What the Largest Restaurant GPO Sees Coming for AI

Matt Wampler: On today’s episode of the Restaurant AI Podcast, I am joined by John Davie, founder and CEO of Buyers Edge Platform. John started the company in 1998 at 21 years old, going door to door convincing independent restaurants to pool their buying power. Today, Buyers Edge is a leading digital procurement network, serving hundreds of thousands of food service locations, with more than 40 acquisitions. And what John believes is the largest, most diverse data set in the industry. I love talking with John. He’s smart, decisive, and always looking to the future. As a company, they’ve been working with machine learning for over a decade. But post-ChatGPT, John has personally been in the weeds and is building his vision of what AI will become for the industry. We get into it all, including the upcoming launch of CollectivIQ, the first step in the creation of the world that John is bringing to fruition. But before that, a quick thanks to our sponsor, ClearCOGS. ClearCOGS is the leading provider of decision intelligence for restaurant operations, turning the data they already have into practical daily guidance for prep, ordering, and labor so that operators can make better decisions before the day gets away from them. Now, here’s my conversation with John Davie.

Matt Wampler: Well John, I’m excited to have you on today, man.

John Davie: Yeah, I’m excited to be here. Thanks for having me.

Matt Wampler: All right, so just to start, can you walk me through Buyers Edge? Because every time I go to NRA, Buyers Edge, giant booth, and then you guys have like AI attached to it, booth there, you’ve got the emerging fund, you’ve got your tentacles all over the industry. So walk me through. What do you got going on?

John Davie: Yeah, it’s both a blessing and a curse. We have a lot of awesome stuff going on. We’ve made a ton of acquisitions of different companies and brands. We go to market under a lot of different brands and labels and teams. But the curse to that is it’s hard to keep everybody educated on everything that we have happening. So we just try to refer to ourselves as sophisticated instead of complicated. But yeah, it’s why we gotta go to these events.

Matt Wampler: I like what you did there from a marketing standpoint. It’s we’re not complicated, we’re just sophisticated.

John Davie: Yeah, you like that? Yeah, exactly. Doesn’t really solve the problem.

Matt Wampler: Well dumb the sophistication down for me. Give me the plain speak. What is Buyers Edge as a whole?

John Davie: All right. So we really at the core of it, it’s around digital food service procurement. Starting way back in 1998 when I started the company with my father, we set out to help small independent mom and pop restaurants get the same buying power, leverage, and advantages as the large mega chains. My father had started companies similar in totally unrelated industries, leveraging buying power together to drive down pricing. And so we wanted to apply the same thing to help small independent restaurants. Now we didn’t have a bunch of buying power to kick us off to go get the contracts with all the vendors to drive the better pricing. So we had to go to the restaurants first. So we had to go kind of chicken and egg. We went to the restaurants and said, hey, share some information with us, work together with us, join up, can’t save you any money today. But we’re gonna work hard and collect everybody together and we’re gonna eventually drive down pricing and get you better pricing.

Matt Wampler: Well, it’s one of those businesses where once it exists, it’s amazing, right? It’s hard to compete against, but how you get from A to B and get all of the restaurants on your side, that seems like a monumental task.

John Davie: It was tough. It was five years of just grinding it out, knocking on doors, convincing restaurants to team up with us. I was twenty one years old at the time, so didn’t have a lot of credibility, no resume, just out of college. So didn’t know anything about the restaurant industry. So yeah, it was tough. My father would just limp in, a thousand dollars here, two thousand, out of his retirement account to keep us afloat and allow me to pay my five hundred dollar a month rent. But eventually we got there. Like I said, it was tough. It was a long long road.

Matt Wampler: The nineties, five hundred dollar rent. That sounds nice. Twenty-one years old, fresh into your career, and you’ve got dad as a boss. What was that like?

John Davie: Well, I was actually the president from day one. So I was technically on paper his boss. He didn’t want to run the company. He just wanted to invest a little bit and be on the road selling. He loved to sell. So he liked to get in front of restaurants and convince them to come on and he taught me to sell. But we’ve been great partners. He’s still a partner. I mean, he’s eighty four years old, so he likes to come to board meetings and contribute where he can. But we’ve been good really fifty fifty partners from the beginning.

Matt Wampler: Why’d you do it? I mean, that seems like an interesting choice of all the things to create.

John Davie: Yeah, so my father has a passion for helping small businesses. Like I said, he had a couple other companies helping small office supply stores, helping small plumbing electrical distributors. I grew up watching that. I got a passion for it. David and Goliath story, help the small guy compete with the big guy. And back then it was all Olive Garden, TGI Fridays, and Applebee’s that were growing like crazy back then and taking market share and putting the small guys out of business. And so that was the focus and grinded it out for five years. And just to bring it back to your original question, we did learn early on that we needed to gather and understand the data. An operator handed us stacks of invoices that were all coffee stained and crumpled up and said, okay, here’s all my invoices, go figure out how to save me money. So we had to go back home and type them all into an Excel spreadsheet and try to make sense out of all these different disseminant invoices and product names. And that’s what ultimately has led us to being what we say is a digital procurement network, because the data ties it all together, of how we drive savings and value across the whole platform.

Matt Wampler: It’s a unique challenge in the restaurant industry because if you’re a data platform and everybody is on paper. I mean, when I got out of the industry in like 2017, 2018, I still had a clipboard with my invoices that we’d enter. It’s not a very digital industry. How’d you deal with that?

John Davie: Yeah, it was again, it was a big grind. A lot of fat fingering of this stuff into spreadsheets, trying to figure out how to make sense. We hired a developer early on that we could work part-time after his full-time day job and try to build us a rudimentary data ingestion engine. We did that for a long time. Then we met our first company that we ultimately acquired called Inside Track. It’s still a really big part of our company today.

Matt Wampler: When was that, that you acquired Inside Track?

John Davie: We acquired them about fifteen years ago.

Matt Wampler: Gotcha. So you guys have been doing acquisitions for a long time.

John Davie: Yeah. We’ve made over forty acquisitions. That was our first one. And it was definitely needed. They had a whole database, they had figured out how to normalize all the products. When you talk about so many different naming conventions and distributor identification numbers, and normalize those to manufacturer numbers, they had figured a lot of that out and we were paying them a large SaaS fee to help us. And we bought the company and now we’ve been able to really scale that throughout most of our customers.

Matt Wampler: All right, so take me back to the nineties. Your dad had done this in office supplies. Did you guys ever dream that Buyers Edge would be what it is today? Or did it just slowly grow into what it is? Or did you know going into it, like this is huge if we can do it?

John Davie: I think we knew obviously there’s a lot of restaurants, and that’s why we looked at other industries like the shoe store industry and sporting goods and other things, and we just kept coming back, like no, restaurants. There’s a lot of restaurants, and there’s gonna be a lot of restaurants, and nothing’s gonna put restaurants out of business. So we knew the market was huge, and we knew there was a need for it, and we knew we could create a multi-million dollar company. But no, I guess we didn’t know quite the size and scale, we didn’t quite know how much we would expand it. We hadn’t thought about an M&A strategy. We hadn’t thought about going international or Europe. So it’s definitely grown a lot bigger than probably we ever would have imagined.

Matt Wampler: Do you remember the point in time where you kind of realized you weren’t just a buying group, but you were really a data company? Because it feels like that’s what you’ve transitioned into in some ways.

