Building a Private AI Brain for Procurement with Shannon Copeland

Shannon Copeland (00:00.28) Good. Russ Fordyce 🏆 (00:02.398) Hey, everybody. Welcome back to the Winner's Circle. Today I'm here with Shannon Copeland from SIB. They are the latest AI Excellence Award winner. And we're here today to talk about AI on the supply chain and in procurement. Shannon, welcome to the Winner's Circle and congratulations on the award. Shannon Copeland (00:18.616) Thank you very much. Thank you, Ross. We're excited. Russ Fordyce 🏆 (00:20.714) So before we get too deep into the AI pitch, you guys have been at this a long time. You've been doing cost reduction work long before AI became part of the conversation. Tell us a little bit about what an SIB looked like maybe five years ago and what you guys were doing. Shannon Copeland (00:35.862) Yeah, we've been in business for 35 years. We are a product of a roll up. There are so many firms that have combined with each other. over the last five years, we have combined seven companies in the procure to pay as a service space. And one of my clients advised me, and so I will take that advice. And they said, OK. when you say procure to pay, describe what it is you actually do. And we actually do the procurement. So we go on behalf of our clients. And it may sound really boring to some of your audience here, but we actually go and do the procurement, meaning we negotiate with vendors and supply chain partners. We do that in a, I'd like to say a loving way. It's not acrimonious in any way. Our clients, allow us to go and take our experts in those dirty procurement spin categories and educate the vendors and suppliers what the real needs are for each of our clients. And that ends up helping the clients become more competitive. You can say saving money, can say reduce, you can say it's reducing prices. 99 % of the time, it's a lot of things, not just price that's involved in that. helping our clients become more competitive. So that's what we've been doing for 35 years across as many as 70 different spin categories. Russ Fordyce 🏆 (02:10.282) Well, that's an important role. I've bought a lot of stuff in corporate America and in business. it's one of those things, you know, it's kind of like home buying new tires for your car. You're never sure you're getting a really good deal. You look all over the place and they all seem to have different options and different variables that make the price, you know, something different than the last guy. And imagine having kind of a central figure to kind of organize that just has got to be that alone's got to be, you know, kind of monumental shift in how they operate business. Shannon Copeland (02:42.412) Yeah, I have an interesting way to brag on our procurement experts, or at least I find it interesting, but our clients do too. And that is some might believe, and I know you don't Russ, cause I now I know your background, that you can just call one of your vendors and say, give me 10 % off. And it turns out that that's just not the way the world works. First of all, the vendors are very big, very sophisticated. They are interested in having a deep relationship with their customers. So if you want to more competitively procure from your vendors and suppliers, you have to speak their language too. So you really have to know that space really well because there are so many different bundles and pricing. rubrics that it really pays to have an expert on your side on your team. Some very large companies have those experts full time. I mean, this is maybe an overused example, but if you're a car manufacturing firm and you're buying steel or aluminum, you have those people on your staff full time doing an amazing job sourcing those raw materials. But when it comes to some core spend, and some indirect spend, what's often referred to as tailspend, almost no company on planet Earth can afford to employ experts across all the different spin categories. So that's where we come in is we help clients because we employ those spin category experts full time. So it's. Russ Fordyce 🏆 (04:23.144) Yeah, I think it makes all the sense in the world to me. takes that kind of that stress out of you're trying to build a relationship with a vendor and then you just start, you not nickel and diming them for a better term, but you start talking to them about price and, know, it's it becomes a very acrimonious relationship. and now all of sudden, you're not friends again. Shannon Copeland (04:38.318) Correct. Shannon Copeland (04:43.854) Correct. that can really, our clients are most concerned with, you know, disruption or the avoidance of disruption. And so we make sure not only is there no disruption as bad of an English as that is, but we make sure that things get better, their operations become more competitive. And so sometimes that might mean a vendor change because another vendor looks at things differently. They have a better offering. They're more innovative. And on our best day, we're helping vendors become more innovative