Transforming Mission-Critical Processes with AI-Powered Hyperautomation

Barry Condrey: Hello there. Welcome to today’s webcast, “Transforming Mission Critical Processes with AI Powered Hyper Automation.” I’m Barry Condrey, Senior Fellow with the Center for Digital Government, and I’m excited to serve as the host for today’s webcast. Thanks for joining us. We’re gonna have a great session over the next 30 minutes.

We have a few housekeeping notes for you on the screen, and there are additional resources available for your reference there as well. You can participate in the question and answer with us by submitting your questions using the Q&A box. Please make sure you have your questions ready for this amazing panel. Send in your questions as they come up throughout their presentation, and we’ll address as many of them as possible. Remember, there are no dumb questions, only questions that don’t get asked. If you’re thinking of a question someone else is too, someone has to ask it. Your questions are the most important part of this discussion. At the close of the webcast, we’ll encourage you to complete a brief survey about the presentation. We’d like to hear what you think.

Joining me today to discuss this topic are Kim Evans, Director of Family Support Services for the State of Missouri, and Nelson Munn, Public Sector Director for Hyperscience. Kim, can you take a few moments and introduce yourself and talk about your background?

Kim Evans: Sure, Barry, and thank you for having me on the panel today. I’m Kim Evans and I’m the Director for Missouri’s Family Support Division. In the Family Support Division, we determine eligibility for our benefit programs for SNAP, Medicaid and Temporary Assistance. Very glad to share our experience today with the Hyperscience automation.

Barry Condrey: Great. Thank you. Kim. Nelson, can you take a few moments and introduce yourself and talk about your background?

Nelson Munn: Absolutely. Barry, thank you again for having me here today. I’m Nelson Munn. I’m responsible for our Public Sector business at Hyperscience. Got a long history in IT. I started my career as a developer, moved into sales and account management about 15 years ago. One unique aspect of me is I did public service for about five years under Jeb Bush here in Florida. I was the Chief of Staff, the Secretary, and I was the Board Chair for the first state data center here in Florida. So, got a great connection to public service, love the business we support. And again, happy to be here, Barry.

Barry Condrey: Great. Thank you, Nelson. So we have a great combination of backgrounds here with our panel, both experience on the customer side, delivering services as well on the technology side with a focus on public service. That’s excellent. To get us started today, we have a poll question for the audience for you to answer. Please take a moment and review this poll question: What is your biggest challenge in managing administrative processes? Please feel free to pick all of these that apply in terms of your largest challenges. I’ll give you a moment to read through those.

While you’re answering the poll question, we’ll go ahead and get started. I’ve worked in local government for two decades, and I’ve come to understand just how much we love our paper and our forms. Agencies at the state and local level process many millions of pages of paper every year. We’re drowning in a sea of paper, processing this information manually indexing it, updating it, creating workflows with it. All these things are very staff intensive. This administration of the paper robs agencies of vital resources and impairs the agency’s efficiency in delivering these services to residents. Automation has become a solution for this in the past, but now we have new technologies to take automation to an entirely new level: hyperautomation. How can we make use of this new combination of technologies to improve quality and efficiency of processing resident information? Our expert panel is here to help us with that.

Let’s take a look at our poll results and see what everybody said. Okay. Maintaining Accuracy and Reducing Errors, 78% picked that as the largest challenge. And if possibly we could scroll down a little bit, or I can scroll down here on my screen… we also see high scores for Staff Shortages, Raising Efficiencies and Handling of High Volumes of Paperwork. All those things sort of tell a very common story.

Let’s move to our panel discussion now. I’ve got questions for our panelists here, and they will be considering each other’s questions along the way as well. We’ll start with Kim. So Kim, talk to us about what drove Missouri to explore the need for hyperautomation and a solution like Hyperscience. What problems were you trying to solve here?

Kim Evans: Barry, the results that we saw in the survey were exactly the things that we were trying to solve. The lack of staffing, the volume of work coming in, trying to meet our timeliness standards for our federal partners. Also, providing that good customer service because providing accurate and timely services to participants, that is great customer service. And to do that, we were doing that in a manual process, which the data entry piece was pulling my staff away from having time to be able to process and making sure that we were paying attention to processing and accurately providing benefits to participants.

