If there’s one word that’s on the lips of credit union executives everywhere, it’s AI. AI is repeatedly being talked about as the way forward for credit unions, but many misconceptions about the technology persist. Those will have an impact on crafting your credit union’s AI strategy.

Host Sarah Snell Cooke, co-founder/publisher of The Credit Union Connection, talked with Great Lakes Credit Union COO Elizabeth Osborne about her credit union’s AI strategy and how they implemented it. They also discuss Great Lakes’ AI assistant Olive and how other credit unions can go about creating an AI strategy that helps create efficiencies and better serve members. Watch!!! —->

Read the full transcript:

Disclosure: Transcript is automatically generated

Sarah Cooke 00:11

Hello and welcome everybody to The Credit Union Connection. I am Sarah Snell Cooke, your host. I am here today with Elizabeth Osborne, welcome.

Elizabeth Osborne 00:23 Thank you for having me.

Sarah Cooke 00:25

Thank you for joining us. Elizabeth joins us from Great Lakes, Credit Union. She is the COO there, and she's going to talk about AI strategy with us, which I know so many people have so many questions, and there's, there's misunderstandings and misinformation out there, so somebody who's actually doing it, let's talk to, let's talk to Elizabeth about that. And so first I want to kind of back up a little. I mean, I know the conferences I've been to this year have all been AI, AI AI, as like every other word you hear walking down the hallway, but it does feel like there's a lot of misconceptions about it, or misunderstandings and, or just the fact that it's a buzzword, and we don't really know the practical reasons why we would want to have an AI strategy. So my first question for you is, what is AI strategy?

Elizabeth Osborne 01:13

Yes, happy to share so it's I've had a lot of discussions on this in the past, and I'm going to share with you what I've heard, and then also our approach. So I've heard a lot of credit unions have taken the approach of, kind of tying it into your technology strategy, right? AI is a technology, it kind of makes sense starting there. A lot of people are setting, standing up a separate AI strategy, and others are doing what I'd love to start doing in the future, which is creating an overarching innovation strategy, which is, which really one, AI, is one of the drivers of and then that overarching strategy, innovation strategy, aligns with your business strategies, your technology strategies, etc. So that helps to ensure it's broad enough where it's not just one area, because it's not, it shouldn't just be in your technology department. There's so many benefits across the organization, both for your employees and your members. By having a broader approach like that, you can ensure you, that that actually occurs.

Sarah Cooke 02:19

Yeah, and that's, I mean, it may be one of those things that helps break down the silos at some credit unions, too, everything. So that's, that's wonderful. Thank you for that. And so, can you talk a little bit about, specifically, what is Great Lakes' AI strategy?

Elizabeth Osborne 02:35

Yeah. So we started down the path about a year and a half ago, really focusing more on the customer or member service side of AI. So if you think about AI strategy, it typically follows like a few different tracks or paths. You have the member service, you have fraud detection or deterrence, and then you also have lending decisioning. And actually you could say there's a fourth too, which is more the RPA. AI, such a broad term as your you and I were discussing earlier, more the automation and efficiency side. So, for Great Lakes Credit Union, we started more in the member service. We, there was such a need. Our members were kind of struggling to get us on the phone and to get the right level of support. And so we identified that member service is a channel where, by incorporating AI in a solution with that, we could reach more members more effectively, and at the times they need help. So 24/7, which is a great benefit. Absolutely,

Sarah Cooke 03:42

I love that you mentioned too, the RPA, because, you know the the Chat GPT, you know the member facing side gets so much attention. But you know those efficiencies, they may be less sexy, but certainly just as important, if not more important, they're the foundation for everything else. Very. How are you all doing that? That piece of it yet?

