AI and Automation for Promo Suppliers | skucast Ep 376

"If I've got AI and I'm looking for a problem, I think I'm approaching it backwards."

— Mike Knapick, SanMar

Ask skucast

Want more of Mike and Phil, or any of the 350+ guests who've been on the show? Ask them directly. It's free.

Try Ask skucast →

[00:00:06] Bobby: What do the industry’s largest supplier tech leaders think about AI, automation, and data today and for the future? Well, you get to find out. Today we’re hanging out with Mike Knapick, CIO at SanMar, and Phil Gergen, CIO at Koozie Group. Welcome to the skucast, the podcast for innovators and maverick thinkers in the brand and merch space.

[00:00:27] My name is Bobby Lehew. I’m glad you’re here. Mike and Phil both sit on the board of PromoStandards, and between them, this year has been full of big, complex projects: acquisitions, integrations, and a ton of automation. We’ll talk about SanMar’s most automated distribution center, Koozie Group’s MCP connector Kai Link, and how each company is structuring its team for AI adoption and transformation.

[00:00:50] Today’s episode is brought to you courtesy of us at commonsku. Over 950 distributors powering 1.8 billion in network volume rely on commonsku’s connected workflow. Process more orders, connect your team, and dramatically grow your sales. To learn how, visit commonsku.com.

[00:01:06] Oh, one quick note before we dig in. Many of you know the term MCP, but for those who are new to it, MCP stands for Model Context Protocol. It’s an open standard that lets AI tools like ChatGPT or Claude connect directly to a company’s live data and tools, so it works from real records instead of, like, guessing.

[00:01:23] I have MCP connectors for my Google Suite (Gmail, Calendar, Docs), for HubSpot, and a few others. commonsku is also building an MCP for customers that will let you connect Claude or ChatGPT to your live commonsku data. More on that later, but for now, that’s the backdrop for part of our conversation with Phil and Mike.

[00:01:40] Here’s that chat.

[00:01:41] Hi, Phil and Mike. Welcome to skucast.

[00:01:44] Mike: Good afternoon. Thanks for having us.

[00:01:45] Bobby: Thank you also for hopping on.

[00:01:47] It’s always a busy time. You guys have probably one of the busiest and most stressful roles in the business. I say that with much admiration for what you’re doing. And speaking of the first thing I wanna talk about, talk about busy. Mike, SanMar completed the BELLA+CANVAS acquisition in June. What does bringing a brand like that on board involve on the tech side? I know it’s a very broad question, and I know Phil is very empathetic ’cause he’s done some of these kinds of things too, so...

[00:02:13] Mike: Yeah, it’s a great question. We got a great brand in Bella, and we’ve got a great set of styles and a great team out of Los Angeles. Our key technology pieces of work right now are ensuring that customers continue to have access to Bella styles and inventory as the transition happens.

[00:02:29] We’ve got inventory in the SanMar distribution network. We wanna ensure that our customers have access to that. You know, they’re used to doing their business through a website that’s run by the combined BELLA+CANVAS company. We’re standing up new technology to front those orders and send them through our distribution network, and that’s where most of our focus has been: how are we ensuring that that great product gets into our customers’ hands in a timely fashion?

[00:02:54] Bobby: Yeah. Phil, Garyline began operating under the Koozie Group name in January. You were the executive sponsor for that integration. What did you build or learn through that process?

[00:03:03] Phil: Yeah. You know, we’ve done a couple, and every one is different. You come in, you have this playbook, and you start walking through this playbook, and you’re like, “Okay, but that’s different. Okay, that’s different.” There’s its own challenges, but I think the point that Mike made there is that you want the experience for the people who are customers to be as seamless as possible. So you look at their technology, what they’re offering, how they’re doing it, and then you work to try to bring it into your system. I’m sure Mike has run into something that BELLA+CANVAS did that SanMar didn’t, and they went, “Oh, that’s a good idea.” And that’s really what comes into the technology side: the business says, “I wanna operate the exact same,” which we know isn’t a real answer, but as much the same as possible, because that’s what customers expect.

[00:03:43] There’s a reason that they bought from that company in the past. There’s a reason that they’re customers. How do you keep that and bring it on as much as possible?

