The Shareholder Is the Member

we've been thinking about so much of the stuff that you shared with us when we last spoke. And just to kind of recap, we were excited about this idea that credit unions had kind of a unique opportunity in terms of AI adoption because they don't compete with one another. and as you pointed out, there already is some infrastructure in place for sharing technology. Also, credit unions exist to serve their members, which makes them unique too, you pointed out to us that one of the major stumbling blocks is that capital is a problem, even the really well-funded credit unions don't necessarily have enough money to start building technology and we've been working with a credit union in fact, focusing in on knowledge management, which is something that we feel like is pretty important to AI adoption. what would happen if a credit union really succeeded with knowledge management and then was able to share that information with other credit unions. And pretty soon all these credit unions could even have a shared pool of knowledge management that could improve offerings across all credit unions. would that maybe start an effect where you could have different credit unions kind of working in different areas of AI, working on different types of automation or implementations of the technology and then sharing that. And so you almost have a whole cohort of credit unions chipping away at the same problem and sharing information.

So I think what you describe is absolutely feasible in terms of a more organized structure where you you I'm gonna make this up. Divide into regions. 'cause they do operate in regional corporates, meaning, you know, the West Coast they have you know, bankers' banks. So they have corporate credit union that serve credit unions. and they have regional, whatever you wanna call it, CUSOs that work together and talk together. So you could be very specific we can make this up north and west, you know you're gonna be super knowledge experts and develop tech for customer experience. And then over here you're going to be on the right side, east side gonna do everything that is to do with underwriting. 'Cause underwriting is another one of those things that constantly change and it's very difficult. It depending on your risk level, you wanted the bear, the type of products, very complex. and you know, you could definitely partition it that way. Another one you can do, I don't know, talent management, successful succession planning, building your grassroot talents through. How do you really, really bring talent into this space? 'Cause people don't know about this space. I didn't know about it and I was in banking or finance forever. So was not aware, like, Credit, what's that? I know, shame on me. But it would be really great if there could be a catalyst and maybe AI could be your catalyst that drive that need because this is their opportunity to perhaps really work together and get ahead of the commercial side of things. you know, like I said earlier, credit unions are very bound on mission, The Customers and the shareholders are the exact same people. So they have a natural orientation to align the profit conversation with the service conversation. You don't see that in the other side, right? Your profit people are your biggest owners, your institutional owners in the very for-profit side, and your customers. may not own a share of anything you do. So, it's very disconnected. It's two separate groups of people. if there is such a group that is willing to take the charge or a small institution, this could be kind of a turning point if if we all work together on this one.

in our space, in like in our world, a lot of people talk about the people will do the judgment, the AI will do the activities, And it's critical that humans are responsible for the judgment. not because they're better at it, because it's to serve us, right? It's key to keep humans at the center of the judgment. and from the structural standpoint that you're talking about, from how how these organizations are structured, credit unions are structured for better judgment by humans because because of what you said, their goal is to not deliver profit to shareholders, it's to deliver value to members, which makes their the human judgment as part of the system, which is a huge part of the knowledge, right? Is the judgment.

Yes. And it's a very interesting flywheel. You know, the f the flywheel if you think about it, the credit union flywheel is not too different than the commercial flywheel that Jeff Bezos talks about. The only difference is it when it when you make a profit and you circle back, you are actually within the same population. You know, when make a profit, you're you're supposed to, you know, service your customer with your with your products. They like it, they come back. You make money, you improve, you lower your costs, you improve your product, you proliferate, you get more customers and that's that flywheel that he talks about all the time and everybody, you know, have adopted it in all the MBA classes. But credit unions are the same, except it's the same population. Customer shareholders are all in there. I think when I look at the flywood sometimes, you know, the shareholders are sitting on the outside a little bit. 'cause it

Yes. It's not a flywheel, it's a facade of a flywheel because i because y you've

It's

got this gerbil that's fine that's spinning in a circle, but there's somebody like counting the money as it's flying off the flywheel going faster, faster.