because they hear about different needs and service offerings from our clients. And then everything gets better from there on out. Russ Fordyce 🏆 (05:30.527) Well, and I imagine when you go through the work, I remember, you know, we outsourced part of our customer service overseas. And as part of that project, you had to have your entire process really mapped out. as it so was, we didn't, right? So we spent a good six months really mapping out processes and making sure we knew exactly how stuff was going to work, because then you had to ship that over to somebody else. Shannon Copeland (05:50.776) for us. Russ Fordyce 🏆 (05:56.969) And I bet it's the same for you guys as you're really kind of narrowing down what is it that we need out of these guys? What is it that we're trying to do in the business? And let's not overspend. Let's not go buy Salesforce when you need a database or something. Let's narrow it down. so you almost just that work alone has got to be able to narrow your spectrum and narrow the playing field of vendors that can support exactly what you need. Shannon Copeland (06:21.258) Exactly. Yeah. It's a good point. The other story our clients share with us is that when they don't have an expert procuring whatever. Shannon Copeland (06:37.058) They will also note that often they have purchased something from a vendor from a very young person with a rate card. And of course, that can lead to overspending and overbuying. The worst part from the vendor's perspective is that the vendor doesn't get feedback from that client. So the vendor doesn't really know enough about that client. And thus, they don't even have the telemetry, for lack of a better way to say it, to learn from that client. And so we, in some ways, you know, we go above that 20 something with a rate card and we start talking to people who are interested in feedback from the client. So it really is a, it's why we've been successful. We, we sell about 900 new engagements a year and every one of those engagements, we help the client quote unquote, become more competitive and we build, we build a deep relationship with those vendors as well. And so they'll take our call the next time. Russ Fordyce 🏆 (07:44.159) Yeah. Now, as you said, you've got 30 years of data and experience in this market and helping people with their supply chain, with procurement. Now you've come into this AI world and developed SpendBrain. Walk us a little bit through kind of what SpendBrain is and kind of how it's transforming that model. Shannon Copeland (08:02.06) Yeah, so I'll stick with hopefully the unusual way to describe it. And I'll start with, you just said the word data. So historically our industry has said things like we have lots of procurement data or we have a benchmarking database. Largely those days are over. We no longer say we have a benchmarking database because the industry has historically built a benchmarking database by harvesting the pricing from every engagement. In other words, harvesting the pricing data from each client's interactions with its vendors. That no longer is acceptable. clients do not want you to harvest their pricing data, even if it's anonymized and aggregated. So we do not do that anymore. Now, the knowledge of what's happening in the market with regard to pricing, and then I'll move away from the data piece just for a second, of course resides in our procurement experts. So our procurement experts, because they're negotiating 20 to 50 deals a day, they know what's happening in the market. And so they can advise each client, each subsequent client on, hey, do you have a good deal or do you not? But of course the permutations, again, much more than price, the permutations are the variety of what is purchased from each vendor across businesses is so different that even if you had a benchmarking database, it doesn't do a lot of good. That's one, and the reason of course I say all this is because AI has made clients keenly aware that data is extremely valuable and that it is private to them. So now what do we say? Now here's what we're doing with Spinbrain. The first unexpected thing that we're doing with Spinbrain is we're not LLM based. And I'll repeat that because customers may... Shannon Copeland (10:22.166) or your audience may or other clients or your colleagues may say, wait a minute, they're not LLM based. We use LLMs, but they are no better than the backup singer in the band. The lead singer is an ontology. So we build privately a semantic ontology for each customer. So each client gets a private brain that no one learns from, including us. So we can contractually commit. to wall off their brain even from us. So we do not learn from them as well. We do use an LLM in the stack as the quote unquote backup singer, but that's only to vectorize the queries. So that way we can say to the client the following. So this is spin brain in a nutshell. Hey, name your own brain. We had a client name their brain orange, too many syllables, but that's what they wanted to name it, which is