[00:03:50] Bobby: Yeah. Mike, any thoughts on that before we move on to the next question?

[00:03:53] Mike: Yeah, I think that’s the key piece. If you start with, like, what’s the customer impact and how do you make that as seamless and friction-free as possible, then it leads you down, you know, working the most important stuff to the least important stuff.

[00:04:06] Bobby: I know you’re both working at a scale that we can’t imagine, but mergers and acquisitions are hot on the supplier and the distributor side. So even if it’s a small business, people think of a merger as a business event only, meaning only that type of transaction at the business level. And you both sort of hinted or said that a lot of it is about technology. For a supplier or distributor heading into their first merger or acquisition, we have had a lot of distributors on commonsku love the fact that they’re all on the same platform, because if they acquire another distributor, then it just makes it so much simpler.

[00:04:40] But for those that are trying to do something, maybe they’re in acquisition mode, because we have a lot of business owners that are considering that, what should they get right about their systems before they do this? Mike, just a couple of tips they should follow?

[00:04:51] Mike: Yeah, I think when you enter into these discussions, there’s a lot of business conversation. I think it’s worthwhile to come up with at least the beginnings of a technology plan, you know, assumptions around technology and two companies coming together, like, “Well, whose system are we gonna use for X, Y, and Z?”

[00:05:09] It’s probably not a deal breaker one way or the other, but it’s good to have just ground in, like, let’s get some basic decisions made so that once the deal is done, we have a path forward that we’ve already agreed to. I think the last thing you wanna do is get into it and then start playing that game of, “Okay, whose are we going to use for...”

[00:05:28] Whether you’re merging two companies, like, “Well, whose CRM system are we gonna use?” That’s an easy conversation at the beginning. Again, it probably doesn’t stop the deal, but at least once the deal’s done, you have a clear path forward.

[00:05:38] Bobby: Yeah. How about you, Phil?

[00:05:40] Phil: Yeah, I think that’s a great point, ’cause timelines get set prior to acquisition, right? Everyone has a timeline in their head, and if you don’t have some of those conversations about what’s the most important thing, technology or business value, what does that acquiring or acquired company do best, what’s the reason for the merger, and how do you tie that into the technology that’s helping them do that? Because if they’re really good at something and the technology doesn’t help that, that ends up putting a wrench in everybody’s plans if you don’t plan ahead. So having the discussions: what are you using today to do the thing that you’re best at, whether it’s one CRM or it’s another? It’s probably not the actual technology that is making them better or not, but the process within, and how does that merge with what you want the company to do going forward?

[00:06:26] Bobby: We’re going to get to a topic on AI, but before we do, I wanna talk about a really big topic that’s been in manufacturing for a while now, just around automation. Mike, Richmond opened last year as the most automated building in your network, with Texas and Reno coming behind it. Can you just give us an idea of scale?

[00:06:41] Mike, I mentioned in the intro about SanMar, but can you give us an idea of scale of your distribution network?

[00:06:49] Mike: Yeah, I think it’s nine distribution locations across the country. In some of those locations, it’s multiple buildings. I think the last total I saw was 12 or 14 million square feet of distribution space. And as you mentioned, Virginia, our newest distribution center, went online last year.

[00:07:08] It’s our most automated distribution location, and it is significantly different than the other locations in that we think about goods-to-person instead of person-to-goods, which is... it’s not new, but the way we do that is through this really large three-dimensional grid. You imagine totes stacked on each other in three dimensions, with people standing on the outside of that grid, and robots, really a box.

[00:07:35] It’s a super-sized toaster, maybe an air fryer. It’s running around on top of those grids, and it’s pulling these totes out of the three-dimensional grid and bringing it over to the side and dropping it down in front of a person. And when you look at it, it’s like, what happens if the tote’s all the way at the bottom of this grid?

[00:07:52] It’s really interesting to see how these robots are programmed to work together. One comes, pulls the top one, then another one pulls a second. They move everything around till you get to that tote you need, brings it over to the edge, drops it down in front of the picker, who ends up with a tote in front of them, and on a screen in front of them, it says, “Pick two of these and put it in the box.”