Elizabeth Osborne 04:03

We are, yes. So we actually started with Microsoft Power Automate. So a great RPA solution out there. There's a lot of different free, open source softwares that you can leverage, but Power Automate, we've been able to apply to a few different areas that tie into that member service. So help streamline receipt of information and then the output of it, also helps to alleviate some manual efforts for my deposit operations team, which reduces errors and then frees them up so they can spend time on stuff that that we really need them to focus on more the service side and doing some of the cool things we're doing at GLCU instead of data input and verification. So that's was how we started. We're expanding that as part of our ongoing strategy, our ongoing AI strategy, to continue to identify areas in which RPA automation can help. Help do just that, streamline, automate and ensure it's more accurate.

Sarah Cooke 05:03

And especially, you know, right now, incomes a little bit snug. Those margins are little tight. You know, being more efficient certainly can't hurt. So, um, you guys started integrating your AI into online and mobile banking a couple years ago with the chat bots. And from what I understand, last year, you guys introduced an AI assistant who's got a name. Olive.

Elizabeth Osborne 05:30

Yes, Olive.

Sarah Cooke 05:32
the first, how'd you come up with Olive? That's the first question.

Elizabeth Osborne 05:35

Yeah. So Olive, we actually engaged staff across the organization. We did a survey. We're like, hey, we have this really cool conversational AI agent who's going to help assist our members, but we need your help to figure out a name. So we sent out the survey. We got a ton of responses. They varied a bit. Some were, you know, like a GLCU bot. Others were a name like Olive, Lily was another one we looked at. And then it took some back and forth between myself, Steve Bugg, our CEO, and Patrick Basler, our CXO. And we determined by listening to the names and, in the solution. So Interface AI solution, we asked them for sound bites, and Olive was by far the best. It just brought a sense of calm, too. So,

Sarah Cooke 06:30

yeah, exactly what you want with, you know, dealing with, you know, financial situations, yes. And so, you know, Olive is kind of next gen, if you I guess, allows for natural language conversations a little less robotic. I mean, can you talk a little bit more about the the advantages of that? Yeah,

Elizabeth Osborne 06:51

absolutely. So where we started with Olive and where we're at now is quite different. The beauty of of Gen AI, so generative AI, using LLM is that she's constantly learning and becoming smarter, and so we're finding that her response and interaction with members is becoming smoother. It's, it's still clearly not a human, alright? And that's okay, alright, because all of all of this works, 24/7, so I don't know any human that can do that. Yeah, exactly. We all, you know, want to sleep and you know, take breaks. PTO is important. So, so Olive is constantly getting better, but we're seeing just of recent since we we upgraded the platform to a new version of Gen AI earlier this summer, and really seeing some of those responses, and just she's answering more, she's able to assist little more. So that's been a huge benefit for our members, too.

Sarah Cooke 07:54

And you mentioned LLM, I just want to make sure, large language model. Correct. Thank you. I want to make sure everybody knows all the terminologies and stuff and just explain that a little bit, if you would quickly.

Elizabeth Osborne 08:06

Yeah, absolutely. So large language model is really about gathering information and learning from that, right? And so there's different variations out there, but for that, you know, our use of it is a little different, because we you'll see some like Open AI, for example, like Chat GPT is tied to the entire internet, right? So they took a download of the internet, it does this often, right? And kind of learns from it. For Olive, she's very, a little more segmented, where she has access to our core. And so our core, through use of very predefined channels and information that are fed into her, she then learns that. And so it's a bit more structured than what you'd see for some of these more open models that are out there. So it does vary. It really depends on some of the guardrails and controls that you implement when you introduce a solution like that. And so

Sarah Cooke 09:05
before we get on to the benefits, I would like to talk about the guardrails. What were you all considering and what did you land on?

Elizabeth Osborne 09:12

Yeah, so we wanted to make sure one, is that, and I think this is important when any financial institution is vetting partners, is that you go through a very structured, well defined process. So utilizing our vendor management process and treating AI as a critical vendor ensures that you have the appropriate checks and balances for us, it was also important that one guard rail's that is more the reputational risk. You know, one concern I had is I wanted to ensure that Olive isn't going to, you know, have a conversation with a member and say something inappropriate or incorrect, right? So guardrails by limiting the data sets the types of information provided. Thourough monitoring an QA constantly for that solution, just like you would hold any employee accountable to the conversations are having with a member. We do that with Olive and so those guardrails are key to her success. It's not like a plug and go and you just kind of forget about it. This is something that you have to really keep a close eye on both in that initial due diligence before you pick a partner, and then ongoing maintenance and support monitoring to ensure that the AI solution is meeting the needs and is performing at the level of what you would expect for any member facing employee

Sarah Cooke 10:39
And who is the partner you have used?