Barry Condrey: Okay. When you’re at the state level, Kim, it seems to me that you have the feds above you and the locals below you, that you probably have to have partnerships. You’re sandwiched in the middle there, right? Okay, very good.

Nelson, that’s a great backdrop of problems that Kim and our audience has laid out for us now. Hyperscience has been providing technology solutions for over a decade and what’s the Hyperscience position on hyperautomation and how has that come about?

Nelson Munn: Great question. Thank you for that. If you think about hyperautomation, what is it? It’s a big buzzword. So if you look at the definition, it’s pretty simple. It’s this relentless pursuit within organizations to improve process, improve their data quality, be more efficient. It’s music to my ears, Kim, I know it is to yours. Back when I was a state official, believe it or not, we were resource constrained. We did not have enough people to do these functions. The more efficient we can be, the faster we can deliver services, the better for everyone.

The challenge, believe it or not… nine out of 10 leaders today believe that automation improving processes is important. The challenge is, it’s hard. It’s really hard. Deloitte will tell you that only 13% of these initiatives actually are successful. We’ve gotta ask why that is. We all know it’s important. What’s the challenge? I will tell you, I’m guilty of being part of the problem. Back when I was a young developer, we built line of business systems. We didn’t think about interconnectivity, we didn’t think about how they had to work together. We were just proud of what we built for that business unit. And over time, we built connectors and we built bridges. Nothing to support what’s needed today.

Kim, I’m sure you will tell us that some of those systems still exist. They’re still out there. Barry, when you start layering on now that the business has figured out that IT is important in driving the services, and we’ve got a seat at the table, but that comes with more and more requests for services, more that they’re demanding on IT. Lastly, you layer on top of that all this unstructured data that comes in that’s needed to drive the business, whether it be videos, whether it be PowerPoints, chat, GPT data now coming into the system. That unstructured data is really hard to manage in legacy systems. We will tell you that 90% of the data coming in is human readable, but only about 10% supports those legacy systems. That’s how we got here. We’ve got all this data, we’ve got legacy systems, but we know we’ve got improved processes to support the mission that Kim’s challenged by the legislator to deliver.

What’s Hyperscience’s position? It’s gonna be done through comprehensive AI infrastructure. Not a bunch of APIs that you program, but a suite of products that allow you to define a business process, assign digital AI agents to that process, give them responsibility, give them measurements, allow them to raise their hand when they have a problem and treat them just like we do our normal human employees today. What that’s going to do is drive the efficiency automation that we’re seeing across our customer base—98% automation, 99.5% accuracy—the kind of levels that our people are looking for. By the way, Kim, I think you would agree, if we can turn those humans back to serving the mission into the agency, we’re doing a better thing for the community we support.

Kim Evans: I totally agree.

Barry Condrey: Nelson, that’s a lot to unpack. You really got us off on a good foot there. I love what you said about human readable versus suitable for legacy systems. I also love what you said about the demands on IT as a CIO, I can definitely relate to all of that. But the first thing I wrote down and underlined three times was the 13% of process automations that are successful. Now, Kim, I’m willing to bet you count yourself in that 13%. Let’s talk about what separates the other 87% from you. You have a clear commitment to the use of hyperautomation. Let’s talk about your benefits. Nelson will tell us a little bit later about how we can help that 87%. What benefits is Missouri seeing from utilizing Hyperscience hyperautomation?

Kim Evans: We are on our journey is what I call it with Hyperscience and really automating all of this. We started this journey during the Public Health Emergency and during the Public Health Emergency, not only were we experiencing all of those changes, but Missouri also expanded their Medicaid program during the middle of that. So we added an additional 300,000 individuals to our Medicaid rolls while we were in the middle of the Public Health Emergency. Our rolls were growing, our paperwork was growing.