John Davie: Yeah, I mean we definitely started out very focused on sort of the GPO model, a buying group, leverage the buying to drive better pricing. And we still do that. That is still core and that’s how we bring a lot of value to operators. But I would say we were probably eight or nine years in when it really clicked that if we didn’t figure out data at a really high level scale, we weren’t gonna be able to compete with other much bigger GPOs. The large catering companies and large hotel groups had created mega GPOs at that time and we were nascent compared to those guys, because they came out of large parent companies that already had many billions of buying power. We were in the single digit tens of millions of buying power, they’re in the many billions. So we were saying, we are gonna win on the basis of being more transparent, being better with data, better with technology, if we have any chance of catching up to these mega GPOs that are out there.

Matt Wampler: So you’re like 30 years old and you’re gonna go take on the behemoth GPOs. How’d you come out on top? Was it just the data side of things or is it incentives? What changed?

John Davie: Yeah, I think it was one, we had been working on that smaller, independent restaurant. Everybody else was focused on bigger hotel groups, bigger casinos, bigger multi unit chain restaurants, healthcare companies, nursing homes. Nobody really could figure out how to tackle the smaller independent mom and pop. And since we had already been focusing on that from the very beginning, we eventually really clicked and figured out how to master rolling out independents across the board. And once we had tripped over kind of our first billion dollars of volume in that smaller mom and pop space, it dawned on us like, wow, we can move up market now and help a 20 unit operator and then a 50 unit operator and a hundred unit operator. And so we moved up market. So that was I think one big thing, we figured out how to scale many different types of operators where even the big GPOs were just trying to go after the big fish and ignoring all the smaller guys. Secondly was that data advantage, that first acquisition of Inside Track and really leaning into bundling software and data and transparency tools with the GPO business model, bringing those two together. Plenty of big operators, especially like Las Vegas casinos, had no interest in a GPO, were turning down all the big GPOs because they already had hundreds of millions of buying power. They didn’t need a GPO in their mind. But they did need software and they did need tech and they did need to normalize all their contracts and data and their price auditing. And that’s where we brought them with Inside Track. That got our foot in the door. That built a relationship. And then over time we showed them how our GPO contracts could save them money. And so that was a big win.

Matt Wampler: Well, and it makes sense, but I think one of the harder things is like how do you make money? I get that there’s a need to save money on the independent side. If you can get everybody together, you’ve got this giant cohort, but how does the company make money?

John Davie: Yeah, so we have a lot of different ways we make money, but the core way is that we get transaction fees, admin fees, and rebates from the vendors as we drive down and negotiate a contract price, which we call typically a deviated price. We negotiate a rebate, and then we might negotiate another administrative fee with these suppliers. And so that way we can sign up operators at no cost, so that the restaurant chain can join totally for free. It’s a very simple sign up digitally. And then we collect their data, we normalize the data, and then we get the operator better pricing, rebates, and then we get administrative fees and transaction fees from the vendors.

Matt Wampler: All right, so let me get this straight. You guys basically have all the gigantic vendors that have all the money and all the power paying you to go save money for all the little guys, the little independent restaurants out there. It’s like the Robin Hood of business models.

John Davie: Yeah, I like that analogy other than Robin who was stealing. I’d like to think that we’re not stealing from anybody. We’re bringing value across the supply chain.

Matt Wampler: Even better. They’re choosing to give you the money.

John Davie: Yeah, they’re very willingly paying us and saving operators money and it’s a great thing.

Matt Wampler: All right, so you’ve got 40 acquisitions in. You’ve got your tentacles everywhere. You started out as the GPO, you’re on the digital side, procurement, but how far do the tentacles reach? What are you involved in?

John Davie: A lot. So we have multiple GPO brands that go to market for different segments of the market. We call that our digital procurement network. That’s a major division, our biggest division. And then we have our fresh division, which we’ve acquired two companies there and we manage about four billion of fresh produce across a network of about a hundred and sixty local produce distributors and wholesalers. We have a couple hundred grower shipper contracts and we bring a huge amount of value in fresh and transparency and food safety to multi-unit operators, large customers across fresh produce and fresh meat and seafood through a network of regional distributors. That’s our second biggest division.

Matt Wampler: Love that division. Prior to Jimmy Johns franchise, we had our whole supply chain and all of our procurement contracts. But when it came to produce, we were still picking up the phone and calling local places, like what’s your lettuce price today? It’s the wild west.

John Davie: It is. We bring all the sophistication there. We bring a full program, farm to fork, contracts all the way back to the farms, the transportation across the country to a local, say, distributor in Boston, and then contract that distributor in Boston to deliver the last mile, and traceability and food safety all the way through. We manage huge operators in that space and bring a whole food safety team, huge amount of savings and visibility and transparency to it.

Matt Wampler: All right, outside of procurement, what else?

John Davie: And then our third largest division is our SaaS software division. So that’s where we have Arrowstream, our largest enterprise supply chain software. About sixty out of the top one hundred restaurant chains in the country use that tool. So it’s used by many of the biggest operators in the country. And then Inside Track, which I’ve mentioned was our first acquisition, is used by a lot of hotel companies, most of the casinos in Las Vegas leverage our Inside Track enterprise tool. And then we have a full back office inventory, recipe costing, and full-blown restaurant accounting software in our back office set. And then our last division I’ll mention is our supply chain management division, which is hands-on white glove full service supply chain management. Any operator who has, say, a smaller procurement team can lean on our supply chain team to RFP contracts, negotiate specific proprietary deals, and bring them even more value at a higher service level. So those are our four key divisions.

Matt Wampler: I don’t get the sense that you say no that often. It seems like you guys are open to like, if there’s an opportunity, let’s go check it out.

John Davie: Yeah, that does bring me to a very big recent acquisition called Dinova, which is a company that brings corporate diners to restaurant companies. So that’s our first kind of non food service procurement acquisition, where really they are helping the restaurant operator gain very high value customers. Think of the traveling sales rep for Pfizer or Merck or the corporate catering for Microsoft or Amazon. We steer those level of high value diners into our restaurant customers, so new revenue in the front door.

Matt Wampler: Shout out to Laura, one of my good friends over at North. How do you deal with acquisitions? Because it’s something that not many people are really good at. I see more failed acquisitions where they try and bring the software in and it doesn’t work and they can’t integrate. You guys have done it more than anybody else. How do you make it work?

John Davie: Yeah, I think it starts with, we like acquiring founder-led, great management teams that are already running a strong, well-run business. And then we just keep those people in place and we empower them to think bigger. We empower them with more resources, help them build bigger sales teams, we help them lean on the platform for all the things that are not core to their business. Most of them don’t have large sales teams, we help them build bigger teams. If they don’t have big HR, we have more HR. They don’t have a sophisticated finance team, they can lean on our finance team. But then a lot of them don’t have all the tools, technology, and resources. So when we plug in all of our technology tools, and they don’t have to now build or buy all these other tools to try to integrate them, it just empowers them to grow and focus on bringing more savings and value to the customer. So that’s kind of some of our strategy.

Matt Wampler: I imagine after like three dozen times you’ve gotten pretty good at figuring out what makes a good company or what’s real technology. How do you figure that out?