Elizabeth Osborne 10:41

So we use Interface AI, so Sri, their CEO, he's been he's so visionary. He's been a fabulous partner to work with. We vetted quite a few different partners and competitors in that market. You know, at the point of time, I'm not as close to the competitors now, so I can't speak to current, but at that point in time, Interface AI was was well ahead with the development of this particular model, and I like that it was, also included a chat feature, if that's something we would like to switch over sometime in the future. Currently, we use chat through a different provider, but with Interface AI, the conversational agent, that virtual assistant was well beyond, and the data segments that were available to us more out of the box, meaning we did not have to really configure too much. We were able to integrate it into our core, refine it to ensure that the responses are on brand and meet GLCU's needs, and then testing it. Because of that and the well established nature of their solution, we were able to stand it up in seven weeks, which is, yeah, unheard of, yeah, and who's and who's your core to? Just, we are Jack Henry Symitar. Our clients, we have, we're one of their first 100 so we've been on that a long time, they're great partners, and it was very easy to turn that on with Interface. Ai,

Sarah Cooke 12:06

so what are the benefits that you all have seen as you've turned this on? Have, like, record of hours saved by your staff, or, you know, things like that that really demonstrate the ROI that you're looking for? yeah.

Elizabeth Osborne 12:21

So prior to introducing Olive we had a solution called telephone banking, non AI. That solution provided members with the ability to make you know selections on the phone and gain some kind of limited information, like your account balance, last five transactions, something that effect. At that, right, prior to our launch of Olive that solution was supporting, fully supporting, less than 25% of all calls that came to our call center. By introducing Olive, and Olive sits at the carrier level, meaning that our Primary, our our main phone number, the first interaction a member has when they call our main phone number is with Olive. We have seen an over 200% increase in self service calls since introducing Olive. So currently, during business hours, she is fully servicing, meaning that the, the employee, the member, does not need to speak to employee, because they receive the information they need. We're seeing an average of 63% of all calls self service, compared to less than 25 and then also, in addition to that, after business hours, because Olive is 24/7. That is averaging above 70% so members are now we're starting to see more calls on weekends and after hours, because they know they can have a conversation with Olive and get the information they need. So that channel has been wildly successful.

Sarah Cooke 13:52

Yeah, no, that's amazing, and especially those after hours. I know you know, when this is a, I'm going to date myself, maybe, but the Home Depot cyber attack occurred, and my card got caught up in it, you know, and calling to to speak to someone at that time, and nobody, you know, you were put on hold for 20 minutes because everybody was calling in, and then you you got hung up on eventually, because, you know, you were on hold for so long. And so, yeah, I totally get that frustration, and how that must be, you know, it's a solution to those after hours or over, over, what's the word I'm looking for? Over, yeah, yes, the overflow, yes, yes, to help reduce that. So, yeah, that's awesome. And so what are your future plans? Where are we heading with this? Yeah, so