What we’ve benefited from this is really the timeliness issue of being able to accept documents, Hyperscience being able to read those documents for us, and then us ingesting that into our eligibility system. It’s what I call “full kit.” It provides that to my worker. We have a tasking system that says once it’s registered in the system, says, “Hey worker, you have this task, you can work.” I as a worker—and I’m gonna tell my age here—but when I was an Eligibility Specialist, I had to do my own data entry when that was assigned to me. Now I just get a task and all of the electronic verification is there, the application is registered for me, and I just have to look at everything and ensure that everything is there and everything is accurate. Are there other pieces that I have to verify and I could process?

We’re seeing that and speed of processing starting to increase. Like I said, we’re just on our journey of updating our system, but also really looking at the accuracy of the data that is entered. Because we’re all human, we’re gonna make mistakes when we do data entry. Just the accuracy piece of the data entry is really starting to help us as we’re looking when we’re doing our QA and QC piece on that. Also, staff love this. They didn’t hire in to be data entry clerks. They hired in to be Benefits Technicians, and they get the better fulfillment of knowing that when they’re processing the applications, they’re processing more timely and they’re helping participants the need that they have.

Barry Condrey: Kim, between you and Nelson, I’m getting the definite message here that 30 minutes is not gonna be enough to really unpack all of this with you two. I just wanna call attention to the fact that you mentioned the term “journey” several times. So often when we’re looking at transformative efforts, we miss the fact that it really is a journey. You never really get there. You’re always improving, you’re always looking for things to change. That’s such an important thing that you mentioned there.

You mentioned accuracy. I think Nelson mentioned 98 or 99% accuracy is the metric that Hyperscience puts out, which I can imagine is a vast improvement over human effort to do. I know my typing is nothing near 98 or 99% great. But the thing I like the most about what you said, Kim, is “staff love it.” That has been mentioned as a barrier a lot of times with staff and automated and artificial intelligence systems coming in, that there might be some trepidation there on the part of the workers. But you’re saying you’ve got a great story to tell with that, and that’s awesome.

Nelson, let’s go to you and talk about the AI for a minute since that’s where Kim ended. Generative AI and Large Language Models, they’re everywhere now. We see stories about them every day. Talk to us about how Hyperscience is using Generative AI and Large Language Models and what makes your approach so effective to using them?

Nelson Munn: Great question. Everybody’s talking Gen AI. It’s the buzzword that’s out there and we support it fully. But if you look at Hyperscience and our ten year journey and our heritage, we’ve been building on an AI platform for 10 years. Our point of view is very mature and in our system we have over 30 proprietary AI models. All of these capable reading very complex information to get to those levels of automation and accuracy we’ve been talking about.

We fully support Gene AI and LLMs as part of Gen AI. They’re really part of the umbrella because they’re so good at speaking the human language and generating content that’s useful for business. Lemme give you an example. We have a user that’s reviewing medical claims and part of that claim is a small piece of text generating a summary of a handwritten multi-page assessment of a client’s issue. That’s the LLM taken and summarized on behalf of that customer so a good decision can be made. We’re fully supportive of LLMs, but the enterprise still struggles with structured data. Once you have that data, you still have to be able to take it, interpret it, tag it, and pull the content out that’s useful for the business process. That’s where a lot of the challenge is.

Just one more point, Barry, ’cause I’m very verbose today, but we just launched a Gen AI solution along with our Google Cloud partner as well as Hewlett Packard Enterprise. It’s gonna provide some really exciting Gen AI experiences, so please look for some more on that coming up soon.

Barry Condrey: Great. I think I heard you say the models are mature and you’ve been working on them for quite a while now. That tells me that, and I want you to be clear about this, that I think there’s a lot of misconception out there about private Large Language Models and the public Large Language Models. Your information is not making it into ChatGPT 4.0 right now. You’re not using that Large Language Model, you’re using your own internal models, correct?

Nelson Munn: The answer is it depends. Certainly up to the customer. We continue to build our product that it can be fully encapsulated on-prem. If the user wants to plug in ChatGPT for a function, more than happy to do that. But for our proprietary system, it’s self-contained, it’s trained on your data and it’s contained. It just depends on what the customer wants. Barry. We’re flexible in that regard.