[00:08:10] And so it greatly reduces, right, travel time. The robots are very efficient. And over time, the product that moves quickest kind of sorts to the top of the technology that we use there. It’s called AutoStore, and it really improves the time it takes us to pick an order and get it in a box and ship it out the door.

[00:08:30] It allows us to be more efficient in the space we use in the distribution center, so we can have more styles and more inventory without necessarily having to have larger and larger buildings, which bring their own inefficiencies. And it’s a great way to serve customers quickly. And Virginia is a strategic location in terms of serving areas of the United States that gives us great coverage for one-day service to folks.

[00:08:53] Bobby: Hmm. SanMar has been deeply respected for the distribution network and speed for such a long time. So what makes this one really different than the other distribution centers?

[00:09:04] Mike: It is that goods-to-person idea. It is, yeah. Our distribution folks are fantastic. Their ability in large buildings to get orders out the door... You know, we have this promise that if you place your order, we’ll get it on that truck that day. That leads to us having a lot of volume at times that are coming up against those deadlines.

[00:09:23] This AutoStore technology allows us to be really efficient in processing those orders quickly. Sending the product directly to the picker allows them to have multiple orders sitting in front of them, and they’re just waiting for totes coming in, with screens in front of them that tell them, “Pick two of these, put them in this box, send it over here.”

[00:09:40] It really makes it easier for us to be fast, but also make sure that we’re picking the right product and putting it in the right box at the right time.

[00:09:47] Bobby: Automation is transforming both of your businesses. Phil, how is it changing the way you’re doing things at Koozie?

[00:09:53] Phil: You know, we spend a lot of time on decoration automation and a lot of investments there with large machinery that does large runs. And it’s great for really large runs, but a lot of the business gets small runs. So as we work on that technology, we work on how to efficiently batch small runs to turn them into large runs, to keep those machines running as efficiently as possible and as long of runs. And as Mike’s talking about, delivering product, loading and unloading, we do a lot of plastic water bottles and different things, capping and uncapping of those bottles. How do we do that more simply and more efficiently, in addition to people, but to get them out the door faster?

[00:10:29] ’Cause turn times are quick, but not quite as quick as Mike’s when we have to decorate, but we’re still doing a lot of same day, next day, two days in a lot of different areas. And to continue to have the efficiency that the machine was made for, you know, multimillion-dollar machines that are made for 10,000, 20,000 at a time, how do you batch your orders, get them organized in a way that it can run as efficiently as possible to deliver for everyone as quickly as possible?

[00:10:53] Bobby: One of the promises we put at the beginning of this podcast is we were gonna talk about AI, and of course, everybody wants to talk about AI. We’re in a special spot to be able to talk to you two guys, and I really mean that.

[00:11:01] You’re both the apex of technology innovation. You both sit on the board of PromoStandards, so you’ve been thinking about these kinds of things for a long time.

[00:11:11] Mike, I’m curious about how you’re thinking about AI at SanMar. And I know it’s a broad question, but you also talked a little bit about the hype cycle from the Tech Summit days.

[00:11:22] You mentioned that too. But what’s your general thought about how you’re seeing AI begin to impact the industry?

[00:11:28] Mike: Well, I think at SanMar we think of AI at a couple of different levels, right? And so we don’t have one approach that solves all problems or implements all AI, ’cause I think it can be dangerous if you think of AI as one solution that everything fits into, right?

[00:11:44] First of all, we think of it as an individual productivity tool. Way back in the day, you rolled out Microsoft Office or something to your users, started using word processing, all that stuff. You go all the way back to that time. We’re a Microsoft shop. Copilot’s on desktops for our office folks.

[00:11:59] And we made that investment and then realized, like, this is a little bit harder. Let’s get some training out to our folks. So we’ve got a great learning and development team here at SanMar that put together training so that individuals know what that tool can do for them on their individual productivity.

[00:12:15] That’s the first way we looked at it. The second way we started saying is one of the efficiencies is not just, “Hey, I can be more efficient because I use AI to help me do my job,” but that groups who have teams of people who do work that’s either workflow or similar activities: how do we enable AI to help groups of people, as opposed to just as an individual productivity tool?