Yeah. I like that. I like that analogy. You know, that's a that's a really good point because when you introduce a product in the for profit space, you wanna get it as quick as on the market as possible. First to market, it may not be perfect, it may not exactly serve all the things you wanted to to to serve. because you have a fiduciary responsibility to investors, like the people that gave you the money. now I sometimes I think about it on the not for profit side like a credit union. You don't have that push to do something for the sake of something you borrowed. You actually are doing it for the longer term of your members. Is this what my members want? Do they really need this? Maybe the answer's no. Maybe what they really need is something more simple, maybe not so fancy, but maybe really functional and easy to use. You know, so you go for that rather than I just borrow a ton of money or I got investors that that's gonna push for a return and they own half my company. And if I don't deliver, I'm gonna be sold. You know, you have a different type of pressure. Now you don't have a lot of capital. You only have what you earn. You can't raise capital. The only way you can get more capital is maybe if you merge with a larger crediting union, you combine your retain earnings and then you can then be more synergistic literally with your investments, right? Because you are investing in the same things. So that that also makes you more prudent in your development efforts.

That makes sense. you know, AI doesn't doesn't need the kind of motivation that humans need to basically burn electricity and and burn energy. Like we we are designed to highly moderate our energy usage and so know competition is a tool to motivate humans to do more things faster because we're naturally throttled internally, right? You don't need to worry about throttling us. We when you do nothing, we throttle ourselves. and W with AI, a lot of people don't understand that if it wasn't for telling AI to stop typing in the window, it would just keep going and generating and generating and generating like literally there's mechanisms within the system to say stop talking. It would just keep generating forever and ever and ever. so you have this like inverted problem, which is optimal stopping. and so you have to start thinking about a structure for AI to operate in that that stops it, competition is great to motivate humans. but in a credit union space, with AI, you kind of have a better mechanism for s for slowing AI down, I mean the the slow is fast kind of way, meaning fast is break things and then turn it off. So it really is slow. Slow is do it carefully and actually release something.

There is a bit of opposing forces in what you just described that we shouldn't ignore. So you do have the careful, deliberate thinking, all good intention, credit union innovation streak, I'll call it that. And then you have the for-profit, fast pays a lot of talent, optimizing streak. Though their purposes may be different, but they do force each other to do things that the other side would not consider. Right? So the credit union, it would force them to move a little faster, take a little more risk. On the for-profit side, maybe the social conventions will come back and say, hey, you know, you really should think about a more economic way of building those data centers. Do you really need water to cool? Are Are you investing enough to discover? We will discover. We will figure a way so that your token that you call is not very expensive and it doesn't consume as much energy. The data center constraint is electricity, one of the biggest constraints. And they've done a lot of innovation to pack compute power within a constrained environment. They've done that already in the last five to ten years with the different chips that you hear about sometimes. The same technology and data center and instrument is there. The question is, are they pushed to just make capacity right now because the demand is so high? And what can we do to incent them to spend more time so that it's more I use the word green. so that even if you let it run forever, it doesn't consume that much energy. We will get there. our innovation streak over the human race is this way. somebody will discover something and we will figure out another way. so I think there's there are benefits on both sides.

Maybe it's a little imbalanced now. We need a little more credit union influence. But you're right. one kind of pushes the other and the and maybe maybe it is two halves of a brain in a way.

in an ideal world they would actually come together and work together like normal people and civilized and and kinda share learning. I mean, think about it, we we always think about shared learning with just the credit unions. Well why can't the the credit unions have a shared learning experience with the for profit? Both sides can learn something.

that points to another interesting question about incentives, it's not hard to imagine that if if Chase had some major breakthrough in AI adoption and was able to release some products that really changed customer experience, all their competitors, all these other big global banks would would have to quickly try to catch up. maybe one credit union gets knowledge management right and that spreads around. But are there other levers that can be moved to help spur adoption in in the space?

there there has been a trend in the last ten to fifteen years where the smaller credit unions that cannot scale have gone away. So if you look at the number of credit unions versus the total asset side, the total asset side grows normally, but the number of credit unions have shrunk quite a bit, like significantly more. So what you're seeing is a lot of mergers, etc. that in itself has I think driven some of the at least the more the folks that have the capacity to go ahead and scale to be more aggressive, to compete with the banks. Think of the banks as their competition. so today I think last time I quoted a while back when I was there twenty years ago, ish you know, we were six credit unions with six percentage of the markets there. I think today it's closer to nine. still small but slowly growing.

And that's due to I think what you said, like exposure to having more products to offer, correct?