Barry Condrey: So you can use it where it’s appropriate. Great. I think that’s an important distinction for the audience to have. Now, Kim, social work seems to me is all about improving outcomes for clients. That’s what we’re all after with social work. Talk to us for a minute about the hyperautomation tools that you’ve used and how you’ve improved those outcomes and how you’ve impacted the individuals that Missouri serves your customers. How have you impacted them?

Kim Evans: It gets back to that accurate benefits and more timely benefits. We’re always trying to improve our timeliness. We have standards that are set by the Fed, but we are always trying to improve those. We wanna get those benefits out as quick as possible because those individuals need those benefits. If they need food on their tables or they need those medical services. Part of this is the applications that come in or the verification pieces that come through that are scanned, read, and put into our systems. The quicker that information can get to my staff, the quicker that we can process.

Those are the things that we’re trying to improve our timeliness on: from the moment that a participant turns something into us, how quickly can we get it into the hands of our staff? And how quickly at that point can we process to get the benefits out the door to them? Because individuals need these services to be able to assist their families. That’s the whole thing that we’re looking at when we’re looking at all of our range of services or technology that we’re putting together to be able to serve Missourians.

Barry Condrey: Great. So do you track that with metrics? Do you track like the cycle time?

Kim Evans: We do. We set our base metrics as we started with Hyperscience and then we’re measuring our timeliness about the accuracy of the documents, the reading that Hyperscience is doing. We do a quality check on the accuracy of the data coming over. We also do checkpoints on how quickly is it getting to our staff. Like I said, we set a base point so we knew what we were doing previously and how quickly are we processing now. And then we also have our federal metrics that we’re comparing that to also.

Like I said, we’re on a journey and staff really enjoy this because it has taken a burden off of them. Because with that data entry, that was an added step that they had to do. Now their step is really focusing on the processing piece, not the data entry. They just really have to focus: Did they issue the correct benefits?

Barry Condrey: So it sounds like your staff are happier with this using this tool in general.

Kim Evans: They are, and it was the way we presented it. It could have gone two ways. It could have been staff thinking that we were gonna replace them with technology, or it could have been “it was the thing that we’re giving you an aid to help you.” And we spent a lot of time—I spent a lot of this during the time of COVID, so we did a lot of WebExes and town halls—me reassuring folks that I’m not here to replace you. I need every one of your warm bodies doing work. But this is me giving you a tool to help you. Rather than doing the data entry, helping you be able to process quicker, and they’ve been very receptive to it.

Barry Condrey: Great. You know, I think it’d be fascinating down the road to take a look back at your retention and your staffing metrics and see how the use of hyperautomation that your staff appreciates and your communication strategies that you’ve used for the change management, how that’s helped your staffing and your retention. I bet you’ll find that you’re doing better than a lot of peers in that respect. I know how hard it’s to keep social platforms and social services types organizations fully staffed in the public sector. So kudos to you, Kim, for that implementation.

Nelson, coming back to you for a minute. Kim was talking to us about the return on investment here, and she mentioned it in several different ways. Customers are always looking for return on investments they make in technology. So how does Hyperscience create value for the customers like the Missouri Department of Social Services and what can customers expect here?

Nelson Munn: Great question and I think I’ll put it in the context of Public Service. We are dedicated to public service at Hyperscience. My team understands the language of government. Many of our customers don’t talk profitability and revenue. They talk mission and how to drive that mission and the success of that. We talk that language. Kim and I were just talking about legislative mandates, unfunded legislative mandates. We were talking about how hard it is to get funding. All that unique conversation comes together on my team with people that can speak that—a lot of us have served, myself included. We bring that value in front of the product.

But when you look at the Hyperscience platform, it fits perfectly for government. Whether you’re regulatory, whether you’re health and human services, public safety, you go down the list. There’s people, to your point at the beginning of the call, that’s consuming documents and content that we can help them with. If we do that well and get that level of automation and accuracy right, we put those people back serving the mission. For me, there’s no better cause than to help customers that affect the community I live in and the families that I know through these government services we support.

Barry Condrey: It is a different value equation, isn’t it? Support of the mission. When the mission is so profound. Everything. Mission is everything. Absolutely. Thank you, Nelson. That’s a great perspective.

