UiPath Fusion 2025 | Andrew Kemble, Greenshield & Al Ochoa, Southern California Edison

Clips
More from UiPath Fusion 2025

Andrew Kemble

Manager, AI Enablement & Delivery

Greenshield

Alfred Ochoa

Principal Manager, Predictive Analytics & Data Science

Southern California Edison

Smarter service, stronger systems: Inside the rise of agentic AI at SoCal Edison and GreenShield

Customer engagement and operational efficiency have been identified as key innovation vectors for organizations implementing artificial intelligence strategies.In harnessing that opportunity, teams from GreenShield and Southern California Edison are leveraging agentic AI, low-code tools and UiPath Inc.’s orchestration capabilities to reshape customer service, empower employees and set new standards for operational efficiency.GreenShield’s Andrew Kemble and SoCal Edison’s Al Ochoa discuss AI for operational efficiency with theCUBE.“At Southern California Edison, we started working on knowledge management

play_circle_outline Introduction of guests Al Ochoa and Andrew Kemble at UiPath Fusion event.

play_circle_outline Al's role at Southern California Edison focusing on customer service and automation.

play_circle_outline Importance of data quality and centralized knowledge management in AI initiatives.

play_circle_outline Discussion on low-code tools empowering business units to solve their own problems.

play_circle_outline UiPath's vision and partnership impact on both companies' automation success rates.

Info
Transcript

Andrew Kemble, Greenshield & Al Ochoa, Southern California Edison

Andrew Kemble

Manager, AI Enablement & Delivery Greenshield

Alfred Ochoa

Principal Manager, Predictive Analytics & Data Science Southern California Edison

In this UiPath Fusion 2025 segment from Las Vegas, SoCal Edison’s Al Ochoa (principal manager, predictive analytics & data science) and GreenShield’s Andrew Kemble (manager, AI enablement & delivery) join theCUBE’s Dave Vellante and Rebecca Knight to explore how enterprises are evolving from traditional RPA to agentic automation. The discussion covers GreenShield’s early production rollout of communication mining to transform high-volume email support, why centralizing knowledge and customer data is prerequisite for trustworthy AI, and how low/no-code is empo... Read more

explore Keep Exploring

What are the roles and organization details of the guests featured in this segment? add

What is the role of Al at Southern California Edison and what does his organization focus on? add

What steps were taken to ensure successful training of AI at Southern California Edison? add

What is the impact of low-code and no-code tools on business ownership and problem-solving? add

What is the relationship between the company and UiPath in terms of automation technology and partnership? add

bolt Powered by CUBE AI

Andrew Kemble, Greenshield & Al Ochoa, Southern California Edison

search

Rebecca Knight

>> Hello everyone, and welcome back to theCUBE's Live coverage of UiPath Fusion here in Las Vegas. I'm your host, Rebecca Knight, alongside Dave Vellante, my co-host and analyst. I'd like to welcome two guests to this segment. We have Al Ochoa, he is Principal Manager Predictive Analytics and Data Science at SoCal Edison. Welcome, Al.

>> Thank you.

Rebecca Knight

>> And Andrew Kemble Manager AI Enablement and Delivery at Greenfield. Thank you so much for coming on.

Andrew Kemble

>> Thank you. Thank you so much.

Rebecca Knight

>> I want you both to set the table for our viewers and talk a little bit about your organizations and what you do, what you do with them. So Al, let's start with you.

Andrew Kemble

>> Great. And so at Southern California Edison, I work in the customer service organization. And so we're really looking out for the contact center, for the billing organization and for customer experience and marketing. We do automation such as predictive models, proactive communications, artificial intelligence, large language models, robotics process automation and desktop automation.

Dave Vellante

>> Okay. Are we going to get enough power to power AI, yes or no?

>> Yes. Yes, we are.

Dave Vellante

>> Good answer. We need you.

Rebecca Knight

>> Andrew, tell us a little bit about Greenfield.

Andrew Kemble

>> Yeah, so GreenShield.

Rebecca Knight

>> GreenShield, sorry.