Yes, and being more competitive and stepping up that tech roadmap, you know, getting mobile banking on some getting your app out there. You think that was easy? It's not easy. and you know, I it's hard to think about what incentive would work. I I hate to say this word, but you know, people use the word socialized banking out there. Kind of is a bad thing. The word socialize, you know, it's a bad thing. but in my in my opinion, personal opinion, everyone should have access to banking. Everyone should have access to banking. Not as a free thing, but it's money. This is how we live in this world.

Yeah, like you think of your your child, like one of the first things that you have to teach them so they can be self sufficient is to get a bank account and ha yeah. And it's and often it's a credit union and

Great, you put your first five dollars in there.

it becomes a member of the family, you know, that credit union, yes.

Yeah, so you know, is there an incentive for both sides to come together and really address that? What if I want everyone to have banking? What does that look like? What technology do we want to share? for profit's gonna wanna make money, which is fine. You know, 'cause that's what credit unions do. We usually buy technology or we run subscriptions, we you know, buy licenses. And we very seldom do any legacy building. Very seldom. So why not? Now, okay, tax exemption is a big thing. Credit unions don't pay taxes. That's the big whatever you want to call it. anger the banks have on the credit union side because we sell they sell typically the same product, they have the same service, they have branches, ATMs, they s they have lesser fees and all that stuff. make money, but they don't pay federal tax Banks do. So could there be a tax product that could be out there that would be attractive for banks? to do something with a credit union. Kinda like affordable housing. I don't wanna get into that tech side, but they use tax credits specifically to incent private investors to invest in affordable housing. That's why you can do affordable housing.

Interesting. yeah, this would allow that capital centralized investment into the technology or change.

Yeah, and share at a much lower cost, right? if there is an incentive for the banks that would sell or service their technology to the credit union side, knowing that it's a not for profit, or non profit, excuse me, at a much lower cost, is that competition or is that something different?

Right. Interesting.

So I'm not so sure.

And this is why I get excited about credit unions, because it feels like the right model for AI. because if now we have to worry about slowing down, then you'd be like, Well, credit unions are better designed for slowing the pace than speeding it up. So you have something better designed for slowing something down than speeding it up.

I have a interesting thought for you on that one.

Sure, yeah, please.

So instead of slowing down, have we addressed what is lacking in AI, which is the social context? It lacks that social context, it doesn't have the ability to create those fuzzy boundaries that humans can create. maybe the question isn't to slow it down. Maybe the question is are there ways to build and learn social contacts within the AI environment so that it doesn't follow just what you want it to do linearly. it surrounds itself with enough social context. So you have to teach it that, right? So maybe this is where you can see the two industries juxtaposed, which is one will learn social context from the other.

yeah, I think yeah, it's a really good point innovation brings us slower, more careful AI. We need to go forward in innovation so that AI is more careful. What we see with hugging face and what happened is it it needs to be more sophisticated. We need to go forward through this period of danger i from an innovation standpoint. and yeah, the speed I'm more referring to is organizational speed we need more innovation to get better simulations.

Uh-huh.

so that we're running simulations before we're putting it out in the wild. So yeah, that's a really good point. I I don't make that distinction enough when I talk about people slowing down that I'm I'm really talking about slowing down the deployment of a a solution that was created by some team in your organization and not properly tested.

And I think that's where knowledge management actually comes back into it in an interesting way. And and Mona, maybe you can help us with this question because if we go back to that example we were talking about earlier, if there's one credit union that really figures out knowledge management for their credit union, obviously some of that information is going to be unique to their credit union. But I would think within the credit union space, there might be also a big portion of that information that is similar across credit unions. the way it's structured is similar. And, you know, not only does that allow information sharing to improve, but it might make it easier perhaps for a credit union to now offer auto loans as well, because they could borrow the information they need from this knowledge model that is sharing information across all these credit unions. And and I think the other thing that does too is it feeds this other thing we're describing where then automations aren't leaning as much into individual models and using their reasoning to to power a lot of what's happening. They're able to actually create agent memory from from these knowledge models. So the model becomes more of the interface and it's figuring out what the customer wants and then you get to use the verified information to decide what to do. And then the model once again is there to kind of feed it back to the user. I mean it's not that cut and dry, but I think you start moving closer to that, which feels slower and safer in a way.