John Davie: Well, one, we usually have such a good pulse in the market. We already know they’re a good company from competing with them or talking to their customers, engaging with their customers. We get to hear from the supplier community so we know if they’re well respected with the vendors out there. We look really closely at their data. We even approach M&A very much from a data perspective. So even if the revenue or EBITDA is not totally where it needed to be, but if we can see value in the size of the data or where the data’s at, even if it’s dirty data, we say, okay, we can take your data, we know we can clean it and make it more magical and squeeze more value for the operators out of the data you have. So those are some of the things we look at that is kind of unique, compared to how some people just look at an acquisition and say, can I cut a bunch of costs and drop some money by reducing headcount. We don’t really look at it that way.

Matt Wampler: Yeah, and the fact that you guys have such a sprawling organization now, I’m sure there are multiple synergies to at least a few companies.

John Davie: Yeah, we really think mostly about the revenue synergies, right? Is there cross selling? Can their customers participate in some of the other products or services we have? Can our other software tools or GPO brands bring value to their customers and drive more revenue overall and more value for the operators and the suppliers that we work with?

Matt Wampler: You brought up having a finger on the pulse of the industry. You guys are so well positioned for that. I’d love to just hear your perspective on the industry today. What’s going on in the industry? What’s working? What’s not? How do you feel about today’s restaurant industry?

John Davie: I’m an optimist, so I always kind of have the most optimistic bent on everything, typically. So with that being said, I think the industry is an amazing industry. I think it’s gonna continue to thrive and grow for a long time to come. I think it’s one of the safest industries. It’s not going anywhere. It’s not gonna just get disrupted by any one technology or some shift in the market. People have to eat, they’re gonna eat, and they’re not gonna eat out less over time, I don’t believe. I don’t think they’re gonna learn to cook in their kitchens a whole lot more. So long term, the industry is in great shape and very much worth investing in, especially if you put a long term lens. It’s always gonna have its challenges from a macro perspective of the current consumer sentiment and how people are feeling and how much of their dollars are they willing to spend outside of home. But those are typically gonna be, I think, relatively short lived and it’ll affect parts of the market. But food service in general, I mean we’re in not just restaurants, we’re in food service in general. When one segment, if fine dining is getting hurt, the other segments will oftentimes be okay. Or if fast casual is being hurt, there’s other segments of the market that are gonna thrive. So yeah, there’s definitely some negative trends this year, but overall I’m not too concerned.

Matt Wampler: It always feels like now is different and then in ten years you realize it was kind of the same. But it’s always been hard to make money in restaurants. It feels like it’s harder today. And I don’t know if that’s just third party fees or the amount of work and technology and everything it takes to run a modern day business today is more. But does it feel harder to you for restaurants to make money today? What are you hearing?

John Davie: Yeah, I think so. Definitely when I started the company back in the late nineties, operators would just focus on a couple of things, right? They would focus on their costs and their labor.

Matt Wampler: Live within your four walls. It’s like I gotta manage the customers in the kitchen or the customers in the lobby, my food cost and labor, and that’s the business. Today it’s like you also need to be a master at TikTok and make sure your CDP is set up correctly.

John Davie: Yeah. I mean, just the amount of data points and analytics you gotta track, and everything. Like I said, back when we started, most operators were just good old fashioned pen and paper on a clipboard. There wasn’t a lot of BI and data analytics and data science when you’re operating on a clipboard. So now you gotta be paying attention to a lot more statistics and trying to figure out what’s the highest priority.

Matt Wampler: It feels like things are starting to get bundled more. Ten years ago, there weren’t that many software providers, so you were just happy to find one. And then all of a sudden you had all of these new companies come up. It was like a renaissance of restaurant technology and there were these point solutions that were best in class, and now it’s overwhelming. People don’t have the time and they’re trying to kind of consolidate and simplify. I would think you guys with your unique relationship with restaurants would almost be one of those, across all your portfolio companies, like here’s your prepackaged software so that I’m not thinking about my whole technology stack. I go back to the food costs, labor and keeping the customers happy.

John Davie: Yeah. I mean, it’d be nice if it was that simple, but like you said, people are tied into one thing over here and this over here, but we can bring them this and that. There’s sometimes where we can certainly bundle a whole package and make it very simple, and all of our tools obviously talk to each other and are integrated with each other. And so that’s what we’ve been bringing together, full back end, full blown supply chain software tools with now a full inventory, recipe, accounting software. We don’t have all the front of the house tools though. So they’re still gonna have to pick and decide all their partners on the front of the house. We’re not too much on the front of the house software side.

Matt Wampler: Hey, that makes me happy because, as an ex franchisee, franchisors were always buying everything you can imagine to try and increase revenue because they didn’t live on the bottom line. But all of us, the bottom line, the back of house, it got neglected for so long.

John Davie: Yeah, it’s a good point. There’s been all the evolution of point of sale systems and all the third party delivery services and third party marketing and customer engagement. There’s a lot there. And I’ve talked to them and we engage and we partner with a lot of them, but man, there is a lot to unpack there. We’re pretty much solely on back of the house related softwares.

Matt Wampler: So a good friend of mine, Derek, who’s a customer with us, he’s an evangelical supporter of Bikky. He loves the CDP and he’s like, man, we have data on everything. I know who my guests are, what they like, when they’re gonna come in, all of that. And then I go into the back of the kitchen and I’ve got Bill over here, who’s just making tuna salad or chicken salad based on what he thinks he should make. And he’s like, I got all the intelligence in the front of the house. But all the money is in the back of the house. And it’s the area that we are poorest in knowledge. It’s like, where is the operations version of the CDP? What’s my operational decision platform? How do you approach the back of the house? And what are your thoughts on the digitization of that?

John Davie: Yeah, I mean I think it does really start with getting your data digitized. The average operator buys from a lot of different vendors, has a lot of different suppliers, and that data is not really consolidated. It’s in Sysco’s eSysco and it’s over here in this spreadsheet and they don’t really have a database bringing it together. They have small meat vendors, seafood vendors, that stuff’s typically all over the place. And so it’s pretty hard to really have any intelligence or AI or real decision making around it when you have very disseminant procurement data. And so that’s really where we start helping operators, is bring it all together in one place, clean it, normalize it. The data also comes in very dirty, so even if you just get it in, it doesn’t all line up. There’s nomenclature, it doesn’t line up. So we clean, normalize data, and then from there a lot of magic can happen.

Matt Wampler: Walk me through that, because I think one of the biggest opportunities is restaurants do have a ton of data. So much data that it’s almost overwhelming to do anything with. And the one thing AI is really, really good at is actually being able to translate that data into something a twenty two year old general manager could just understand.

John Davie: Yep. So that’s where we get really excited about AI. We’ve been spending decades, number one, collecting the procurement data in a digital format. Number two, normalizing it, cleaning it, making that data clean and actionable. And we’ve been deploying AI for over fifteen years against that data. Way before ChatGPT or any of it was a massive buzzword, we were building mini neural nets and doing product matching and data cleaning on that data. And so then, obviously with the advent of ChatGPT about three and a half, four years ago, and all the layers of evolutions of AI now, we’re sitting on a hundred and ten billion of cleaned, normalized food service transactions. I think it’s the largest, most diverse data set in the whole industry. Even bigger than like a Sysco’s data set. And it’s very diverse, across independent restaurants up to large chain restaurants to hotels, casinos, and many thousands of different vendors. And so now we’re in this phase with AI of taking all of that and not just rendering it in sort of dashboards that have a whole bunch of analytics for an operator to try to unpack, but just bringing them proactive decision making, right to their fingertips, actions they need to take to move the needle on their revenue or profit.