Andrew Kemble

>> We're a Canadian based company. So we're Canada's first nonprofit payer provider. So we've been around for almost 70 years now. We were founded in Windsor, Ontario, so two minutes from Detroit, right across the border south of Detroit, as the Windsorites will say. Really for us, it's about how can we drive efficiencies to help Canadians' lives. How do we help provide better for all, better health for Canadians?

Dave Vellante

>> So my understanding is either one of you or both of you started with email, is that right?

Andrew Kemble

>> Yeah.

Dave Vellante

>> Was that GreenShield or SoCal?

Andrew Kemble

>> That was GreenShield. GreenShield.

Andrew Kemble

>> Yeah. So within GreenShield, I think we were the first team in North America to go live to production with a communication mining. We did a proof of concept back in 2023. And truthfully, we were helping our contact center out with things, so customer service center, and we had seen these solutions before and I wasn't buying. To be very frank, I'm like, "This isn't going to work. Yep. We'll do a proof of concept. Fine. I can say I did a proof of concept this year. Let's go back to our PA and continue some improvement processes." There's too much other stuff going on. Well, the results of what we saw in the proof of concept kind of floored us from holy smokes, like we need to do this, we need to get it to production. Look at the extra value that comes through of just having this level of insight. Then also what we can layer on with these other automations.

Rebecca Knight

>> So drill down exactly what you were trying to get at and what you found, what the insights were that were so valuable.

Andrew Kemble

>> Yeah, so we had six performance indicators that we were looking at. There were things like finding automation opportunities, proper classification. Really, we were just trying to understand the nature of email communications. So what we had seen over the last, I'm going to say five years, was our email volumes kept growing and growing and growing, but our call volumes were staying consistent. We were putting a lot of effort and energy to helping with the voice channels, but we were kind of ignoring email. We just thought, "Yeah, what worked with voice will work with email." Well, when we did the proof of concept, we realized no, people are emailing in about very different concerns than what they call in about. So now, how do we address that? So now we had all these new opportunities that came to the board of saying, "Let's do that, let's dive in here."

Dave Vellante

>> So Al, the research that we've done suggests that, obviously, legal and compliance are big concerns for people in this AI race here, but we consistently hear that data is the key. You got to have your data. And the fact that your title is predictive analytics and data science suggests to me that you might know a little bit something about data. So we talk a lot in theCUBE about bringing the AI to the data. Is that what you're doing? What does that mean? What is the state of your data? How important is all that? I wonder if you could talk to that.

>> Yeah, absolutely. I can tell you that is the most important part of making this all successful. And so at Southern California Edison, we started working on knowledge management, and so that's unstructured data. So if you think through and training a AI expert, you really have to first make sure that all your training and policy procedures are accurate and they have to be stored in one place. Because if you feed the AI agent bad information, it's going to give bad results. So for us, focusing on knowledge management was critical. Also making sure that our data that we expose customer data was in a space where we can go to one location, one centralized warehouse. So that really was our strategy that we worked on for probably about a year and a half prior to even building an expert or agent, AI agent.

Dave Vellante

>> Okay. So you actually have the data sort of consolidated?

Rebecca Knight

>> That's right, centralized.

Dave Vellante

>> And cleansed, harmonized?

Dave Vellante

>> That's right. Yep. You have to have governance over that, making sure that people are updating documents consistently and that there's oversight over all the material.

Dave Vellante

>> And then how do you handle access control?

Dave Vellante

>> Yeah, and so each organization has experts within those organizations that have the ability to update their policies and procedures, but we still have one centralized area that governs to make sure that only certain people have access.

Dave Vellante

>> Interesting.

Rebecca Knight

>> So Andrew, you talked a little bit about the experience of deploying UiPath communications, mining support on your customer support inbox, but I want to go back to the humans actually who are part of this too. What impact has it had on team members, and how has it shifted their tasks and roles as well as how they do their jobs?