wonderful. Name your own brain and we will build this private instance for you for say orange and it will be accurate. we're, because we're not LLM based, it's non-hallucinatory. There are a lot of syllables there as well. And you can begin conversing with that brain verbally, even in multiple languages as needed. You can converse with that brain. starting in about 30 days and you can have back and forth conversations with that brain in about four months. So very accurate, very low cost, very fast to deploy. What do back and forth conversations mean? Meaning the client in four to six months can start teaching the brain as well verbally. And so for example, if you wanted to inform the brain on how to calculate something properly. So we take in both unstructured data and structured data into that brain. And so they can begin doing forecasting and budgeting as well. Again, non-LLM based and walled off from the world. It's that simple. Russ Fordyce 🏆 (12:37.15) Well, yeah, well, it makes sense, especially given your clients. You guys are working in public sectors and healthcare, higher ed, and these aren't typically, you know, the bleeding edge technology companies. They're usually, you know, highly regulated. What made these environments kind of the right place to build this? Shannon Copeland (12:56.023) I think we just checked all the boxes. And again, 90 % of your audience will just shake their head and go, yep, that's what's expected. I want to go back to what you said earlier. We do work for companies that are $50 million in revenue to $5 billion and above in revenue. We do. But many companies across that spectrum, they're already so sophisticated. And to your point, let's take health care, for example. Healthcare is a pressured industry, statement of the obvious. It's a different industry. They really do only think about patient outcomes. They really do. And of course, they care about the health and wellbeing of those frontline workers, but they also know that everything does have to be private. And so I think by having this basic idea that clients should have their own brain at very low costs so that they can get operating leverage from tech finally. We can go into that in a minute. We just kind of naturally check those boxes. Yep, private. Yep, we don't learn from you. Yes, here's the other element. Can I ever take it back? Can I get rid of Shannon one day if for whatever reason I'm tired of a Southern accent? Yes, you can get rid of me one day and take it all back. And that's a big deal. It's hard to, I won't name names, but. If you fill out your data in a typical SAS platform, which is a bunch of empty fields, sure, you can always extract that data out. We're offering the ability to do much more than that. You can keep your ontology and truly get rid of us if you want to. And of course, our hope is that we're adding enough value that a client never gets rid of us. Russ Fordyce 🏆 (14:54.174) Well, yeah, I would think the procurement piece alone is worth its weight in gold. Now, in my experience, again, coming out of marketing department, sales departments and tech companies and telecom companies, you know, I would buy stuff from a vendor and I'd stick the contract in a PDF on my file share and I, you know, and then I'll send over an invoice over to somebody. It seems like data is all over the place. I mean, I worked for giant companies and there was no central repository. There was no kind of central format or anything and invoices coming in from all over. And it's such a big fraud, waste and abuse area for these enterprises. How do you collect all that data and messy data environments and get it all kind of indispensable? Shannon Copeland (15:37.25) Well. That is a beautiful question. And we have to sort of nudge ourselves to make sure clients are not overwhelmed by our proposition, to your point. And that is because clients' data is everywhere. I mean, there are all these fancy terms. my client, my data is federated, meaning it's in multiple systems. Part of our secret sauce is we can take the data in any form from anywhere. So we can say to the clients, and we do this every day, hey, leave all your current systems in place, all of your data stores in place. We will API into all that data, pull it in in a disorganized way, and then we'll reconnect it in our brain. We have a little bit of an advantage in that we've been doing this as an AI team for 17 years across three other previous companies. So I brought my whole team here from Accenture. And we've had a lot of experience. We have a little bit of an unfair advantage in that we know how to take in that dirty disorganized data, process it really quickly, connect it all and build that ontology. And then of course, start that iteration of humans in the loop. It sounds like a little bit of a marketing scheme, but we just trademarked this term handcrafted AI just to make sure people understand that if you don't, our clients particularly, if you don't have humans in the loop, even if you're non-LLM based, it's still not going