Yeah, or like you said more I prefer the words more context. let's say we were successful and we built this knowledge base of everybody's head within the very certain sector. can we use that to train AI to learn that social construct? And what would that do? What would happen?

in our world it not to say this is just sort of for us. I'm not sure that everybody else agrees with us, but we sort of think of knowledge as like a long term storage or long term memory. And then context is a feature reduction or a reduction of that knowledge to something smaller to do a specific task. So you kind pull that knowledge into something all the relevant memories just the context for the system to do that one task. and what seems to be popular out there is everyone just building context randomly in all of these different agents, but not having a centralized canonical knowledge base that they're drawing from. So now they're creating a a bunch of knowledge systems that are unmanageable and get outdated and like what you know what's worse than no knowledge is the wrong knowledge, right? So in our minds you have this like this canonical knowledge piece that's kept up to date and then you have these tasks that are done where you've reduced and and you've queried that knowledge and pulled what's relevant. Because in our brains like yeah we have the knowledge but then we have our short term memory where we pull the context to do something. and so I think, where Josh is talking is like if you have the wider context that's shared, then other credit units can then pull from that and reduce it, combine their context and then have tasks done without each one having to manage the broader context of like let's say policy or law or regulation.

Yeah. And I think the feedback would also be very, good. You know, so you're talking about big brain. Gigantic brain holding all this knowledge. you know, knowledge are interrelated, you know, even though we it distilled them down to a very specific block. they're not mutually exclusive to all the other things around it. there is a connectiveness to it. And so the feedback coming back, did it work? Did it not work? New information from the interaction is actually very critical to refeed that. And you know, then then technically, theoretically speaking, you would naturally auto enhance each time you create a new product.

Yeah, it's it's a fantastic comment and something I just add a little to. In our brains when we read, we always write, even when we're not writing. You know, we rewrite our memories every time we retrieve them. so some people underestimate how often that feedback loop should close. Even just simply that one agent or person retrieve the knowledge should go back into the knowledge system so that it knows that that knowledge is being used.

Yeah, I agree. Yeah, no, I agree. I you know, sometimes the description seems to feel a flow out only and we never talk about that reflow back in, that it comes back in it has new information.

Yeah, and that's part of what makes the work so complex is that is that everything has to be logged and verified and accounted for. And the deeper you go into it, the more snapshots of what's being remembered are being taken, and then those become part of it too. So it it does start to grow in complexity as you're building it. Which

Yes, it it could I could see what storage is going to be so maybe invest in storage.

Yeah, the the t the toughest part that we had to tackle was like it's one thing to store data, that's simple in in today's world. I mean simple-ish in comparison. and we have this concept of data becomes information, information becomes knowledge, and knowledge becomes wisdom, right? And it's it's this idea that A number starts as a piece of data, right? And it's a number. And then it's information when that number is a birthday. And then it becomes knowledge when that birthday says you're old enough to have a bank account. And then it becomes wisdom when you can start offering them products that make sense for where they are in their life, right? And it all starts with data. So a piece of data And managing data is much easier than managing wisdom. Because you're like, what products are right for where they are in their life And so when you try to manage the canonical, meaning the source of truth for knowledge, ideas, you're starting to like try to hold ideas and And have a source of truth for an idea. Like is this or is this not a good product for this person at this phase in their life? Has to exist in one place in the brain. It can't have ten different converging ideas, right? and so once you start looking at at keeping knowledge up to date, to your point, the feedback loop, it also has to know what it knows and know what it doesn't know so that it can update itself on wait, this no longer is a good product because they've changed the law around this product for this age, etcetera. So it gets interesting as we start realizing how much we just take for granted as humans that we just do in these jobs.

It's true. know, you flip a light switch, your light goes on. that's it. Do we know how works? No. We take it for granted very much and I think knowledge is the same. I think we don't know where it began or it began somewhere long ago or somewhere far away and so to to do what you describe to me sounds amazing and kind of a large undertaking.

It is. What what I like about it, it would makes it less ominous, is NASA did it and they succeeded. And they did it without AI. and so I feel a lot of hope 'cause I've walked with Roger through this in great detail, you know, as we've like gone through how they did it with humans, how they did it successfully, and how they broke it down and and then we basically applied it, just taking the humans and saying, well, okay, what if we automated you know what the humans did at NASA? And there's a couple of things that make it less ominous and less big feeling. and the the two things Roger did was first of all, like just find the the role in your organization that depends on knowledge the most, and and focus on them. Don't focus on all the knowledge of your organization. Just like who which role uses knowledge the most? And then take more of a training curriculum approach to and what knowledge is most critical to them. And then just try to manage that. And then you just sort of chip away role by role at who depends and you sort of grow it organically over time instead of take all of our documents and and dump it and and that's how they pulled it off and and it worked quite well.

how do they like account for people's instincts and making decisions with minimal data?