[00:12:37] Two things that we did. We challenged teams to identify not AI opportunities, but really process improvement opportunities that we then look at and say, “Is AI a way to make this better?” So don’t start with AI in mind, but start with, like, where is there a place we think we can be more effective, more efficient, give us more time to do things we want to be doing and value-add and less mundane?

[00:13:00] And then we set up an IT team. We call it the AI enablement team. It’s not big, but they’re smart in the way of using AI for groups and teams. And our business partners can come to us and talk to us about, like, “I’ve got this idea about... we do this thing every week or every day, and we’ve got three or four people doing it, and we think maybe AI might help us out here.

[00:13:20] But can your AI enablement team kinda help guide us along the way?” And sometimes that’s building an agent for them, sometimes it’s helping them build something themselves, and sometimes it’s working with a third party who would come and help that group. And so those two levels are the way we’re approaching AI.

[00:13:36] I’ll say that the third one, the huge AI big enterprise projects, they just go through our normal enterprise project portfolio evaluation. We think of AI as one of the tools that we can use to solve business problems.

[00:13:52] Bobby: I was listening to the head of Google Labs, and he was talking about the very thing you mentioned, which is to make sure you have a group, it can be a small group in your organization, that is the testing ground for a lot of these AI things, and then bringing that to the organization. Is that kind of what you’re talking about?

[00:14:07] Mike: Yeah. Yeah, we are. In addition to this enablement team, we have this concept of AI ambassadors. So each group nominated AI ambassadors in their business, and these are folks who, you know, in their spare time, or they’re just really enthusiastic about exploring AI as an opportunity, and they all work together to identify, call it, proofs of concept: small opportunities that we think can grow into large ones.

[00:14:32] Bobby: I like that. I also love how you noticed right away that training was big, because we’ve seen Silicon Valley suddenly invest billions of dollars in integrators and implementation companies, because we saw how difficult it was to deploy some of these tools for all, even though the promise was there that it was gonna be easy. So that’s huge. Phil, how are you guys looking at AI, and what kind of structure have you set up with teams with AI? We’ll talk about MCP in a minute too, but structurally, how do you guys think about it there?

[00:14:57] Phil: I think similar to the way that Mike says in a lot of ways. We do have an AI personal productivity group that is specifically focused on personal productivity, because there are a lot of tasks that individuals do on a daily basis that are not necessarily scalable. And those are always the easiest and the first things to pick off, ’cause people know. But we also created a group of people to talk cross-department. There’s no two people from the same department, and they just share what they’ve done that’s unique. And the ideation that comes from that has been huge. We’ve seen our marketing team bring stuff to our finance team, say, “I did this thing over here,” and that light bulb went on. If you talk about training, training is so hard in AI because it changes every day. And it really does, and I think it’s a little cliché to say that, but this is a time and space that is real, that it’s different, it’s new, and something you did two weeks ago that didn’t work for you may work today without you even realizing what happened or what’s different. Or the opposite happens too. Something that worked fantastic two weeks ago, you do it again, and it doesn’t work.

[00:15:57] Bobby: Right.

[00:15:58] Phil: And that’s the really hard challenge for, I think, everybody who’s doing AI projects. You talk about AI being non-deterministic and not always doing the same thing, and how do you—

[00:16:06] Bobby: Can you explain that real quick, Phil? Keep going, but for our audience.

[00:16:09] Phil: Non-deterministic means it doesn’t necessarily do the same thing all the time. I had a really good explanation that someone said: “It’s like having 1,000 experts you could ask the question to, but you don’t know which expert is gonna answer.” And every expert has the exact same expertise, and they might do it the same way or they might do it slightly different. And you can give them instructions, and people talk about prompting and prompt engineering, and I don’t even think prompt engineering is a thing anymore, because the AI doesn’t listen to what you want to do.

[00:16:41] It likes to do it the same way.

[00:16:42] Bobby: Right.