Matt Wampler: Right, so I got to catch the first couple of years in restaurant technology pre-ChatGPT. And so I’ve gotten to follow the arc for the last five years, and it’s been crazy. But I gotta imagine from your perspective, if you were playing around with neural nets 15 years ago, running probably on local servers, not even cloud-based, what’s the evolution been like?

John Davie: Yeah, it’s been wild and fun. I’m not a developer, I didn’t train on technology, but I’ve dived into it deeper and it’s just a passion of mine. I enjoy geeking out with my developers around how to apply AI, and I can explain it to, say, manufacturers or restaurants in an easy way that they get excited about even though they’re not really deep into it. And so especially in the last few years with ChatGPT, we’ve really quadrupled down on our investment in AI, even though we’ve been investing in it for 15 years. And we’re launching a big product that we just launched at our big trade show, our big summit, called CollectivIQ, where we’re leveraging all the AIs that you’re familiar with, ChatGPT, Gemini, Grok, Claude, and a bunch of others, to make operators’ lives really simple and not have them have to pick which AI frontier lab to go with.

Matt Wampler: Right, so I want to get into the CollectivIQ thing, but before we do, there’s something fascinating there with programming. You say I’m not a programmer or a tech guy, right? We all had to rely on what, like three percent of the population that spoke a very specific language. They could write Python, right? And they were always the translator. And maybe there’d be multiple people involved before that actually gets written. And now all of a sudden it’s like we had all of the software created from a very specific personality type, because programmers do have a specific personality in general, not to overgeneralize. And now we’re gonna open it up, we’re gonna let the other ninety-seven percent that didn’t speak that language create what they want. I feel like that’s gotta change completely the type of software, how things are built. It’s a totally different experience. It sounds like you’re playing around with it quite a bit.

John Davie: Yeah. The challenge always was, if you had an idea, a feature or something, as a business person or salesperson or a non-tech person, you want this cool feature in this piece of software, it’s very complicated to articulate that to these developers and translate that language so that they would actually be able to code what you wanted. Now I can go in, I can vibe code exactly what’s in my brain, iterate on it by myself, using CollectivIQ or other tools that are powered by AI, that now spin up not just a wireframe, not just a better mock-up, but a full working prototype of what I want before I even give it to the very expensive developers that are hard to talk to.

Matt Wampler: Well, and there’s this whole process of when you’re working through it, you realize, oh I was wrong, I didn’t want it that way. And that would have been like six months you threw away. And instead you figured it out in the first four hours.

John Davie: Yeah, we figured it out in four hours, I iterated, I can show it to a friend who has a restaurant, say, what do you think of this? Play with this for a second, text it to him, and you already have customer feedback before it even goes to an expensive developer. It’s really incredible what you can do now.

Matt Wampler: So ChatGPT was great, right? When it first came out, you could write things and there was some marketing stuff. But I remember when I first got Cursor, my co-founder got me set up, and I sat down at 7 a.m. at my computer, in my underwear having coffee, and literally didn’t move till like two o’clock. Something went off and I’m like, I’m supposed to be in a meeting. It was like this changed everything in my world. Do you remember the first time that you were given that power?

John Davie: Yeah. I didn’t get it in Cursor. This was maybe about a year and a half ago. I had my developer basically just show me how to plug in one of the LLMs and set me up with a little mini IDE, and I built my first game with the family. I built like a Duck Hunter game. It was clunky and it was hard. I’d copy paste and I’d learn some basic software jargon and code, but ultimately, after like you say, three, four hours, I had a Duck Hunter game that I sat and played with the kids. And yes, it was eye opening.

Matt Wampler: Same deal. And I have mixed feelings about this. I sat down with my daughter. We did one of those maze runner games where things are popping up. And they could say, no, I wanted to have palm trees and it’d be snowing and there’d be elves around. And it was this really cool process. And now I’m conflicted where it’s like, my daughter wants a laptop and she’s like nine. Do I really want to give her AI to go build games on? It seems like it’d be a good thing, but also it’s like we’re just gonna give the nuclear codes to the nine year old.

John Davie: No, I’d let her run with it. I’ll sit with my seven year old and I’ll say, we’re just bored, what do you want to build? What kind of game? She’s like, I don’t know. I’m like, well, what’s your favorite animal? An axolotl. I’m like, okay, how about a game with an axolotl? What’s your second favorite animal? I think she said a zebra. Okay, so let’s make an axolotl and a zebra game together. And who’s the bad guy? Who do you think the bad guy should be? And she’d be like, the dragon. Okay. Then we just, all right, type it in, tell me what you say. And lo and behold, three minutes later, we have an axolotl zebra tag team game against the dragons. Pretty amazing.

Matt Wampler: Well, and it flips everything on its head, right? Because it used to be the barrier was the technical capability and how hard it’s gonna be. Today it’s like the idea. When you have the power to do anything, you sit there and go, well, what do I actually want? It’s the creative problem solving that’s hard.

John Davie: Yeah. It’s creativity and then execution. The AI doesn’t solve total execution. It can shorten some execution steps, but you’re still gonna need probably a human for a long time to actually go out and sell it and convince people to buy it and all that. One day we’ll turn that over to the robots, but right now, for the foreseeable future, we’re gonna have to still execute as humans.

Matt Wampler: Yeah, everyone that talks about AI being the end of white collar jobs, I always sit there and think, now that we have the capability to do anything in the business, and there’s no excuse as to whether we can get it done or not, you now have so much more work to do. I feel like I’m working more than I’ve ever worked because of AI, but on the flip side, now it’s more work. It’s kind of crazy.

John Davie: Yeah, I don’t buy into that. I think the white-collar jobs are gonna be fine. I think when a white-collar job becomes twenty-five, thirty percent more productive, then the whole economics change around that. And businesses naturally, no matter how big or profitable they are, they want to get bigger and they wanna become more profitable. So you’re just gonna hire more white-collar jobs as long as they’re that much more profitable to the company. And so I ultimately think we’re gonna see job growth out of it. At Buyers Edge we’re not slashing our workforce even though we’ve been investing in AI for fifteen years, and the workforce just continues to grow throughout that whole time. Will things move around? Will you hire less of this role or will you cut back on this role and hire more of this role? Yes. So I can’t promise every single title and every single company will grow over time, but the overall workforce I don’t see going down over time.

Matt Wampler: You know, I get kind of a kick out of the whole how you bring AI into an organization, because I feel like there’s not a correlation between how many tokens you burn and how much more productive your system is. I run into people that are running twenty unit brands where they had the right tech guy with the right personality who understood AI, that stood up some incredible stuff that the Coca-Colas of the world wouldn’t even dare try, because it would be kind of experimental. It’s almost leveled the playing field in some ways where there’s a guy with four units that’s got a fully automated marketing engine that’s just killing it. And then there’s giant organizations that have no idea and are just burning tokens.