Elizabeth Osborne 14:46

you know, at Great Lakes Credit Union, we are very focused on growth and expansion to continue to offer services to our members. And so as we continue down the path of M and A and looking and helping credit unions that, that there's a lot of credit unions that need assistance. So as we merge in those additional institutions, you know, I see great opportunities for Olive to help onboard those members. So whether it be outbound calls to remind them that their new debit card arrived, or to, to remind them of a location or to send them a text where they can find information to log in to online banking. I think there's great opportunities for mergers and acquisitions to use a solution like Olive for that. But then I also am in conversations with, with Interface AI, as well as other partners on areas like fraud. You know, fraud is rampant throughout the industry, and any place where we can find a way to detect and prevent before it becomes a fraudulent event is really a gold mine for anyone that is running a bank or credit union. And so AI can do just that, because they they learn about, you know, behaviors and members. And so one partnership we have is with Defense Storm. So Defense Storm, we use them for fraud monitoring and cyber monitoring, and they do a really great job of detecting events today. So if we could have some type of AI overlay with that, whether through Defense Storm or someone else, to even better, detect it and block it. That would hopefully decrease costs and the very negative member experience that we have when no one wants, it's a terrible feeling to have fraud on your debit card, money leave your account because account takeover anything that affects so anything we can do to prevent it is a great opportunity. And then we're also continuing down the path of looking on how we can lend beyond the credit score. So, you know, we are a low income designated credit union, so we have a lot of members that either have very limited credit history, or their history needs some improvement. So if we had the ability to use the wealth of data that we have access to as a financial institution to say, okay, maybe your credit score isn't where it should be, but you've had 12 repeat payments to your utility company, you've maintained a positive balance of your account for nine months, whatever the case, right? Whatever those factors are, using that data to make better credit decisions presents a huge opportunity to provide assistance and help to our members that need it. So we plan to continue down that path. That's

Sarah Cooke 17:37

excellent. Yeah, the more we can help more members, especially those who really need it. Truly need it. Yes, is really good, good way to serve the mission of the credit union movement. So what are some of the best practice recommendations you would have for other credit unions who might be looking to enhance their AI initiatives?

Elizabeth Osborne 17:56

Yeah, so I'd say, you know, all of us are in strategic planning right now. Some have already finished. Good for you. We're in, we're in the heat of it right now. And so I would take a hard look at your strategy and say, Okay, if we, is there an opportunity to introduce AI to provide an even better service to our members. I really feel strongly that if credit unions and all financial institutions do not adopt AI in some way, you will get left behind, because this is the future. It's really about how, where does it fit best for your institution? You know, if you are, if accuracy in efficiencies is a, it's a huge opportunity for you, because the staffing or hiring the right talent, then RPA is probably a great way to get started. And so I'd say just kind of look at what's partners that are out there. There's so many great solutions. Do your due diligence and really identify what are those top strategies that AI could help us propel forward. That would be my best advice. And then just make sure, you know, talk to other credit unions. That's the beauty of our industry. We're so open to talking to each other. I came from banking, and I love banking, but that was not the case. It was like, you know, publicly traded can't really share secret sauce. Credit unions, we are so open to share information. So leverage those contacts and connections. I talk to credit unions all time about AI and different solutions that are out there. So I encourage you to do that we learn best from each other, and that will help propel the industry.

Sarah Cooke 19:29
Yep, love it, especially the cooperative nature. And so yeah, with that, I'll let you have the final words. What are your your final thoughts for our audience today?

Elizabeth Osborne 19:39

Yes,

Sarah Cooke 19:40

all about AI,

Elizabeth Osborne 19:41

so I have a very cool announcement. So Olive is now bilingual as of this week. So she speaks Spanish, and so we're interested in the Chicago land market, very large Hispanic population, and Olive is now able to have a conversation with that portion of our membership base that, it's in their preferred language. So I'm really excited to start seeing some data and see that adoption occur. It'll take some time because, you know, got to get the word out, and members may not know that Olive can speak to them in Spanish. So I'm really excited to introduce that and also to continue to look at other ways to expand different languages that would meet our members needs. So that's, I would say, that's the best news we have at Great Lakes Credit Union right now. Just yeah, keep an open mind. Talk to your regulators. You know the NCUA is very open. They will partner with you. So don't be afraid to pick up the phone, send an email and ask question on hey, I'm looking at this option. What should I be concerned with? So keep that open line of communication, but don't be afraid to to make a connection and find that best solution for your organization.

Sarah Cooke 20:57
Awesome. Thank you so much. Elizabeth, I appreciate your time today.

Elizabeth Osborne 21:00

Yes, thank you for having me.