I mean again, they the feedback loop, as you pointed out, was probably I think Roger would say was one of the biggest challenges that they had. he has a story of how just something as simple as two engineering teams, one on metric and one on imperial caused a huge disaster. and saying like better knowledge management would have avoided that, it is key that feedback loop. is key because otherwise instinct is something you have to trust on the end and it never gets fed back to the system. So you you can be feeding the wrong information over and over and over and over again and have no awareness that the system is feeding the wrong information. And and that was I think he would say one of the hugest struggles. And I'm not sure he would say that they solved it, but something AI can do a lot better at is pulling back that feedback loop and and feeding it back to the system. So so I mean, gut feel is is more context. And so I think y you're s you still need a person at the end, right? That that ultimately makes the judgment. This is where judgment comes in. You're not trying to remove the judgment. You're just trying to provide provide better context to the people.

I think that's a very important distinction. You're providing context. The judgment is not necessarily dependent on that context. Depending still on the human. I think that's that's a big distinction to draw. The human has that other thing, intuition, instinct, judgment. Some have it really sharp, some don't have it, but in different areas. And we as a society of humans haven't figured out how that actually works. You just you know, some people are you just really in like you just know. I don't know how you know. You just know. And we haven't figured that out. that's very that's very interesting. I I would love to hear when people talk about AI to be more deliberate about, you know, what AI actually does. What does it actually provide you? Because they say we replace humans. Well humans what?

Yeah. Well there's an interesting thing in this yeah, in this judgment let's look at context, environmental context, not necessarily mental context, right? there's one really key piece of context that we have to keep in mind that probably won't change for a long time. And that is that when that person makes that judgment call, they are ultimately taking responsibility for the decision. And that person has something to lose. They have a paycheck, sometimes a big one, depending on importance of that judgment call, like how impactful that might be. and and so when they make that call, they have something to lose. AI has nothing to lose. AI has no job to lose, no paycheck to lose, they can't be sued for getting it wrong.

They don't have a guilty conscience.

So irrespective of whether it can make a judgment call, that doesn't matter because it can't take responsibility for it. And therefore, humans will always have to be there to take responsibility. Some human with something to lose has to f say, I will I will be on the hook for this decision. Which makes us like reading the decisions and approving them and And the the fear being like, Did you read it or did you skim it?

You know, knowledge management is a very human endeavor, even when you're doing it with AI. And his point too is that at at NASA, like they they needed a culture shift so that people realized engineers, anyone, that not saying something could have fatal consequences. 'Cause at NASA it could. If you if you felt like something was not right and you didn't say something and something went wrong then you were left holding that. And so, know, it's a culture shift, which is I don't know if credit unions are better at this than other organizations, but just the ability for people to speak up and the incentive to have them speak up and like a culture where you want people sharing information and talking about what they're seeing, because it helps other humans, but it also helps knowledge management.

And this is where AI can really help because at least it's been shown in a lot of our work that people are more comfortable speaking up to AI than than their boss.

know, okay, this is one of those funny questions. If everything was paid two dollars, what would you actually do? You know, 'cause you're worried about your boss 'cause I don't know, you're the breadwinner or you're financially depending on this job, must most of us are. You're gonna have constrain within for your own self interest. Right? So you may make a safer decision for yourself than may not be the right decision for the company. That happens all the time. that's why we don't speak up, because you know we'll get fired. then the question is I think you one of you said it, what incentive what is the right incentive that allows you to provide that feedback without a negative consequence. But still though hold you responsible, right? Because you don't want to s just say things. you wanna be very mindful of the feedback you're providing. That's a really difficult thing to navigate. What was the incentive? Because you know incentive can go the wrong way. just like AI, incentive behavior that you necessarily don't want to continue.

Yeah, it's interesting, like I'm I'm I'm just thinking about this for the first time, but if my AI is constantly flooding me with decisions that I have to take responsibility for, so my job's on the line, whether it's an incentive, you know, based system or a punitive based system, either way, why wouldn't I just say no to everything? Like why would I just not be like I'm not gonna take any risks? Right. Yeah

No no no Need more information. That's the feedback. Isn't that what you do when you don't want to make a decision? I don't have enough information.