John Davie: Yep. You’re seeing a lot of different levels. I think it’s a problem that every business, not just restaurant or food service businesses, are grappling with. How you’re gonna leverage AI, where you’re gonna leverage it, how much you’re gonna spend on it, what are you giving your employees as far as AI tools. We have investors asking us, and I can only imagine any other business with investors is being asked the same thing. What do you do with AI? So it’s a big challenge. It’s not just something for the CTO or CIO or the IT guy to go figure out. I think this is CEO on down has to figure out how, what, where, when, and how much to spend on AI going forward.

Matt Wampler: How do you deal with the amount of data you have? Because it would be easy to have an AI tell me how much I ordered last week or how much did you save me on a specific question, location, but you’ve got so many companies and tentacles everywhere. And if you could have AI that really truly understood the entire picture, I would think that would be incredibly powerful, but that’s really hard to do.

John Davie: Yeah, it’s hard to do, and obviously we’re very sensitive about our customers’ data. We keep our customers’ data separate and private. We can aggregate certain things and get certain industry trends and see what’s happening across the industry with the data we have. But where we’re really focusing, we’ve had, and the industry’s had for a while, really good BI tools and industry tools that a human has to log in, look at it, check it out, and decide what action they should take based on the data they’re being shown, say in a dashboard setting. The next evolution we’ve been really focused on for the last year is, how do we just have our tools and the AI tools, with the clean data we have, just tell them what it really thinks, with a high probability, they should do to move the needle the fastest. And then tell them exactly the steps to get them there.

Matt Wampler: Answers. They need answers.

John Davie: Right. Not just dashboards and BI, but the answers, and then potentially even agentically offer to go take some of that off their plate and go do it for them.

Matt Wampler: Well, walk me through that, because I think that is partially the CollectivIQ that you’re working on, or at least that’s the beginning of that process.

John Davie: Yeah. So CollectivIQ was born out of the fact that we were trying all the different AIs out there and we had a number of different licenses. Some employees were using Claude licenses, some were using GPT, some were using Gemini. People had their preferences. We did realize that we needed to own the enterprise license, or else if our employees were putting our company data into a free license or a non enterprise secure license, it could train it, and then all of a sudden our competitors could learn our most sensitive stuff.

Matt Wampler: And not to cut you off, but that is the one big one. Everyone is always like, we haven’t figured out our AI policies. Well, you better figure it out now because your team is probably using it.

John Davie: Yeah, your employees are doing something. A lot of your company data is ending up in an AI, whether you know it or not. It’s called shadow AI use, right? And so we had to lock that down when we learned that about a year and a half ago, and say, okay, I know I’ve been telling everybody to try whatever you want, but I gotta put my foot in the mouth on that one. You gotta hold off. We need an enterprise. So we went with ChatGPT at forty bucks a month per employee, fourteen hundred employees, seven hundred thousand dollar bill.

Matt Wampler: Ooh.

John Davie: Yeah. So that was hard, but I didn’t want to only give it to some employees and not give it to every employee, because it feels like it’d be like hiring a new employee and being like, you’re not worthy of a laptop. So we’re not giving you a laptop. You’re not paid enough, you’re not high enough titled to get a laptop. No, of course we give a laptop to every employee who joins our company. So I felt the same way about AI.

Matt Wampler: Also, if one in ten or one in fifty of them happen to come up with a great idea, that’s pivotal.

John Davie: Right. So I felt like I had to give it to everybody. I didn’t want only executives or only people of a certain title to get AI. No, everybody gets AI, but it costs a lot. Then it was like, okay, we knew everybody was using it a lot. We were pumping a lot in. And I called the sales rep for ChatGPT and I asked for him to give me some analytics. How were our employees using it? What were we getting out of it? What are we putting into it? And his answer was pretty blunt. It was basically like, can’t give you anything. They’re all separate instances.

Matt Wampler: By the way, that makes me feel better that that was the answer.

John Davie: Yeah. And I said, look, I’m not asking you for people’s individual prompts. I’m not asking you for secrets here. I just want to know in general, give me some analytics, right? He wouldn’t do it. So it really hit me after that conversation, it’s like, we have to own this. We can’t just rent this from one vendor and get vendor locked into just one ChatGPT. We have to be able to be agnostic. We feel very vendor locked with Salesforce.com and that bothers me, that I don’t have a lot of bargaining power with Salesforce.com.

Matt Wampler: Because that’s what you do, is bargaining power.

John Davie: Yeah, I know. I would like to do an RFP every so often and bid them out, but Salesforce will call you out and be like, yeah, good luck trying to switch off our service. I don’t like that. And I’m very much not trying to get into that with anybody else. And I was feeling that with ChatGPT and this enterprise license. So I went to my CTO and I’m like, how do we build our own LLM? And he goes, John, we can’t build our own LLM. That’s gonna cost billions of dollars. And I’m like, okay, all right. And then a month later I come back, well, what if we did our own open source and we stood that up, and then at least we’d own our open source? He said, well, we could do that. It’s probably gonna cost a million dollars a year. You’re gonna have to hire three, four people, we have to stand it up, and then you’re gonna be like four generations of intelligence behind the latest frontier model. And I was like, shoot. Well, that doesn’t sound great. So then I kind of woke up one morning, in the middle of the night, and it dawned on me. I was like, why don’t we just stream together all of the main AIs and then create a consensus engine that gives us the best out of each one of them. And so that’s basically what CollectivIQ does. You prompt it, it goes out smartly to the right LLMs at the right price, depending on what you ask, and it gets all their answers and it pieces together the best of the best answer, the most accurate, the most honest, and it basically fact checks the underlying LLMs. Does that make sense?

Matt Wampler: Okay, so many questions. All right, first question. That’s probably a great way to get the right answer, the mixture of experts, you get everybody in there. Also a great way to just burn a ton of tokens. So the question is, do you even care, because this is the beginning of it and you can work on efficiency later? Or is there an elegant way to do it cost effective?

John Davie: Great question, and that’s kind of the number one question we get. So number one, when I built this, I did not care about cost. I was already spending seven hundred thousand dollars on a ChatGPT enterprise license. So cost was not the main concern. It was all about not getting vendor locked to one LLM. Seeing all these announcements, back then it was Gemini three that blew up the market. And I’m seeing, I was like, wait, I don’t want to keep employees stuck on the oldest, the not so smart one. I wanna be able to jump to that, but I can’t keep jumping around on enterprise licenses and paying for multiple enterprise licenses. So it was less about cost and more about maximum intelligence that was going to give them the right answer. Because there were a few scenarios, even I was using it for a major acquisition, and I was twenty five prompts deep on a major acquisition and data and everything, and I challenged it on something. Because I talked to a human that told me, no, no, no, John, that’s not right. And so I went back in my AI and I challenged it. And it goes, yeah, I have to be honest, I’ve been fabricating a number of facts throughout this thread. And I was like, what? So I was like, what do you mean? Give me details of what have you been fabricating. I’ve been fabricating certain specifications. I mean, it went through a laundry list of things that it fabricated throughout the thing. That was a huge wake up call. And if I hadn’t talked to the human who challenged one of my data points and then gone back and challenged it, it wouldn’t have admitted it fabricated. It was basically then paying it to lie to me, essentially.

Matt Wampler: And do it well. That’s the thing that kills you. It’s very convincing.