to work. You've got to keep the experts in the loop and that includes us and our clients. So our clients become a part of that human in the loop supervised learning teaching source very quickly. And so we're both working on that brain and making it a lot better. Russ Fordyce 🏆 (17:31.339) And now I always, you know, I've talked to financial guys, know, finance teams for my entire career, because we always like to know who we're spending money with because we go try to sell to them too, right? And I know when I talk to the financial guys, you ask them kind of like, how are cost controls going? Everything kind of good. And, know, the good ones always are kind of like, well, maybe not as good as it be. But most of the CFOs or financial guys I talk to are, oh, yeah, everything's great. When they get you know, spend brain for the first months in that first four month period, what do they see that they didn't see before? Shannon Copeland (18:05.036) Yeah, that's a great question. We had a CFO say to us the other day and we checked it. Of course we did the, you we went out and did some research on our own, but the CFO said, you know, I bet we haven't done a top to bottom audit of all of our spend because these, some of these companies are big and old and some of them are doing well. When you're doing well too, that's kind of a, you kind of, referred to that a minute ago. When you're doing well, you might have less of an incentive, but here's what the CFO said. She said, I don't think we've done a top to bottom audit of all of our spend in over 15 years. It turns out when you do the research, you never hear, you never see in the research months or days, you see years. It's been years since companies actually do a top to bottom. of every contract, every invoice. So imagine, to your point, imagine this, what our offering does is we're doing that every day in real time, every day. We're looking at every contract, every invoice, every detail. And my joke, and I'll just, this will be fun for this audience. And I do say this to some clients because they love it. We build a C3PO for your spend. It never forgets anything. speaks multiple languages, if you will, speaks multiple spin category languages at the intersection of every industry. That's important. So in other words, a hospital, and we love the hospital clients because it's such, it's easy to talk, to understand what you're really doing there and who you're really serving the patients and the frontline workers. But the hospitals speak a different language. And so if you're building a C3PO for a hospital, it needs to speak their vernacular so they can use it quickly without changing all the underlying systems. So 24 seven a day per day, you know, it's, it's learning. Here's the best part. Clients are very sophisticated. All of them are. Here's what they say. I don't even want to have to ask it questions. I want the C3PO to come back to me when it sees something wrong. Shannon Copeland (20:27.5) So it's true, you need to ask it questions for a year to be simple. Six months a year, everybody asks questions and then sure enough, that brain will come back and say, I found this today. clients really understand what the power of this can be. They just have to trust that it really is gonna work and that it's accurate and low cost. Russ Fordyce 🏆 (20:48.916) Well, that's what we all want. We want that proactive brain out there. was, as I mentioned, I was in telecom for a long time and there was a whole industry kind of around billing errors and kind of what you're doing to the procurement and making sure that your telecom bill was correct. And I also remember getting literally an AT &T bill on a pallet once. We were buying wholesale at the time. there were a lot of calls, but... Shannon Copeland (21:12.227) Yes. Right. Russ Fordyce 🏆 (21:17.046) The fact that you had to go through that and then trying to figure out what was a billing error and all of that. And I imagine that vendor billing errors are still a significant part of what's Ben Bray's catching. How sophisticated is that problem on the vendor side? Shannon Copeland (21:31.95) Well, there are all kinds of stats. So we have our own stats because we're processing a lot of data every day. There's a lot of research stats. I do believe that this quote around 40 % of all invoices across all categories have errors. That stat is holding true for us internally. 