Andrew Kemble

>> Well, I think the very first thing that we looked at was when we saw the nature of communication that were coming through, we realized how much time is being spent just responding back to our plan members and our patients saying, "Can you verify who you are please?" We wanted to make sure that from a compliance standpoint, from a risk standpoint, that we were giving the correct information to the correct person. So right away, all of the emails that come through saying, "Hey, I'm Andrew, can you help me with this?" "Yes, absolutely, Andrew, I want to help you with that. Can you give me the three or five pieces of information so I can help you with that?" So now, instead of I'm going to say if we had a two-day turnaround time on emails, sending an email and two days later getting a response saying, "Sorry, who this?" No. Now, they get that response right away. So that way, I know, oh, yeah, I need to provide this information. And even from an agent compliance perspective, we have a step one. Our step one is making sure that through communication mining, intelligent automation that we're responding succinctly, precisely. And that way, our agents now, they just have to worry about helping the plan member. They don't have to worry about verifying, and am I giving the correct information?

Rebecca Knight

>> And that's the boring stuff, that's not that interesting. So that's helped them do their jobs and find more meaning and interest in their jobs, but what about from the business perspective?

Andrew Kemble

>> Well, from the business perspective, I'm going to say the first set of use cases that we wanted to put in was we saw just how many people are struggling to get into their accounts and registration. How often it's, "Hey, I forgot my password. Hey, I forgot how to register." Well, we're a Monday to Friday is our hours of operations. We're 8:00 AM to 8:00 PM essentially. But if you're a single parent that's coming through, you're not thinking of submitting your claim at 4:00 in the afternoon. You're bringing your kids to soccer practice, you're coming home. It might now be midnight. You wake up and you think, "I got to submit my claim form." So you go to the app, you try to log in and you go, "What's my password with my account information? How do I do this?" Well now, you can send us an email. Essentially, we've got 24/7 support hours because we can answer those emails right away. So that way, you can submit your claim when you need to submit your claim.

Dave Vellante

>> What's your experience with, Al, and maybe you guys as well, but with low code? You hear this term vibe coding. Is vibe coding just a new name for low-code? How does agentic change the low code mantra? How is that evolving?

Rebecca Knight

>> Right. Well, I can say that our experience with low code tools is that it's actually allowed the business portion, not our IT department to get into the development space. And so if you could think about a team of experts that are talking to customers that are actually processing bills now being able to come in and be a part of development of technology, that's how we're getting use cases to get to production very quickly. When we think about agentic AI, leveraging the UiPath tools really does give us that capability to reach what I'll call large language model tool, that's like the ChatGPTs and expose our data to that type of tool, and then get results back from it. It's very powerful. We have a use case where you're a utility customer, you have an electric bill, and customers call in, they're upset because their bill is higher one month to the other. And so we're able to build this use case where we train the model in the training material just like we would an agent, and then we expose it to the customer's data. And the actual model, actually, the AI expert reasons and analyzes the customer's account. So it will go back across and look at usage patterns from one month to the other, or a year. It will look at weather patterns and see what was the temperature during that timeframe? Did the temperature increase? Did the customer fall off any programs? Were they on a low income rate and that rate dropped and that's why their bill went up $40? Then what it does is it produces that result out to an agent, and the agent now just has to read back to a customer what the problem was. So for us-

Dave Vellante

>> That's a human agent?

Rebecca Knight

>> That's right.

Dave Vellante

>> Oh, okay.

>> For us, that's the first stage. We don't want to put this type of technology in front of customers yet. We want to be able to monitor it, and so we're doing that by providing the information to agents first, and then later we may go straight to the customer.

Dave Vellante

>> So the electronic agent has agency to support the human agent.

Dave Vellante

>> That's right, absolutely.

Dave Vellante

>> And then that human's in the loop and the .

Dave Vellante

>> Yeah, we call it their partner, right?

Dave Vellante

>> Yeah.

>> It's the partner analyzing the data for them and giving them the results, and they get to speak it to a customer.

Dave Vellante

>> And Andrew, are you similar kind of situation?

Andrew Kemble

>> Well, I think it's empowering, right? Suddenly now, through no low code tools or vibe coding, it allows the business to start taking more ownership of their problems.

>> That's right.