John Davie: I went back, it was so convincing. The way, I am certain of this, words like I’m definitely this. I mean, it was scary how convincing it was. Had somebody not fact checked it, I would not have caught it and it could have cost us many millions of dollars. So it was less about cost, because the cost of my employees going off half-cocked with the wrong data, or me presenting to my investors a wrong data point, was way more risky from a cost perspective than token cost.

Matt Wampler: Yeah, the cost of making a bad decision is much, much higher. Well, but John, you have the coolest data set ever, which is how all the LLMs perform on the same prompt. So tell me who’s good, who’s bad, who’s in, who’s out. Do you want the high-end models? Do they really make that big of a difference? Give it to me.

John Davie: Yeah. So I hate to pick on any one model, because we use them all and you get access to them all through CollectivIQ. Some are more sycophant-advanced than others. I don’t know if you know what that means. It’s like more agreeable.

Matt Wampler: Just wanna make you happy.

John Davie: Very make you happy, agreeable. It’s a brilliant idea. Should we go sell ice to Eskimos? Yes, that’s a great business model.

Matt Wampler: And we should say that whatever we say right now is gonna be completely out of date in like six weeks when the next model launch comes out. But yeah, walk me through it. What do you see?

John Davie: Okay, so some are more sycophants. They’re basically digital yes men. Some are more just inaccurate. It’s not necessarily malicious, they just get it wrong. An example, my son was asking me for help with his geometry homework. I didn’t know the answer, so I snapped a picture of it. I sent it up to one of the models and it came back with the answer, like sixty-three. And I’m like, John, the answer is sixty-three. I thought I was being all cute and sneaky. And he’s like, Dad, that’s so wrong. It’s not right. I’m like, prove it. Come on, this is the smartest intelligence that humans have ever created.

Matt Wampler: It just wrote me a research paper on why it’s right.

John Davie: Right, exactly. It was very convincing. And sure enough, it was dead wrong. And I don’t think that was malicious. I think the AI just hallucinated and it just straight up didn’t do a basic math problem correctly. But then there’s that example I said, where it was literally fabricating information. And that was the one that really got me. I cannot rely on any one model. And they all do it. As I watch it, CollectivIQ shows you, it flags where they got it wrong. It has argue mode. If two models completely disagree on a number, it will go through another round of arguing between those two. And then usually there’ll be a verdict and one of them will convince the other one that theirs was right. That’s a really fun stat to watch. It shows me what percentage it uses of each one. So oftentimes it’ll just say zero, because one of them just wholly hallucinated and didn’t get it right, so it doesn’t use that at all. And there really isn’t one winner take all. I wish I could say this one’s amazing and this one’s not. They all are good at different prompts. They’re good at different things. Some are good at math, some are good at creative writing, some are good at medical questions, some are better at big data problems. And that’s what I think the whole industry is gonna learn, especially as you see the Chinese models and everything coming out, and open source, and there’s a whole conversation going on about this. And this is good news. I don’t think there’s gonna be one winner take all. It’s not gonna be a monopoly or a duopoly of just Claude and ChatGPT. I think there’s gonna be a lot of models out there, and each one is gonna be better at certain things and bad at other things. The only problem, to your point, is it’s gonna be harder and harder for people to figure out which one to do, who to sign up a contract with, who to pay. And that’s where I’m excited. I have this CollectivIQ, and now it’s gonna just decide and route smartly to the one that is the best intelligence at the best value that is maximally honest.

Matt Wampler: Well, and again, the ground shifts beneath our feet. I remember even when it first started, ChatGPT three came out. We built an agentic system on it. And by the time it was ready, ChatGPT three point five was out. And then it was four. They update these things so frequently that you may be good at math one week and then the next week it’s really good at reading. So the ability to figure out who’s the best and route it makes sense.

John Davie: And the average restaurant operator who doesn’t understand APIs or how to plug in APIs, they don’t know how to bounce it when the next new model comes out from a new company. They don’t really want to go sign up and get a new API and plug it into their thing. And that’s why we’re trying to keep it really simple with CollectivIQ. You don’t have to be like me, up on all the latest, greatest models. We get them plugged in within hours of them launching. And if you wanna use the max tier, which we call the frontier tier, you get Fable Five, you get GPT 5.6. It costs you more. It’s definitely gonna be more expensive. But if you don’t need that, you can just go to auto mode, and you ask it like, what’s the latest news? It’s just gonna go to the cheaper, less expensive models, and it’s gonna do a perfectly good answer on what the latest news is. And that’s, back to your original question, that’s how you save money, even though we’re hitting multiple LLMs. Because it smartly routes up and down from frontier, the most expensive, call it Fable Five, down to GPT two point five flash, which still works totally good. It still uses those models for easy prompts.

Matt Wampler: And in some ways, that’s kind of what the models are now doing themselves, even within their model. You ask Fable Five and it spins up a bunch of agents that are using Sonnet to go do a bunch of work to be more cost effective. So where’s all this going? It’s great that you’re putting the power in the hands of the owners and the managers and the restaurateurs. It’s gonna open doors, they’re gonna do things with it. Is this one of those where you’re just like, hey, people don’t understand AI and we’re going to help bring together all the AIs for them so they can have one place that they can trust with it? Or is there a bigger vision to all of this?

John Davie: Yeah, so the bigger step one was bringing all the best of intelligence together, creating the consensus engine to be at the best value slash price for the prompt they’re looking for, and weed out those fabrications, lies, or misinformation into an honest answer. And that’s done. Where we’re now taking it, and what we are announcing here at our summit, is what we’re calling digital direct reports. A digital direct report, so now all Buyers Edge employees have spun up their own digital direct report. It’s a persona. You go through a hiring process, you go to Alice Hunter, the headhunter, you describe the role you want to work for you, and then she asks you several follow-up questions about what skill sets they need, how proactive they should be, what tools do they need, et cetera. And then it spins up a profile and it gives you a first name and last name. You can change that to whatever first name and last name you want. Mine’s called Max Moore, he’s my main guy. And then it gives you a picture, a persona, and then you hire that person. And then you start chatting in a simple chat format with that digital direct report. And they take, again, a mixture of the LLMs for the task you give it. I have my digital direct report monitoring my email, easy to connect to Microsoft 365. So it monitors my team’s chats, my email, has access to my OneDrive, drafts responses that I can approve. To my own internal company employees, it can respond for me in Teams. It has a ton of context on me now, so it does a very good job of responding on my behalf. As a restaurant owner, you could spin up a menu engineer, in which with that menu engineer, you just point it to your company, your website, your menu, and then you can have it go compare it to all the other menus of similar restaurants in my area. Tell me where I’m overpriced, underpriced, what products I’m missing on my menu compared to others. It’s pretty incredible, the stuff you can do with it. And especially when you tap it into your Buyers Edge data, which you’re now able to do.

Matt Wampler: Okay, so again, I have so many questions. First the hiring. That’s a very specific hiring process. I have not heard of anybody doing that. How did that come to fruition?

John Davie: I played around with OpenClaw, which you’re probably familiar with, right? Quite a bit. Pretty cool. Was blown away with it like a lot of people. It’s cool. I gave it a name and it’s doing some different stuff. But as I play with it, it still didn’t feel like an actual employee that I would really be willing to have talk to my human employees. So I sat back and said, okay, how do we make CollectivIQ feel like it’s a real digital persona that can actually do real work for me? What can I really delegate to this? And the average restaurant owner doesn’t understand OpenClaw, nor should they set it up. It’s very complicated, it’s a lot to do with setting it up.