[00:16:42] Phil: You have a bunch of individuals that think similarly, but not exactly the same. If you think about your problems that way, you can understand why it would answer differently time over time, as opposed to being the same thing, versus deterministic software, which is something that we’re so used to using. I type these letters, they show up in Word. I put this formula in Excel, it’s always gonna do the same thing. I can ask 10,000 experts to make me an Excel formula, and you might get answers on how to do that. I think that was a great use case at the very beginning: “Write this Excel formula for me.” But there’s lots of ways to do lots of things, and understanding what the right places are to use this technology to improve your job... But you still have to make sure that they are answering the right way, or the way that you want to, or the same way over time.

[00:17:27] Bobby: This has to be maddening to some degree. Don’t let me put words in your mouths, both of you. But deploying AI at scale feels like just a completely different animal than deploying AI as an individual. I mean, that’s probably something you’ve both dealt with for a long time.

[00:17:41] Does that concern you, or is that something you obviously have to think about a lot?

[00:17:45] Phil: I think you have to think about it quite a bit when, where it goes into the workflows. Mike talked a lot about how we have different types of groups: personal productivity, workflow enhancement, large-scale projects, and each one takes a different solution and different tools. Everyone says AI; I don’t even know what it means.

[00:18:01] Everyone says an agent. I don’t even know what an agent means these days. And I’m doing it every day. I’m creating an agent or doing something. There’s so many different skills. There’s personal agents, there’s workflow, there’s so many different things, and it’s hard to talk about sometimes, even within these groups, of what did I actually ask to be done?

[00:18:20] What was the results that I got out of it? And is that repeatable, usable, at scale?

[00:18:25] Bobby: Yeah. Our friend Eric Kanata used the term, the first time I heard it, “agent washing,” just for our ability to talk about agents as if that’s somehow magic that you were able to create there. Mike, do you wanna respond to this scale challenge?

[00:18:37] Mike: Like, I think scale’s a real challenge that you bring up, especially if you’re thinking about AI in terms of interacting with your customers, right? At SanMar, we take our customer relationships and supporting our customers really seriously, and so anything when you unleash, as Phil pointed out, non-deterministic, right?

[00:18:55] You can’t always predict what that model is gonna respond with. And so you just have to be very careful about how you use it, when you use it, how you put controls around it, how you continue to monitor it. And it’s not one of those things that you turn on and walk away from. You have to continually keep an eye on it.

[00:19:11] And so I think the challenge at scale used to be, like, just doing the scale work, and then you turn it on, you make sure it performs, and you’re good. You have to do all that and then continue to monitor it.

[00:19:22] Bobby: Yeah. I just read this morning the CEO of a $10 billion company posted his entire product online for anyone to take, ’cause there’s this trend going on around reverse engineering software platforms. It’s funny, he did this to prove a point, and the point was that you build these incredible software systems, but somebody still has to support it and troubleshoot it and do all those things.

[00:19:43] That’s just the beginning. And I think I went through my own sort of hype cycle with this, where everybody envisioned vibe coding everything, and suddenly you realize... I vibe coded a travel app, and then the first time I had a problem with it, I was like, “Okay, I’m out. That was fun, but this is now Mike and Phil’s problem.”

[00:19:58] I know that’s a little tiny tech problem, but you both laugh. But Mike, speaking of the hype cycle, where are we kinda at as an industry, do you think? And can you explain the hype cycle?

[00:20:07] Mike: I’m not sure I’m the expert on the hype cycle. I would just say generally, with new technology, early on there’s a lot of conjecture and projection about what that technology’s gonna do and be for people. And I’ll just say it this way: it’s very difficult in the middle to know, are we in the beginning, the end?

[00:20:27] What does the end really look like? I’ll go back to something Phil said. He said something about prompt engineering: “I’m not even sure that’s a thing anymore.” Like a year ago, we would’ve been like, “Prompt engineering. Everybody should hire a prompt engineer.” And nine months later, you’re like, “It doesn’t even exist anymore.”

[00:20:41] And the thing I try and be very careful about as I read and listen to people talk about AI is trying to understand who’s the source, who’s providing the information.

[00:20:51] Bobby: Yeah.

[00:20:53] Mike: A lot of it comes from companies who are selling AI, and it’s good to understand what’s their motivation for why they’re saying what they’re doing.

[00:21:00] It’s a real thing, so I’m not suggesting that it’s a non-entity. But I do think it’s worthwhile to understand... As a technologist, I’m very interested in it. As an IT person who works for SanMar, I wanna look at it from, like, what’s the business problem I want to solve, and what technologies do I wanna use to solve that?