And so Kim, last question is for you. You got the final word here on the panel questions. Lots of folks are wondering how to get started with this technology. I’ll guarantee everybody in the audience is like, “Wow, we could just achieve the benefits that Kim has articulated. What would that look like in our environment?” What are some hints, what’s your secret sauce here for how they can get started?

Kim Evans: I don’t know if there’s any secret sauce. What I’m going to say is don’t be afraid. A lot of times when you talk new technology, especially when you’re talking anywhere around AI, people could start to get scared and get paralysis. Don’t be afraid of it. Because it can be your friend and use it to your advantage. Look at your gaps, look at your pieces that suck the power out of your staff or suck the time out of your processes. That’s what you need to attack and be open and honest and go to the table with Hyperscience and have those hard conversations, because that’s what I did. I brought ’em to the table and I said, “Here’s my problem. Help me solve it.” And that’s where we started the conversation. Just be ready to have those open conversations and let them help you work through it. Your journey may not start where you think it is, and it sure not going to end where you think it is, because ours has taken a twist and a turn a couple of times, but we’re getting to where we need to be. Just be flexible and be open.

Barry Condrey: Great words of wisdom to end on and find a good partner. Sounds like you’ve found yourself a great partner that’s been very supportive for you as well, so absolutely. Let’s have those questions in the Q&A. We’re gonna transition now to some live Q&A and we do have a couple of questions there. If there’s anybody else out there that’s got a question, please go ahead and put that into the Q&A box, and we’ll get to it if we can. But we’re gonna segue to our live Q&A now.

And let’s see, our first question is a question about staff adapting to the new AI processes and the training that was necessary. Training is so difficult sometimes to provide to staff when you’re already resource constrained, making time for that so important, it’s difficult to do. So, Kim, what kind of training was necessary for your staff to make use of the new tools here?

Kim Evans: It’s interesting that, and thank you for that question, because that’s something that’s very important. During the Public Health Emergency, we sent all of our staff home to work. I replaced all their PCs with laptops and everybody go home to work because we couldn’t shut down. We had to work, and that was not our normal business model. We also had to learn how to have our trainings online. We have switched all of our trainings to an online session and we had to then teach them, “Okay, so you’re not doing the data entry now, so this is what you’re going to expect in the system.” We had to rewrite our standard operating procedures. All of that had to be taught and then incorporate that into our new hire training also. So it has been a major training issue, but it’s been successful.

Barry Condrey: So you talked about new hire training too. It’s not just about training the existing workers, but making sure you’ve got that training built into the process, the evolution of the employee from the time you bring them in until the time that they segue out. That’s a great thing for people to remember.

And okay, one more question real quick. And this one will come to you [Nelson]. How do you see the role of AI evolving in government operations over the next few years? Take about a minute and tell us how you see AI evolving in government operations.

Nelson Munn: I believe fully that AI will become more and more important, controlled AI. Hyperscience through our 10 years believe that “human in the loop” is an incredibly important part of the process to make sure those digital AI workers acting as we expect them to, that whole life cycle of managing the resources. But it’s already being given that traditional RPA processes are falling apart because they’re still rules based. They’re still pretty structured. And what we need is a more dynamic way to manage these processes as AI infiltrates that. So we just gotta make sure we manage it. I don’t always listen to Elon and that AI’s gonna take over the world, but we do have to make sure we manage it and have the tools. I go back to our AI infrastructure to make sure we’re managing those processes and those resources.

Barry Condrey: I love what you said about having the human in the loop. That answers so many of the things that come up when you talk about AI, especially in the social realm. That’s great.

Okay, so I wanna be respectful of our time commitment. We only have 30 minutes and I could spend all day talking to this panel about this topic, but so we’ll have to wrap it up here. I’m sorry. In closing, I would just like to thank our panelists, Kim and Nelson, for sharing their insights on today’s webcast. I’d also like to thank Hyperscience for making it possible to have this important webcast discussion today. And thanks to you, our audience, for spending time with us. We look forward to seeing you again at another government technology event. Have a great rest of your day. Bye-Bye. Thank you.

Nelson Munn: Thanks.