Andrew Kemble

>> I think the closer you can get the end state solution to the business, the better the solution is going to be. The business is going to solve their own problems with help from IT or without help from it. So I think low, no-code solutions, vibe coding solutions, that's empowering for employees to solve their own problems, empowering for business units to solve their own problems, and then now it becomes an advisory capacity. How do we support you in this? How do we maybe level up what it is that the automation you put in place, the AI agent that you've built for yourself? How do we make sure that it's not a single point of failure? All things that on the IT side we're going to be thinking about and concerned with, but we want to make sure that the business can run the way the business wants to run.

Dave Vellante

>> Were you both RPA customers?

>> Yes.

Dave Vellante

>> Okay. So how relevant, important is having that RPA foundation and how does it complement what you're doing in agentic?

Dave Vellante

>> So if you think about agentic, agentic also needs orchestration to get the work done, and so we can get that work done through RPA. So you have an expert going and analyzing data. Now, knowing that it needs to take that result and do something with it, we pass it to an RPA bot and that RPA executes on it. And so let's say that you need an order completed, the RPA will complete the order for you. So it's this orchestration of having all of these tools work together.

Dave Vellante

>> You trust the bot more than you trust the agent?

Dave Vellante

>> You know what, we've been doing RPA for a real long time.

Dave Vellante

>> So, yeah?

Dave Vellante

>> Yeah, absolutely. Yeah. And the agent is something new, and so we want to take a cautious approach to it.

Dave Vellante

>> Is the agent like a smart intern?

>> Way smarter than an intern. It's yes.

Rebecca Knight

>> Well, when you were talking about how it's empowering to companies to solve their own problems, I could also see how it's also empowering to the frontline employees, although maybe that person will soon be cut out of the process, but because they are the interface with the customer, they are having to deliver the human news of why their bill was higher and deal with the aggravation or the understanding of the customer. But that human is also able to notice patterns. "Hey, I'm getting a lot of calls on this. This seems to be an issue within the business." And then alert others in up the chain. I'm just wondering if you've seen that happen as yet, or what you predict will happen there?

Andrew Kemble

>> I think when we're ... I'm going to bring you to 2030.

Rebecca Knight

>> All right, okay.

Andrew Kemble

>> The way that I think the role of a contact center agent will start changing, I think a contact center agent will start managing their own agents that are come to play. They're going to have five agents at a time that are going to handle nature of different communications. And just like now, a contact center supervisor might be live listening to calls and giving feedback and coaching them in terms of like, "Hey, maybe we can deescalate a customer this way. Maybe we can answer this approach that way." That's going to be become the role of a contact center agent. They're going to be watching and coaching and mentoring their agentic agents in terms of their response, and they're going to help prompt them. They're going to help code them. They're going to help say, "Hey, slow down a little bit. Maybe we're over-communicating, over-explaining right now to this plan member or this patient." Or maybe this plan member patient gets it, we can speed things up a little bit. So I think the role of that customer service agent, we'll start becoming an escalation point. They're going to become that knowledge center and that way if a plan member or a patient of a customer is getting frustrated, then that agent will step in. They'll get the indicator saying, "Okay, maybe something isn't quite working with this agentic. Maybe that intern doesn't quite know how to answer that complicated question. Let me step in and help now."

Dave Vellante

>> You know how every time we get a new smartphone and it's like triple the performance, it doesn't seem to change? And the reason is because you're pumping more data and more apps through that and the apps get bigger and smarter. So my question is, will there be more or less human in the loop in 2030? Because I could make an argument that it's going to ebb and flow because the humans are going to be that much more capable and have that much more time to create new ideas that they're going to need to be in the loop as much if not more. Do you have a point of view on that?

Andrew Kemble

>> I definitely think we're going to see more creativity out of staff. We're going to see the ability of staff to get deeper in those problems, because at the same time, our consumers are going to get more sophisticated in their answers, right? They're going to be assuming. Right now, we're seeing a commodification of knowledge. I think it's something that I've heard around here a little bit this morning, and what that means is knowledge is at your fingertips. So if knowledge is at everyone's fingertips, that means they're going to start asking more complicated questions. They're going to have more complicated concerns. Well, it's going to take a human to start answering those complicated pieces again. So I think we're just going to get more sophisticated. Yeah.