[00:21:21] So if I’ve got AI and I’m looking for a problem, I think I’m approaching it backwards.

[00:21:27] Bobby: Right.

[00:21:28] Mike: It’s: what’s the business objective, what’s the value to my employees, what’s the value to my customer I’m trying to drive? And then AI is a tool that I can use to help get there, but it’s not the only tool I have.

[00:21:40] We’ve got years of those tools, and I think the real challenge for IT leaders is figuring out when’s the right time to use it and how to use it. And I guess that’s not different from what our jobs have been for a while.

[00:21:51] Bobby: Yeah. Go ahead, Phil.

[00:21:53] Phil: Yeah. Just to follow up on what Mike was saying, it’s definitely a tool that can solve problems that exist today. It can solve new problems and different problems, but it has to come in from the exact same flow. What is the business use case? The CEOs out there, I’ll say including mine sometimes, say, “We need more AI,” right? That’s not a statement that can be actioned on. What does that mean? How does that work? We wanna make sure we have the structure in place to use it when it’s appropriate.

[00:22:18] Bobby: Yeah.

[00:22:18] Phil: But just saying, “Use more AI,” I don’t know what that means when that gets said.

[00:22:22] Bobby: We’ve been hosting these AI roundtables where we sit down with customers, whether they’re enterprise level, distributor enterprise level, or whether they’re our ICP, and we will just share how you’re using AI in your work. And I’ve noticed the trends after having these for a few months, that distributors in particular, and I can only speak to distributors ’cause I used to live in that world for so long, will almost apply AI in two areas.

[00:22:45] One is client experience focused, anything on the sales side and customer experience, and the other is the ops, finance, and those kinds of things. So it’s interesting to see... I thought there would emerge primary trends, but there were so many challenges people were trying to solve with it.

[00:22:59] And I’ve noticed, going back to Mike’s advice, and Phil, you nodded to this too, that I think the bigger problem is understanding the problems you’re trying to solve, as opposed to understanding the tech and then trying to apply it to everything.

[00:23:11] Mike: Yeah, I think that’s fair.

[00:23:12] Phil: Yeah.

[00:23:13] Bobby: Go ahead, Phil.

[00:23:13] Phil: I was gonna say, and Mike’s probably similar to me, we’ve been hands-off keyboard for many years in actually doing anything that happens. But our core responsibility in these companies is matching up technology to problems.

[00:23:26] Bobby: Yeah.

[00:23:26] Phil: That’s not new for us.

[00:23:28] That’s not new for anybody who’s been working with technology. You have a problem, you have a different use case. Every different business, as we talked about with integrations, different businesses have different problems and different areas that can be solved differently. That’s the biggest focus for your business.

[00:23:43] What is the problem statement that is either most efficient... And a lot of people talk about what’s not fun in their jobs. That’s an easy way to talk about what can be automated for us. But there’s challenges everywhere, and which ones do you pick to solve, and do you have something new and unique to solve it?

[00:23:59] Bobby: Yeah. Speaking of, in August you launched Kai Link, and it’s in... Is it still in beta? I think it’s still in beta.

[00:24:05] Phil: It is in production. We haven’t sent out an official launch yet, but it’s ready to go now.

[00:24:09] Bobby: Do you wanna explain that, Phil, real quick for folks?

[00:24:11] Phil: So as we’re looking at different technologies, we wanted to build a connector into Koozie Group. When you think about what agents can do and how they work and where you want to have that control, we decided that whoever’s using this needs that control layer on their side. So we didn’t offer agents to our customers.

[00:24:30] We offered a technology layer that their agents can work in, and it’s a new way to deliver information to those agents so they can work more efficiently. Agents can scrape websites, and they do it at varying degrees of success. When you give them the data directly, the success ratio is increased quite a bit, because it has access to structured data that’s consistent. We’re not delivering an agent that may or may not do as we hope, or might change, or different things. We wanna deliver information in a new way that is usable for people who are looking to use agents in their own business.

[00:25:06] Bobby: Phil, how’s it going so far?