Matt Wampler: Also a great way to burn a lot of money.

John Davie: Yeah, burn a ton of tokens, you need a separate device, there’s a whole bunch of security implications. We wanted this to be like the easy button of agentic AI. Which it is. It’s all in the cloud. You don’t need a device. You go through hiring just like I would do if I had a company and I want to hire somebody, you gotta go talk to the recruiting department. You talk to the recruiter, you describe what you want, they prep a persona, and you can plug in your tools. You can plug it into your Inside Track, and it can, now you can challenge it to find you products that you’re overpaying on. Which products should I be the most concerned on that are rising in price the most in the last six months, and then what should I do about it? And they go help you figure that out.

Matt Wampler: So this is the art, not the science of AI. There’s an art to putting all of this together. And I’m sure there were some variables that really mattered. Setting up an agent with personality that’s the right agent for you, what were the things that mattered most?

John Davie: The big one, I found my OpenClaw, it wasn’t proactive. It would sit around and just kind of wait for me to come back unless I set up a cron job or something. I couldn’t get my OpenClaw to truly proactively come back to me at a random time and be like, John, you really need to do this, or this is a really big problem. So we are really focused on trying to get these things to be proactive. The way I describe it to developers is, if I hired an intern or I hired a junior employee, and I have aspirations for them to be a great employee, if I don’t give them perfect directions, which as a manager, I’m not great at giving you a laundry list of things to go do. I just kind of give you general guidance and I hope you go do it and I hope you come back to me with some really good results. And that’s what we’re trying to get these DDRs to do. Even if you don’t talk to it for a little while, you don’t give it specific directions like I want you to do this, this, and this, it’s going to look through your data and come back and be like, hey, there’s a big problem. This product is rising far greater, two X faster than anything else. And would you like me to send an email to your distributor rep and question them on why your pricing is going up so much? And you can just go approve, and it will draft the email and it’ll send the email to your sales rep and challenge why your pricing is up so much.

Matt Wampler: All right, so the first thing I’d want to do is I’d end up with like 15 agents just because of natural curiosity. And it comes back to the point of, agents are actually really bad at managing things, right? You’d think, I’ve got this super intelligence, it can do anything, it can certainly manage itself. And it doesn’t. There’s a skill to managing agents. How do you manage agents?

John Davie: Yeah, it’s back to, you gotta really tell them what’s important to you, and you’ve gotta give them the autonomy to do it, even when you’re not giving them specific direction. We’re very much hard coding a lot of this logic behind the scenes, because the average user hasn’t spent time on an OpenClaw and really got it to actually do anything useful. We’re trying to take a lot of that learning and make these really proactive for a non-technical person, somebody not like you and I who really geeks out on this stuff and plays with all the latest agentic tools.

Matt Wampler: And I think that it’s an important point to hammer home. Just because AI can build things doesn’t mean it’s gonna do a very good job. And setting up the architecture, things like permissions and what’s read only and what it can write to, all the guardrails really, really matter. It would be like going into a restaurant and being like, yeah, you don’t need to follow the recipes. In fact, make whatever you want. It probably isn’t gonna go very well. So you’ve now got this framework in place where you can empower somebody to say, yeah, I wanna hire this person. I want them to do these tasks. And they probably do a pretty decent job.

John Davie: No, they do. It’s pretty incredible. And unlike OpenClaw, those are not built for really enterprise. This is built for an enterprise application. So we’ve hard coded all the stuff, like it’s not allowed to communicate with an external party without going through an approval process. A human has to click approve. Internally you can tell it, I’m okay with you communicating with a fellow employee. It’s allowed to do that without approval if you tell it as a user to do that. It can’t access certain things. There’s a whole bunch of security settings we’ve set across that to make it enterprise ready.

Matt Wampler: How do you deal with the whole AI slop where it’s just giving you lots of words that are not overly helpful?

John Davie: It’s all in the architecture in the back end. And you see, you know a lot, because you can even ask that question, because you’ve obviously spent a lot of time with it. We’ve spent a lot of time with it. On our back end, it’s like everything has to be concise. Unless the user asks you to write an essay, you have to keep it under X number of words and stuff like that. So there’s just guardrails, and look, we’re figuring it out. I’m not gonna claim it’s perfect. It’s been in the hands of my fourteen hundred Buyers Edge employees for about three weeks, and we’ve debugged it and we’ve made modifications and we think it’s ready for other customers. So we’re bringing on beta customers onto it right now.

Matt Wampler: I mean, it’s an exciting time, right? We’re all working in public, building in public, the technology is changing so fast. Have you sat back and thought, if this goes well, we’re going to give hundreds of thousands of restaurants the ability to have digital agents overnight. And they’re going to build things with them. And those things, I mean, they have close to super intelligence. They could do almost anything. The industry is going to change as a result of AI. Have you thought about what the secondary effects are and what the world’s gonna look like the next twelve, twenty-four months?

John Davie: I think what I hope happens, and it’s relatively simple, is I just hope restaurants become more profitable. Because to your earlier point, it’s hard to run a restaurant. It’s super freaking complicated, it’s not easy to make money. When you hit it big and you have a great concept that has lines out the door, you make really nice margin. You have a nice profitable restaurant. A lot of restaurants aren’t like that. So if you can take the average restaurant and you can increase their EBITDA margin and put more money to their bottom line, and if you can delegate some of the mundane, annoying tasks, like maybe monitoring your TikTok to your point, or monitoring your Facebook, or monitoring your Yelp comments, and you can delegate that to something that doesn’t sleep, works twenty-four seven, it can ultimately end up with a more profitable restaurant operation. That’s where I hope it is six, twelve months in the future.

Matt Wampler: You know, I think the biggest realization I had when I was in restaurants was how diverse the skill set you needed to have. You’re managing your permits and your health inspections and your health policies, and don’t forget to file your business license, and you kinda gotta be good at HR for hiring and firing. And you gotta know Excel. But you don’t have to be great at it. You just need to be proficient at it. And AI is proficient at all those.

John Davie: It knows everything, right? It knows every health regulation in every specific state. It helps us write contracts in Denmark. We used to have to hire a lawyer in these other countries as we expanded. Now we just have the AI write us the first draft of the legal contract and all the local regulations and laws. So as a restaurant operator, you used to maybe have to hire several consultants to deal with things that they weren’t good at. My hope is they can delegate a lot of that to AI and it’s gonna be way cheaper and way faster. And then it helps them juggle all this stuff. When you just started this restaurant to make great food and serve customers, and then it kind of dawned on you that you have to know so much more than how to cook a good dish. AI, hopefully you can delegate a lot of that stuff to.

Matt Wampler: Yeah, I think just like on the programming side, it used to be you had to go through a programmer to write code. To be a general manager, you really did need to have all the admin and quantitative. You had to be good at inventory and the P&L. There’s a lot that goes into it. But I wonder if there’ll be a shift there where all of a sudden you can start self-selecting for more hospitality people. Good with the team, good with the customers.