Dave Vellante

>> Has AI reduced your workload? To me, it's the opposite. I have now so many ideas. I have a backlog of cool stuff that I want to do that I can't get to that I never would've even thought of before.

Dave Vellante

>> I think we'll see a shift in jobs. I really do. And so looking at the potentials of what this AI can do, I really do think that it will take some positions, but we will see a shift in jobs where now we're seeing some of the mundane kind of repetitive things go away, things where AI can come in and take over easily. We'll start to need to see creativity from people to start come in and to monitor, to do quality assurance, to come up with new ideas. But definitely from what I see, it will take jobs.

Dave Vellante

>> I mean, I'm dying to ask Brynjolfsson about this. Erik Brynjolfsson is coming out tomorrow. I think he's given a keynote. He's like the productivity alpha, and I remember the productivity boom well of the '90s, and it didn't make me ... It made me a lot more productive, but it didn't make me less busy. Maybe in Europe I'm talking about, but we Americans-

Dave Vellante

>> Use this example. I'll use this example. Way back in the day, I was a meter reader and I never thought automated meter reading would come. I walked out to the house, walked in the backyard, I read the meter, and the customer got a bill.

Dave Vellante

>> Yeah, sure. I remember you guys get off my property.

Rebecca Knight

>> That's right. That's right. You don't see meter readers anymore. They're gone. Automation came in, it took over, but there's other jobs that now there's so much data of meter data that there's other positions that it's created. And so that's when I say that I think that there'll be a shift in jobs. It might not be the front line job anymore, but it'll be in another spaces.

Dave Vellante

>> Is there a cultural difference in Canada, because US is in an arms race, AI arms race with China, who can work harder, who can work faster. Is it different in Canada, or is it very similar?

Andrew Kemble

>> I think it's very similar. That comes through, right? We're so close as neighbors. A lot of it, there's some interchange of culture that it takes place, but I definitely think that there is a competition right now of how can we make use of this? What can we do? And there's a little bit of, I'm going to say of underdog kind of standpoint of we're only 40 million people that comes through, but how do we have such a big impact? What can we do differently? So there's a real mindset right now of embracing AI. We understand that there'll be change. We understand that the nature of roles and organizations are going to start changing and shifting, but we're excited by that.

Dave Vellante

>> What's your bottom line on UiPath? I mean, obviously, you've been there for a long time with the company. They've gone through a number of changes. You've stuck with them. Why? What are your thoughts on the future of the company?

Dave Vellante

>> I'll tell you, we have about 150 UiPath RPA licenses and about 900 robotics desktop automation licenses. And as we've moved into now this AI space, the one thing that I can say is that UiPath as a company has helped us with the vision and the architecture and really seeing how we can make it happen. And so we see them as a partner going forward. Being at this conference, I see the new technology that they're going to be providing. It's exciting. And so I definitely see us in a partnership with UiPath for a while.

Andrew Kemble

>> I think UiPath is uniquely positioned right now. You see the other hyperscaler players in the space that are coming through. You've got your Microsoft and your Google and your AWS, that's happening. But I'm just going to say UiPath feels closer to the problem. They feel closer to what we're trying to solve. They feel closer to the return on investment in the process. And it probably comes from their RPA roots, right? Here's the actual hammer that's doing the work that's taking place. So the perspective that they're coming from is one of orchestration. It's one of actually solving the problem, not giving you a tool or not giving you a solution and asking you for the problem to solve. No, this is very practical for us, and I think that practicality is allowing us as UiPath customers to really drive an increased success rate. Right now, everyone is talking about the MIT study that showed AI projects are failing. It's not hard to go on LinkedIn and to see the same pieces. I have not had that same success ratio.

Dave Vellante

>> No, I haven't either.

Andrew Kemble

>> I've almost been inverse of the number that's gone through right now. It's the low, no-code solutions, it's the orchestration. It's just closer to the problem, so the solution just presents itself a lot easier.

Rebecca Knight

>> Right, but the problem-

Dave Vellante

>> You guys, you're saying you're in that 5%.

Andrew Kemble

>> Definitely.