Right, right. Go back and

Yeah, yeah.

and get me more information.

Yeah, and I've seen that loop many times in the corporate world.

Mm-hmm. Yeah, you're you're right. Like the there's a balancing now of incentive you know, with responsibility that you're taking on these things to get people to do things and not just say no to the AI constantly.

That's a tough conundrum, you know?

AI adds another wrinkle too, because especially if we're talking about knowledge management and implicit knowledge, the kind of knowledge that people hold in their heads, where's the incentive to train a model on that if the feeling is that that model's gonna be used to replace you? So that that becomes another challenge. where I I think, we have yet, I guess, to see it fully flower, but you could imagine that if you had a knowledge management model running in your organization that was doing what it was supposed to do, then you would have a lot of new choices to make about what you wanted to do with it. you know, a for profit business might want to use it to get rid of as many people as they can, but you could imagine a credit union space saying, like, well People like us because we have people working here that they see all the time and there's a community and they like that they can call and talk to somebody. So let's use this vast knowledge model now to just make them better at their jobs instead of trying to replace everybody. But how do you get people to believe that going in that that's what you're gonna do? Mm-hmm.

Yeah, that's a tough one. That's a tough one. I I sometimes I wonder, you know, economically there is a efficiency, right? When you go through the human productivity thing and whatever you save on human labor you reinvest and blah blah blah. There's again that other another flywheel. there's a different flywheel where when you do hire more people, you do generate also economic benefits within that particular community and we don't talk about that very much. You know what I mean? Like if you're working, you are a contributing part of the society. You're generating income and you're spending expenses. Hopefully, you know, they they match each other and you're not creating a debt for yourself. When you take that out, where did that capital go? Where did it go? And then, How long does it take for it to come back to that community for it to be productive? Sometimes I'm not sure people thought about that. Maybe some economists have, maybe I could get a PhD in there and really study it. But I think there's something there that we don't talk about in the AI space is that okay, fine, if you really wanna do this, reduce your staff by half and have all the task. But how does it come back? How does it come back to the society? Does it just go in someone's pocket? Or are you reinvesting in the society so that you recreate different jobs, new jobs, new products that actually enhance you know the human cycle? I'm not sure. I think that maybe too non capitalistic conversation, but we don't have that conversation.

Yeah, so Joshua Gans has an interesting approach. and and I I wouldn't say this like is spot on to what he's saying, but it's sort of adjacent. but I can give you a a a fairly recent example that we had where one of our customers wanted to automate and use AI to get a sense of how happy their customers were, so they could, do something or act upon possible churn. and so they're like, okay, let's let's automate this, right? Let's automate how we might measure success of customer satisfaction through their conversation data, right? That they've had with customer support and whatever else. So we go through and we identify like this this huge swath of customers that are like you know potential churn, right? And then we step back and we said, okay, did we just reduce work for people or did we just increase work for people? Because these people were just gonna be ignored. They were the company was gonna do nothing. There was nobody who had to do anything. They were just gonna churn and no one was gonna have to do anything other than maybe, you know, close their account. and now all of these people have all this work to do to try to see if they can try to prevent this churn. and you know, like some of it might just have to be human to human, like a phone call, you know, that says, Hey, I just want to see how you're doing or or to make sure that the next time they called in that information got passed to someone who could, maybe just dig a little deeper or find out, you know, w whatever the thing might be. So it's funny how much how much work AI creates. When you actually are on the front line and you realize knowledge people c say knowledge is power. Knowledge is also a lot of freaking work.

That example actually. You know, sounds like a positive experience for that customer who was who left because of whatever reason.

Yes. And and negative for the employees who's like, my God, more work I thought we were

But that's

supposed to automate our work away.

Well that's interesting, right? Is it more work or is it just different work?

What's your point? More revenue. Like we they keep a customer less churn. It's in the end it's less work, yeah.

It's a good thing. Yeah, it's a good thing. You you change y you know, you you what you used to do may not be needed. What you now need to do is very important. It's just different.

Yeah, 'cause the cost of acquiring a new member and the effort is so much greater than just Yeah.

Very expensive. Yes. Absolutely. Interesting, huh?

