uploaded, and they are ready for import into a Robot Studio automation project. Let's work with the model that I recorded at the beginning of the video. I'll pick up the other one later.

And here's the fully imported project. The AI agents divided the workflow into six separate and reusable automations like login and customer search. These automations use our best practices and are complete with exception handling and proper error messages. This would have taken a skilled automation developer a really long time to do, and it all happened in just minutes and was completely automated.

Now we are ready for our end to end run and it's lightning fast, just like it will be when run as a production bot. The test modeler is a new Pega design agent that really speeds up the process of generating automations and makes it super simple and recording the model doesn't require Robot Studio, Robot runtime, or even a special browser extension. It can be completed by anyone that understands the business process.

And when you are ready to invoke the bots we just created from a Pega application, use the robotics smart shape for an unattended bot. You can also use a data page or a pre or post-processing flow action step for an unattended bot, and the end result is bots that are generated through a task model using AI agents that will run in production the same way every time and lean on our existing infrastructure for security, repeatability, scale, governance, compliance and auditability all things that an enterprise RPA client demands.

[Video Demonstration Ends] Okay, cool. So, Gina, what do you think about this? How do you think that this is going to shape the way that you guys develop automations and do process improvement in the future?

Yeah. So I see, as you know, the task mining capabilities in AI continues to expand. I see a lot less time being spent in that current state analysis, that root cause analysis and repurposing that time to actually drive implementation.

Um, I'll say that there are several business units that, you know, they have a lack of existing process data. It's hard to get to or it's impossible to get to. So they rely on those traditional manual time studies where, you know, they're capturing how maybe a handful of users are doing the process. They're definitely not getting all the different process variation that's happening. They're not getting that full picture. And so with tools like this, I see again where it's bubbling up those most valuable insights for you a lot quicker and saving time in that space.

That's great. And being able to see those insights come, come right out of the gate has got to be invaluable for you, because I know there's several steps after you do the analysis to to really understand. So, Kenneth, from from your perspective, how do you see the the future with AI at design time changing the way that you guys build robots and do improvements?

So I Navy Federal have a few guiding principles around AI adoption in general, making sure that we are maintaining security and respecting privacy. Um, being making sure that everything is transparent and inclusive. Um, commitment to fairness along with making sure that everything is accountable. And most importantly, we're looking for human, uh, human augmentation, not replacement.

So what we've been doing for years, helping our business areas, being able to automate their processes. After watching that, it's like that's the same type of automation we can now bring into our development process. At design time, we're able to take those tools. And if you've had experience before with robotics, the time it takes to sit there and get a recording, watch a user go through step by step of the process to be able to have Wi-Fi either capture that for you or task modeler to be able to sit there and record that and giving you that automation without the time to do the step by step interrogation, it's a great opportunity to be able to speed our delivery up so that we can be able to get the products out there even faster.

Yeah. That all makes perfect sense. So for both of you, how do you guys handle resistance to change in the org when introducing new technology like this? So I'll start with that one. Uh, for me, my experience is getting someone to champion the change. There are sometimes areas that are kind of resistance to adopting a new technology, new change in process, but getting a single person to start utilizing it, showing the difference inside their KPIs and their performance, and be able to show that to others. That usually encourages other people to be like, see the power of it, and they have the desire to be able to adapt that technology and take it on themselves as well.

Yeah, and I'll say getting leadership buy in is critical. And I know that oftentimes that is easier said than done. But, you know, being able to provide various examples to leaders of how it's been used at Navy Federal, showing them so that they can actually apply it to their own business area and understand, you know, kind of the what's in it for me, especially knowing that with, you know, automation specifically, oftentimes you're going to need this kind of data to get that work prioritized. So if you want to get work done faster, maybe try leveraging tools like this.

Yeah. What's in it for me is critical to any change you're doing right? Absolutely. So before we turn it over to the audience for for some questions, there's probably different levels of engagement here from where people are and their journeys with improvement, whether it's they're just starting out or maybe they've been doing this for a while, like like you all have. What are some lessons learned that you, you all could share with the audience that would help them along their journey?

So for me, it's don't be afraid to fail. If you've never used a task mining tool before, it can be very overwhelming to start. But for me, specifically, that constant experimentation, figuring out what does and doesn't work, figuring out what's a better candidate for this versus what's not. Because sometimes Wi-Fi is not the right tool for the job, and that's fine. But as a process improvement analyst, I'm always going to challenge not only how the work is being done, but how my own work is being done, and leveraging technology like task and process mining to improve my own processes.

And I would say, don't be afraid to challenge the business process that you're going to automate. Oftentimes, the way you can do something through automation is different than the way a user can be able to do it. Doing things inside of parallel multitasking in ways with robotics is much different than the way people perform those business processes.

Excellent. Well, Gina and Kenneth, it's been an absolute pleasure working with you all through this whole process. I know I've learned a lot. I'm sure Becky has as well. Um, thank you all for coming up and offering to share your insights. And I know hopefully the audience got something from that. If anybody in the audience would like to come and learn more about what you saw on the video or from from Gina and Kenneth here, please come out to the Innovation Hub and check out the process and task Mining boost. Check out the RPA boost. There's lots of great demos out there as well. So now we'll turn it over for questions. If you do have questions, please walk up to one of the mics. Um, so there's any questions out there?

[Q&A Session Begins] Yes, sir. Please. Uh, I actually yesterday maybe this is a so just a simple thing like how did you manage the test data, the security. Because we are talking about unattended bot things happening behind the scenes. And if the unattended bot is handling the financial transactions or something. We have those use cases as well. But did you encounter any situation like in a normal life, a bot unattended bot processing a check and it redundantly processed same check twice to a customer, and at the back end it is transforming the data on a legacy system like mainframes or something. Maybe different systems for you, but some transactions are right at the moment, right? Did you encountered anything like that one?