40 % of some kind of error. It doesn't necessarily mean that 40 % of invoices have a cost reduction opportunity or an error, you know, where they're being overbilled or underbilled, but there's some type of error. And our job, of course, now with AI is to detect all of those errors. To your point, vendors, when they discover that we're involved with a client, they are very interested. Vendors, we are not acrimonious with vendors. Vendors employ a lot of people. They feed a lot of families. They do want to have more accurate bills and more accuracy in a deeper relationship. So we're helping, we have vendors that have also adopted SpendBrain, by the way. That sounds strange, but we have vendors actively seeking us out to adopt the same capability for them so that they can have an accurate view of what's going over to their clients. So it's a, the world is catching up fast. Russ Fordyce 🏆 (23:03.24) Yeah. Well, I imagine you guys have built this kind of co-creation model with your customers, which I love. I always love that. You're building this for them with their institutional knowledge and working through the process with that. You mentioned earlier that they own the data and data ownership and explainability seem to be kind of the two biggest terms I'm hearing outside of human in the loop with AI companies. How central is that to kind of what you guys are doing? Shannon Copeland (23:14.902) Yes. Shannon Copeland (23:28.14) Yes. Shannon Copeland (23:32.312) For sure, I love that explainability term. they obviously own their raw data, but they also will own that ontology or brain that we build for them as well. We know that now with a lot of clients, we do have to meet with the heads of AI now with clients and educate them on how we're doing this. And thus far, Every head of AI has approved us to get involved because of these factors of, hey, it can be yours, you're not locked in. Explainability, and you might also mean traceability there. Every verbal query comes with an answer, and those answers are traceable back to the detail. We even have a pattern around how we highlight that source of data, that backs up that answer. So we do that. And here's another perhaps unexpected expectation from our clients. Because they know we've been in business 35 years and we have these experts, they don't want to suffer the delay of us not understanding their industry. So we work for a lot of large multi-site firms. and the chief procurement officer of a national very large auto dealership that's, that's they're booming. They've been very successful in a very tough industry. That CPO said to me, well, I expect that, that you're going to know a lot of these terms already, or you're going to know how my industry works. So we do come to the table. We know the vernacular of a lot of industries, 11 industries. And they don't want us to be more expensive while we learn that new industry. So we call that our industry ontology. And so they can ultimately keep that in the end. Again, if they don't want to be locked into us and they are not, they can also, quote unquote, have that industry ontology in the end as well. Russ Fordyce 🏆 (25:52.917) Well, it's got to be very kind of, I don't know, heartwarming when you kind of go into these large institutions and they haven't done an audit in 15 years and they don't know kind of what's going on. And I imagine that kind of that first report you guys pulled, there's some light bulbs going off. Give me some examples of kind of what you guys have seen operationally when that system kind of surfaces something that large, that's been invisible for years. Shannon Copeland (26:19.67) Yeah, I think first of all, our spin categories in the tailspin and even in the core spin space, we love the boring. And the audience in our procure to pay space, they don't call it our clients. They don't consider themselves boring. They just consider themselves fastidious, if you will. So I couldn't point to any of the 70 or 100 spin categories and make it sound sexy. So I'll pick the most exciting. I'll pick the waste realm. you know, restaurants and retail organizations, even healthcare, one hidden spend that is actually an enormous spend is waste. So recycling any kind of waste, medical waste, et cetera. So with one client, They were spending about $9 million a year all over the country in waste. And we turned on SpinBrain and within two weeks, we worked on it for 30 days, just to be sure. We identified over 1.5 million in annual savings. And I won't go into all the ways that we discovered that savings, but back to me bragging on our experts. We think about crazy details when it comes to just to each and every spin category. And it often has nothing to do with price. It has a lot to do with what is purchased as well as price. so, but yeah, we found a lot of errors. We found a lot of non-market competitiveness, if that's a way to say it. And that's available to your point. It's available to... almost every client, 900 new engagements a year, about 450 net new clients a year, and there are savings in every one of those engagements. Russ Fordyce 🏆 (28:28.412) and you've delivered kind of these substantial annual savings across your client base, that really kind of means that you kind of pay for yourself, doesn't it? Shannon Copeland (28:38.52) For sure, for sure. Yeah. And look, even though the largest consulting firms in the world have now over the last, five to 10 years turned to this contingency model, whether in whole or in part, 35 years ago, 10 years ago, to your question earlier, six years ago, we were still 100 % contingency based. Clients now are more sophisticated than that now. They don't