Well almost points too to all to AI potentially creating jobs, right? If it's identifying more opportunities to fine tune an experience and the existing system isn't supporting that, then maybe maybe more people need to be brought in to work on these things. Maybe it's not something that can just be automated away.

Yeah, maybe like you said there so there may be, you know, maybe they automate this one thing, but there is a new segment that's going to be born off of that efficiency that may require different and brand new investments. I think we're all waiting for that right now.

Yeah, and that idea of like not going necessarily bottom up and starting with use cases and going to knowledge, then what happens is the the data drives the use cases instead of the use cases driving the adoption path. So you're like, okay, now we know where our biggest opportunities are to improve our business, whether it's automating something we already do or not. then you work down from the knowledge and data into the use cases and it just makes so much more sense.

you can have your own credit union think tank. this will be very different, very interesting.

we've been talking a lot about service design we actually talked about it with Don And you know, when we think of credit unions and I think in this conversation, in our last conversation, we've talked a lot about how they they come to feel like a member of the family, right? And part of that Probably is because you're a member and like a credit union doing well means your money does better. But there must also be like services baked into a credit union that that give it that feel. That you would if you were bringing AI in, you would be very deliberate about how you handled them. You'd have to look at them and say, like, what specific aspects of this service can we automate? And what aspects do we not want to touch because we don't want to taint this feeling of the credit union being a member of the family, 'cause that feels like a very precious and potentially precarious thing.

Yeah. that is an excellent point because this this is the industry where when you do go in a credit union it's a different feeling. When I go into the bank, even now, I am intimidated. I don't know why. Like why? I don't get it. Like the and

I feel like they don't want me there.

like you know, like like yeah, I wanna get out. I wanna get in, get out as quickly as I can. Like I don't I don't wanna be there very long. but when I do go into a credit union, I don't have that overwhelming intimidation feeling. I'm like, okay, I'm cool. I can stand in on, I'll talk to them like a normal person and it'll be just fine. And you know, I don't understand why that happens to me. It's a mental thing, it's it's what you associate with it, and that's just the way it's been. So if if and when they do invest in AI, which I know they are. I think that is a very important thing to preserve. It's that feeling of ease. It's ease. It's okay. Come on in. We won't eat you alive. It's okay

Or kick you out. Like I I always feel like my bank doesn't want me to come in. You know, they're like they're trying to get me to not come in. So I'm like, Okay, is is it okay if I came in here and asked you to do something in person?

Don't you feel like you always get scold by the s teller? I wanna deposit some money. Did you fill out deposit slip? Okay, where is that slip? why don't you just use the ATM? Okay, where's the ATM? Like

Right. Yeah, it's what do they call it? What did they call it? rationing by friction, they call that. So you don't come in the bank,

rationing by friction, yeah. Yeah, yeah, yeah.

they make you fill out these slips, so you're like I okay, forget it. Yeah.

I'll just use the HTM.

Right, yeah. So I like always it's always a scary thing going into I don't like going into banks. And I work in that industry, so

that's cool. That could be that could just be another focus area, right? You have one what one credit union working on knowledge management and then another credit union really focusing on service design because like where

Yeah, yeah.

where knowledge management is like this key kind of foundational piece, service design, the more we keep digging into it, really provides a blueprint for for what an organization can do. And That blueprint also is a shareable asset within credit unions because a a lot of it is is the same.

Yeah. Yeah, service design is just like another shape of knowledge, just like we have data. Service design is just another memory system for an agent and it becomes their objective. it's a great state management system for an agent for a set of agents to be like this is your objective is to facilitate this journey successfully.

but that's a very different objective you just said. Rob, you said you're your objective, Agent, is to facilitate this journey as best as you can. Right? Rather than you need to Get off the phone as quickly as you can, answer the question within thirty, whatever, five seconds, and you know, your average call time cannot be longer than a minute and a half. That's a very different two d very different set of objectives.

Yeah. Yeah, I'm I'm waiting for the token rationing to come where like you're gonna call and you're gonna get like

My way.

yes no answers from the phone. You know, there it's no hello, how are you? It'll just be like what? One token, okay.

That's the I was gonna joke that's the New York style banking. Just kidding. What do you want?

Yeah, exactly. It's like g

was good.

all right guys. Well this was great. Thank you for joining us. I I loved our research call and love this conversation too.

It was super fun.

Thanks, Mona.