So we've had a couple of things we look into on that. And you talked about test data at the beginning, I believe. So part of our intake process when we're doing assessment to see the fit case use case for automating something is checking to see do they have test data. Is there a lower environments for these applications that they can be building the automations against? That gives them the opportunity to be able to refine that automation, make sure that they're not going to have those types of bugs going on inside of production.

Another strategy that we help people apply when they're looking into robotics is sometimes we build it as a attended robot first, even though knowing that long term we're going to have it unattended. So if you build it attended, then now you're able to get the efficiencies to be able to help the user out, but they're able to go and run that workflow and validate that it is running successfully before we move it over and running it on it to.

Yeah, okay. Thanks. Thank you. What other questions are out there? No other ones. No other questions. Oh. I guess the questions for Gina. So you mentioned that you don't want people just to go to Wi-Fi and keep looking for things, so they should have some purpose or objective problem. How do you define that? So how does does that come to be. So you have something to pursue. And how do you enable people to use Wi-Fi to solve that problem.

Yeah. So sometimes um, most times we're working with process improvement teams. Um, and so I always challenge them. What are you currently working on? What do you currently need metrics for that you don't have? Um, and so that's kind of the first area. The other piece though is that sometimes we've got operations or different areas that don't have a background in process improvement. And so from the operations side specifically, it's a lot easier to say, okay, what are your pain points where. What are you hearing from your end users?

Um, oftentimes it's like, you know, manual cut, copy paste type work. Those are usually easy, easy opportunities for us to identify. Um, but sometimes it does require more hands on work for my team to kind of go in and explain exactly like what the tool can do, because like you mentioned, there's, you know, they want to come in and look at everything. And, um, sometimes there are different views within the tool that aren't going to at all apply to what they're trying to do. Um, so really it's trying to understand where there are pain points from their end users and trying to help them kind of bring it to that.

Another piece of that, though, is sometimes they want to bite off more than they can chew and look at a huge beast of a process. And we really want to say, okay, let's narrow it down. Like going back to that business membership example, that's an hour long phone call. So being able to break that into smaller pieces that are more manageable from an analysis standpoint, we kind of provide support there too. So I think I answered your question.

Awesome. Thanks, Gina. Yeah. Thank you. Yes, sir. So one of the common issues with robotic automation is automating the user interface. So even that some elements change in the user interface, it breaks the bot. Obviously, in a good ideal world, we would know ahead of time so we can put the fix in. But that never happens. So and the bot is now down for maybe one day, maybe a few hours, one day, two day, three day to do the fix, right? So how do you tackle these situations? What sort of process you have put in your place to kind of govern these and manage these.

So that is much challenging and we still run into it. It's never perfect. The idea is that when we are starting a project, we're reaching out to the technical owners of the UI application that we're automating against making sure that we're setting up those relationships so that when they are getting an upgrade that's coming through, that they're passing that on to the people that are automating against it, so that we're not just building automations against applications and those applications owners not knowing about it, but sometimes those are vendor managed applications, and maybe we don't get that type of notification because the owner was inside of our company doesn't even get the notice.

It pushes an upgrade, something that would be relatively transparent to you when using the UI. But if you built an automation that has match rules against maybe some element that's unseen on the screen because they decided to change their naming convention for their IDs, then your automation breaks. At that point in time, it does have to go into a production fix. There's not much you can do at that point in time, but get the tasks assigned to IT to go and resolve and get that change pushed out there.

We do do those types of deployments through incident management as opposed to a normal production release. So we are able to get our releases out there faster without going through all the necessary security reviews that we would go through in a normal production release. We take on those security reviews later on when we do the next formal production release.

Thank you. Great. Thank you Kenneth. Any other questions? Oh go ahead. I had a two part question on buyer and user. Could you talk a little bit about when you first made the business case to get the product, sort of who was the champion within the organization to be able to buy it and then once you bought it. Can you talk a little bit about how you got users to buy in to use the product? You had mentioned that you have some pockets of people that are resistant to change, so they're not using. So could you talk a little bit about the buyer and the user? I'd love to hear about that. Thanks.

Okay. Can you talk about buyer? I'll talk about user. So when we were first bringing in, both Wi-Fi and Pega kind of came in at the same time, we were looking to be able to bring on a platform to be able to provide governed development for our business development or citizen development that was happening inside there. We had a lot of development going on in applications like Lotus Notes, and there was not any governance around it. Many applications getting created. So with that setting up our Center of Excellence and having a platform, we're able to take that unmanaged risk of development happening out there and having a platform that they could build on, and a governance process to be able to review what was going on with inside of it.

And then as far as getting people to actually use it. So with Wi-Fi starting out, we started doing roadshows. So we would go around to various business units and basically educate them on what the tool was, what it can do. And then we stood up a training program. So making sure that we can train their analysts to use it themselves. They don't want that dependency on us to to do the work for them. So that was a big piece of it.

Um, and like I mentioned early on getting some getting some Navy Federal examples. So once we had some teams that had adopted it, it made it a lot easier for the other business units to now understand, because it's one thing to hear Pega present, you know, use cases or generic use cases or ones that don't apply to the financial services industry. It's another thing to hear different business units in your own organization and how they use it. And sometimes it can be used similarly across different areas. Um, so that was a big piece in, in kind of educating. And then once people were getting trained, it kind of word of mouth got caught on because it was filling a gap where a lot of these areas don't have process metrics. And now they could get them.

All right. Cool. Well, thank you all so much. Thank you again, Jenna and Kenneth. Than ks .