want to pay 100 % contingency. So we're now more fixed fees, more services fees, more predictable fees, but we do see 6X to 25X or more on what we do just because look, we don't want to do everything for a client. We want to stick to our knitting here, work in the supply chain, procure to pay, sourcing space with our expertise and just be that weapon for our clients. Russ Fordyce 🏆 (29:39.06) Well, I think the pairing of kind of advanced AI and your human kind of involvement is really kind of central to that approach. How do you decide what the machine handles kind of autonomously versus what kind of the human has to come back and validate have that human? Shannon Copeland (29:55.702) Yeah, statement of the obvious for most of your audience. The spend area, procure to pay, supply chain, whatever that word cloud is, sourcing, we're fortunate in that the data is relatively, I call it laminar, the data is relatively clean. Yes, we do, we deal with unstructured data, like in the contracts or policies, and then semi-structured data. that's in the invoices. But by starting with that data set, there is a lot that we can do with quote unquote, I'll call it advanced technology because AI has become kind of this term for a lot of things that are not AI, of course, as you know. that one thing we said to a client is we can probably take the mundane work and reduce the time spent almost to zero on the mundane work. That way, our clients' teams can do more strategic work. I know that's an overused phrase, but it's true. So every one of our clients, even the most successful clients whose stock price is doing very well, when you start to talk to those clients underneath, they are hard at work. trying to build the next thing or compete, develop a new product, develop a new customer experience for our QSRs, our restaurant clients. They're trying to develop that next new menu item for the next quarter. so their clients have said to us, they'll keep all their current people. don't even, it never comes up labor savings. What comes up is can I take, my experts that I already have in house and have them do something totally different and more competitive. And we, it almost brings us to tears. We love it when they say that because when they see what we can do where, look, our, as we say, our clients know all the questions they want to ask. They really do. They just can't get the answers out of their disparate data. And so we're helping to kind of crack that open. And I keep saying low cost. Shannon Copeland (32:20.344) Cost matters to everything, it really does. And so they don't wanna come in and spend millions of dollars a year on something like this. They wanna be able to do this in an affordable way and so that they can get back to the strategic nature of their business. Russ Fordyce 🏆 (32:36.18) Yeah, I think I've heard that from a lot of kind of AI leadership is that this isn't about reducing workforce. There's definitely that's going to happen. unfortunately, a member of my family was part of the Oracle, you know, where they were Oracle, you know, I guess, write down whatever you want to call it. You know, that that was really a business decision where, you know, they had to raise cash and take that money and spend it elsewhere. And what I see AI doing now more is enabling people to be more productive. Shannon Copeland (32:50.754) Yeah. Right, right. Russ Fordyce 🏆 (33:04.936) And as you say, kind of get rid of that task work that I was doing every day. And now I can focus on something strategic. How am I going to build a better product? How am I going to get out to a customer? How am I going to satisfy that customer need better than I am today? And AI is certainly a great enabler with that. But it's really changing the whole economics of how knowledge work gets done across pretty much every sector. In cost intelligence specifically, what becomes possible in five years that's not possible today? Shannon Copeland (33:33.378) Yeah. Well, and then I'll marry that with finishing the answer to your previous question. And that is the both the vendors, the vendors are innovating every day. They're coming up with new offerings. The clients, because they're innovating every day, they have new needs. And so you have to have an expert and we're hiring pretty aggressively right now, even though You know, we're private equity owned and we're not venture capital funded, which means we have to make money every quarter. We actually have to hit a number every quarter and make a profit. But we are hiring because you have to have that human in the loop. And again, like I described earlier, the client's expert is in the loop. We do not want the client. Please do not. We say to clients all the time, please do not get rid of John or Sally. We need them on their side, teaching the brain, and then us on our side, because that turbulence between need and offerings, that'll never go away, thankfully. And so I think five years from now, you're going to see some pretty amazing things. think that these capabilities, like