John Davie: Right. If you wanna do weekly inventory, oftentimes you gotta hire or pay somebody more time or hours to really have a regimented inventory process. I think AI is something that can take that off the plate, and that’s money that’s gonna fall to the bottom line. You’re gonna have more profitable restaurant owners, or open more restaurants, open a second, third, fourth, fifth, hundredth restaurant, if we can help them become more profitable.

Matt Wampler: How do you feel about restaurateurs using AI to build their own technology? Because I feel like that would be the natural evolution.

John Davie: Yeah, like in CollectivIQ, you can spin up your engineer. And I’d recommend anybody spinning that up, because it basically does all that vibe coding you’re talking about. And then you just natural language talk to your engineer, and you could say, I haven’t spent money on a shift scheduling software yet. Go research all the shift scheduling softwares on the market and build me a shift scheduling tool for my 30 employees. It can build that. And you now might avoid paying a SaaS fee for another tool. And now you have a whole shift scheduling app that you vibe coded and built yourself. And again, it saves money, you’re more profitable. And now you have a tool. Maybe you were holding out, we’ve been doing shift scheduling the old fashioned way on pen and paper. Now you can shift to an actual digital one. You can vibe code that.

Matt Wampler: Yeah, I think one of the things Salesforce did is they implemented the whole headless architecture, this idea of accessing the data through Claude or any other way. They just kind of are almost this behind the scenes invisible data standardization system of record. I kind of wonder what that economy looks like, because I do think people can vibe code a dashboard for themselves or the UI. It’s great at that and you can see it. But at the same time, you don’t want it messing with the numbers behind the scenes. You’d hate the, I can imagine a world where you tell your AI, I need my food cost to go down by half a percent every week. And the AI starts lying to you because it wants to show that your food, it’s like, I didn’t realize that they weren’t really going down. There’s just what the AI was telling me. So I feel like there’s gotta be something there in terms of governance or the master systems of record.

John Davie: Yeah. AI is super powerful, but it’s also dangerous. Like you said, if you said, I want to lower my food cost by half a percent a week, well by that theory, in a couple of years your food costs will be down to like five percent. And so that’s obviously impossible. So you have to be careful with what you tell it to do as we hand over more of the control to it.

Matt Wampler: Or have five of them that are all agreeing that it’s the right answer. That also helps. It was funny, it was Friday night, it was like ten o’clock, long week. I had a glass of bourbon on the back patio, and I’m sitting there thinking about something I built, and like, shoot. I don’t know if the numbers it’s giving me are correct. For all I know, it made this stuff up. It was the first time. Because when you look at something on a dashboard, it seems like fact, but you know.

John Davie: Yes. No, you should go look at the numbers, because I have mine doing a lot of presentation building. I almost never use PowerPoint ever again. CollectivIQ engineer will build me an amazing animated deck. I give it a couple prompts, it does all of its own research and it fills in all its numbers. It looks amazing, but I have to tell it, make it editable. Because when it puts important numbers in, it fills in a number, but sometimes it literally just makes those numbers up.

Matt Wampler: Yeah. Absolutely insane. I was having this conversation with Bill Lindsey, Cogwheel, back of house software. And one of the things, I was sitting there thinking, well, maybe Google could just have their AI build restaurants’ back of house software and give it away for free so they can get more data. That’s something you could almost do today. And I was kind of thinking of like, how far are you gonna let AI go into your software? Part of me thought, having a system of record that AI can’t touch is really important. On the flip side, it’s really nice to be able to tell AI, just go in and change all the numbers because we changed our recipe and that’s all out of date, and have it change everything. Do you get any sense of how deep you think AI should go into systems?

John Davie: I think people will keep pushing it until there’s several tragic examples of where, like I said, it either doubled their food cost or cut their food cost in half, but it also meant the quality of the food also got dramatically cut in half. And those stories will get out and people will pull back. But that’s the advancement of technology. You push it, push it, you hit a wall, or you might have to take a step back because you might have given up too much control. But then you build more guardrails, build more safeguards, and you move ahead. I don’t get too scared. I don’t think AI is gonna put any restaurant out of business anytime soon. And if it did, it would be such a big story and it probably wouldn’t happen again. But I think ultimately, we’ve all been giving over more and more of our data and everything to apps, software, tools, and everything. And ultimately that has been a good thing. It’s created more visibility, more transparency, a little more complications like you’re talking about. But ultimately I don’t think we ever want to go back to pure clipboards and only paper invoices with the coffee stains on them. I don’t think we want to go back to that day.

Matt Wampler: Unless we have our Y2K moment with AI. John, you run a massive 1400-person organization. You’re a multi-billion dollar company, and you guys have gone into AI and are actively using it. What do you say to all the CEOs out there that are just waiting on the sidelines and tentative about this and risk averse?

John Davie: I think my advice would be, you really can’t ignore it. You’ve got to spend the time on it. Even if you say it doesn’t affect my business today, it’s hard to imagine it won’t really affect your business in six months or a year. So the longer you wait to dive in and play with it, tinker with it, toy with it, the harder it’s gonna be when a time comes where you really do need to dive into it. And secondarily, I think it really can be a massive efficiency tool. I talk a lot about speed when I talk to my employees, and I talk a lot of times on stage. Everything’s about pace. The faster we can achieve something translates to growth, right? If you want revenue growth or profit growth, you have to go faster. You have to serve more food quicker at a lower price. You can almost bring everything back to speed if you want to grow. AI is a massive accelerant to access to information, access to intelligence. And so to ignore that, I think, is really just putting yourself at a disadvantage. So step one is, they gotta tinker with it, see it, sit in on demos. People are oftentimes surprised that I sit through certain tech demos. Normally other people would make the decision on these software tools. I like to sit in on the demo. I like to see it myself and understand it. And that’s part of how I keep up the speed on these things. I was impressed, I did a demo with a 300 unit restaurant chain today. He was the founder. He sat through the demo, it was just him. He didn’t bring his IT people or anything. And I give him a lot of props. He’s a good example of a restaurant CEO fully diving in, spending the time himself, learning it. At the end of the day, he’s a restaurant owner and operator. So that’s kinda cool.

Matt Wampler: And he makes the time.

John Davie: Yeah, he made the time. And so that’s fun. That’s a similar situation I’m in. And we have a lot of people that can help navigate and help him understand technology and data and AI. Short answer is, you gotta tinker with it, play with it. Good news, you can do most of it on your phone now. So it’s not like you have to be sitting at your desk to tinker around with this stuff. You can vibe code with voice as you drive around.

Matt Wampler: Put your headphones in and talk to it.

John Davie: Yeah, I mean I’m driving around, my car self-drives. I drive a Cybertruck and I essentially vibe code while the car is driving me around. So it’s pretty fun.

Matt Wampler: Well, John, I appreciate you coming on today. If anybody’s interested in Buyers Edge, how do they find you? How do they reach out?

John Davie: Yeah, I would say if you don’t have connectivity with any of our client managers, any of our business development people out there, just go to the website, inquire through there. We operate under a lot of different companies, so start with buyersedgeplatform.com and see what we do and reach out. And we have real humans that will talk to you. It won’t be a digital direct report just yet. Give us a couple months on that one.

Matt Wampler: Just yet. Until they’re better than the regular humans. Well, John, thanks for coming on. This has been fantastic.

John Davie: All right. Awesome.