[00:25:08] Phil: We’re getting some interesting use cases coming out of it, and I think that’s the biggest thing. Not only this, but I meet with a lot of customers and first explain what it is and how it works and why it would be better. I think it’s interesting.

[00:25:20] It’s always tough for me. As a technologist now, I’m not gonna buy a new software unless it has an MCP connector for me; that’s just a prerequisite.

[00:25:28] Bobby: Yeah.

[00:25:28] Phil: I think that’s gonna be the world that we live in for all things in the future, but there are different levels of AI experience and capabilities within these organizations. And so it’s still in discovery mode quite a bit, and people are figuring out how to use their agents; now they’re deciding how to plug in and what business problems it solves for them.

[00:25:45] Bobby: Yeah. I tried it. The first thing I asked was the orders that were in production last week, as if I were a distributor, and so that was the first thing I did. We’re working on MCP too, and I’ve noticed in my own behavior, buying or sort of user behavior, right, I’ve got a connector for everything now, it feels like: HubSpot, Asana, everything you can imagine that we use in our own workflow. And it’s become that headless kind of option. But I have a question. I don’t know if I can unpack this well for you two, because I’m out of my depth when I’m talking with you two. But data has been something you have been involved in for a long time through PromoStandards. It feels like there has never been a more important time, and maybe that’s too dramatic, to get this right in terms of data and how important it is, because of the ability for agents, in the future and now already, bots talking to bots, interacting with bots. Am I right on that assumption, Mike?

[00:26:39] Mike: I think it’s always been important. Maybe it’s highlighted a little bit more because of what AI can do to expose it.

[00:26:45] I think as an IT person, I’ve always thought that was an important thing. It’s not an easy thing, and that’s why most people who have bad data don’t have it because they don’t think it’s important. They don’t have it because it’s not easy. And so if AI is a catalyst for, “Hey, I know it’s hard, but we gotta go do this,” as a motivator for folks, that’s fine.

[00:27:06] I’ll take that.

[00:27:07] Bobby: Yeah. How about you, Phil?

[00:27:09] Phil: Yeah, same thing. Data’s been one of the most important things forever for anything that we’ve been doing. It does feel more important because bad data gets exposed easier and people find things. This industry has a lot of implied data or implied information or implied understanding that’s not structured in the data across a lot of areas. And that’s always been—

[00:27:29] Bobby: What do you mean by... Can I interject on that? What do you mean by “implied”?

[00:27:33] Phil: I think there’s general expectations that come out. If I’m a distributor in this business and I’m trying to understand... I say all the time, we have 3,500, 4,000 products. Mike has hundreds of thousands of SKUs. Understanding what all those things can do, I think as humans, they kinda hone in on the things that they can understand. I can imply the quality of this shirt. How does AI imply the quality of this shirt? Quality of decoration method, quality of supplier, quality of everything is still... That doesn’t necessarily sit in the data. It can, or there can be judgment and different things added to that. That’s a big thing in this industry as a whole, where people have been able to get away with less quality data because humans can understand or make assumptions that are not necessarily there.

[00:28:18] Bobby: Wow. And so you’re saying, I think, that that is going to be less forgiving as we move forward.

[00:28:24] Phil: Yes.

[00:28:26] Bobby: I’ve described this arc of the industry from word of mouth to printed catalogs to websites. And now the agentic era is either here or on the horizon. And as someone who’s playing with some of these tools all the time, I have Grok Bot scheduling appointments for me.

[00:28:42] I have Muse buying books for me. There’s interesting little fun, easy things. But when you look at promo, this is a complicated business. I always say that we don’t pause enough to celebrate just how miraculous it is to get complex orders across the finish line every day, because it’s a custom order all the time. And I remember talking to someone who came in as a supplier from the retail side, and I said, “What did you think about this industry?” She was amazed. She said, “The complexities we had in retail are nothing compared to what you have going on in this industry.” What do you think positively about this new era we’re entering?

[00:29:15] And then what do you think we should be most concerned about? And Phil, we’ll let you go first, and then we’ll hop over to Mike.