I've described, a quote unquote spin brain on the vendor side and the client side, you'll be able to use those brains to talk to each other, for sure. And that's not a new thing. That's out in the domain already where you have agents interacting with each other. But again, we see no possibility of totally removing the humans from that, the experts from that. One joke that we've said internally is, if we even say this to our children, You know, go get a job and understand some industry really well. First, go, go something physical, whether it's in supply chain or, you know, you know, something in the vendor community, go understand that industry really, what could be software, but, but, but then, then you'll sort of empower yourself or outfit yourself to play in this crazy AI world. Russ Fordyce 🏆 (35:52.459) Yeah, it definitely is an accelerator on your own expertise if you have that expertise. And I know it's been liberating for me because it gives you so many new capabilities you just didn't have time to learn. I didn't have time to learn to code. I wish I would have when I was 17, but I'm not anymore. So I've got other things to do. And so you're almost limited because it was like, oh, I could go do that. And that's an investment. Shannon Copeland (36:05.678) Correct. Shannon Copeland (36:12.504) Right. Correct. Russ Fordyce 🏆 (36:19.274) But now it's less of an investment and it's a little bit of an investment in time. But now that I can send my agent out to do these things, it's really liberating. It's really liberating. Shannon Copeland (36:28.494) Correct, especially if you've probably read about how, you're thinking quote unquote, LLM only, how the LLMs are, whether this is true or not, but it's out in the attention sphere, that the LLMs like to sort of confidently support your question by giving you answers that may not be totally right. And so of course, In our model, we avoid that with our clients, of course. It's got to be correct and accurate. But I guess my point is that, you really do need to be an on-the-ground human-to-human expert first before you can start learning from these. And again, the LLMs are going to evolve and become better and better over time. But yeah, it's a crazy new world. Russ Fordyce 🏆 (37:25.768) Now, if we've got CFOs listening out there who are like, I probably have some cost leakage out there that I'm not catching, other than calling SIB and Spendbrain, what's the first thing they should do this Monday morning if they know they've got something kind of leaking? Shannon Copeland (37:43.232) I would say to a CFO, follow your gut because what we're finding is CFOs, heads of supply chain. kind of referenced it earlier. They really do know their data. They know it. They intuitively know where problems are. Just go attack that one piece of intuition on day one. Use the people you have because it's better for us, frankly. What's exciting What is the most fun we have on a given day is when we talk to a client, say a CFO, of course the intuition is there, the questions they have are there. Where they think the problems are, that's already there. But what gets even better than that is they've already experimented a little bit with AI. They've done something already and they've said, you know, I kind of got this to work, but it's not doing it completely. We had a big state of California institution describe it this way, and I'll be really brief. I won't name the name. But what they said was, is they had bought a bunch of stuff. They bought four five different things to solve a really big problem, about a $45 million a year problem, just for one location. California is enormous, of course, as we know. Bigger than Texas, for sure. Russ Fordyce 🏆 (39:11.412) Fourth largest economy in the world, right? Shannon Copeland (39:11.79) Correct. And what they said to us is we've tried a bunch of things and we've spent a lot of money and we still can't get the answers out of all of this capability we've purchased. And we said, wow, wow, first of all. And yes, we're not just saying this, you're missing one element. You've got to like invest enough into building your own quote unquote brain. in order for you to be able to understand that data that you have. You can't just throw disparate systems out of it. So again, some clients are very, very advanced already and don't need us. But even those clients, when they realize, no, we're trying to be very pragmatic here and low cost and also move really fast, even some of those clients, they'll choose a firm like us. Scrappy industries is my favorite phrase. Russ Fordyce 🏆 (40:06.942) That's amazing. Well, I think that's where we are right now. it's, you know, a lot of people are figuring out what we can do with this. it certainly seems like you guys have done that in a pretty significant way. So congratulations on all the hard work and effort. I'm sure it's paying off on the bottom line too, but congratulations. Shannon Copeland (40:26.574) Thank you. Yeah, thanks for paying attention to us. We're excited.