Dave Vellante

>> Yes.

Dave Vellante

>> And it's because you're sort of automation aware. UiPath obviously has a perspective on that. It's not just the tool, it's the mentality of automation, the culture of automation.

>> It's the use cases that you're picking and to make sure that they're successful. They have to have the return on investment, and so you have to be very careful about which business processes you're going to automate.

Dave Vellante

>> When Daniel Dines, I don't know if you saw his keynote this morning, he was saying that part of the problem for the failures is people aren't aspirational enough. They're not aggressive enough. I interpret that you think about ROI. You guys know ROI. There's many dimensions to ROI. There's an internal rate, there's a percentage, which is just math. There's a breakeven, and there's a number. How much money is it going to put in my pocket? Probably the bigger the number, the bigger the risk. So how are you guys balancing time to the value and IRR and the NPV, if you will? How do you guys think about that?

Andrew Kemble

>> I can say that I've made my mistakes over the year with going after the easy to implement automations with breakeven ROI, and you don't get a lot of momentum around that to actually do the next thing. And so if you want to sell your executives on the purchase of new software technology, you have to go after the hard-to-do things, the things that are going to show benefit in dollars. And so the things that they were saying today, the keynote speaker was saying was right on target. You have to challenge yourself. You have to go after it. And this way, when you show success, then you have opportunity to do more.

Dave Vellante

>> How do you de-risk that? Because if you're going after the moonshot, you mess it up, you're going to get fired.

Andrew Kemble

>> Well, but we're going to the moon and no one's been to the moon before. There's a little bit of that risk is baked into these projects that we're doing.

Dave Vellante

>> And so that's the part of that just setting expectations.

Andrew Kemble

>> Exactly.

Dave Vellante

>> Join me in the risk.

Andrew Kemble

>> Yeah. I'll say I'm very fortunate. I have an exec team right now that understands we're swinging big, so let's keep swinging big. We've got a culture of change right now within GreenShield of trying to make things better, which just allows us to be innovative. It allows us to take some ... I don't want to describe it as gambles, but it allows us to see opportunities and potential and to really chase after it.

Dave Vellante

>> Are you able to quickly ... So the whole fail fast mantra, but maybe it's not fail, maybe it's course correct, so we hit the moon.

Andrew Kemble

>> Exactly. Well, that's all it is right now is it's as you're going through proof of concepts, you're making these minor adjustments. So you're like, "All right, the moon's shifting a bit. Maybe the rocket's sliding out, how do I reposition to get to where I want to land?" And really, the analogy that I've been using a lot is we're on a moving train, golfing, but our hole keeps moving at the same time. So how do we get a hole in one?

Dave Vellante

>> So to make those micro adjustments, do you what? You start with the North Star and you work backwards from that, and then you start in little micro segments?

Andrew Kemble

>> I think you take an agile approach, that's a project management methodology.

Dave Vellante

>> Yeah, okay.

>> And you release some sprints, because then you can make change quickly. And so that's our approach is that we don't try to bake the cake all at once. We take small releases to try to get them right, and then we add on to the technology. And so through that agile methodology is where we see success.

Dave Vellante

>> And then constantly ask yourself, is this still headed in the right direction?

Dave Vellante

>> That's right, because you may have to make adjustments. Yeah.

Andrew Kemble

>> I really think it's about empowerment too. It's about empowerment and direction and knowing where it is that you want to go, right? We want to make things better. Well, what do we want to make better? Here's the experience that we want to have, right? Our focus right now at GreenShield is very much about how do we make that patient experience better? How do we help healthcare? Well, that's a great vision to have. It's a great vision to march towards. So every step we take in that direction is a step in the right direction.

Rebecca Knight

>> Indeed. Well, that's great. Andrew, Al, thank you so much for coming on the show.

>> Thank you.

Rebecca Knight

>> Really great conversation.

>> Yep. Thank you.

Rebecca Knight

>> I'm Rebecca Knight. For Dave Vellante, we're going to take a quick break and then we'll be back here with more from UiPath Fusion. You're watching theCUBE, the leader in enterprise tech news and analysis.