[00:29:21] Phil: I think from a positive piece, I think it’s going to expand the ideas and expand the horizons of the offerings that are out there. I think people are gonna do a better job of finding that uniqueness that they’re interested in finding, versus trying to understand that. I think that’s really exciting for the industry as a whole. You have a custom product ’cause you want something unique and memorable. That’s the core premise of the entire industry. And as a human, you’re only able to capture a certain portion of that or understand what that means. I think there’s gonna be additional drivers that’ll continue to help people come up with something unique and individual.

[00:29:57] There’s amazing products in the promo space that everybody has that I’m sure never made it off the ground because of visibility, or people didn’t understand what they could do. And I think it’s really exciting that some of these new and unique things are gonna surface their way into the marketplace and into people’s minds, and onto people’s desks, where they’ll stay and show good value.

[00:30:18] Bobby: Any concerns? Or, I should say, primary concern?

[00:30:23] Phil: I guess in the same thing, the one thing that concerns me just in general about AI is, does it turn into groupthink on its own when you don’t understand it? The other side of that is, even though it’s non-deterministic, as we talked about earlier, can it be generally the same thing?

[00:30:38] And will we lose creativity on that? And I think that’s a big driver for people who are shopping for products, making sure that they are not losing their own creativity as they’re thinking about what it is and what they need. And if everybody asks for the same items for the same events and the same things over and over, then everybody’s gonna have the same thing back.

[00:30:57] Creativity is still, in my mind, a human piece. There’s pseudo-creativity in AI and LLMs, but as people use the same models and do the same things and start getting the same answers, then you end up losing that creativity.

[00:31:09] Bobby: How about you, Mike? Most positive, and then maybe your overarching concern.

[00:31:14] Mike: Yeah, I think I echo what Phil has to say. I think people’s ability to use AI to help them, augment them, as opposed to replace them, is an exciting opportunity. And in our industry, there’s, I think, lots of places where that can be true. But I share Phil’s concern, right?

[00:31:32] This idea that there’s a uniqueness value to what we do in the promo industry, and if everybody has the same thing, that it loses some value. When you say my biggest concern, my biggest concern is that we as an industry, but even more as a society, strike the right balance of what’s AI here to help us do versus what do we think is innately human, and therefore is necessary for us to maintain the human piece of things.

[00:31:59] Bobby: Yeah.

[00:31:59] Mike: That probably sounds a lot deeper than I mean it to be, but that’s where my head’s at.

[00:32:04] Bobby: No, a lot of people, I think, are thinking that way too. I’m curious, one last question on this with AI. The channels have gotten far broader, in the sense of, like, just talking about Phil and your example with MCP. You have MCP now and this sort of channel there, if you will. Then you have platforms like commonsku that serve nearly 1,000 distributors with thousands of users, and you as a supplier are trying to get your product on all these platforms in all these different ways. Are we just gonna see this continued dispersion, if you will, or diffusion of different channels and things?

[00:32:37] How do you think that’s gonna impact? How do you guys think of getting your product and technology right for all of these different ways people wanna use it? Is this a blue ocean opportunity or challenge, or both?

[00:32:49] Phil: Yeah, I’ll go first. Yeah. It’s really interesting. That’s a challenge that we have, picking and choosing technologies and places to deliver our products on different platforms in different areas in different ways. We wanna be the easiest to work with and have an experience that can match everybody’s experience, from the large online technology platforms to the person who’s sharing with their small business, “Here’s what’s gonna make your business work better.” And that’s a struggle that I think every supplier has, and then the entire industry. What platform are you on? How do you find products? I don’t know which direction it’s gonna go, personally. There’s a lot of possibilities for this to flatten, increase.

[00:33:27] Any new technology is a place for the best people and the best industries to accelerate, and I think that’s the exciting part. If you think about it, you have something that’s good, use this technology to accelerate what you have and be better at it, and that’s going to help you be there.

[00:33:42] There’s not one website that sells stuff, right? There’s not one Amazon. Even how amazing Amazon is, there’s challengers, big and small, that are out there, and I don’t think we’ll ever solve all business problems in the world in a way that’s gonna fit every model that’s out there, from size to scale to speed to selection to service. That’s not gonna be the case, at least not in a way that I imagine.

[00:34:06] Bobby: Yeah. I sit in Fayetteville, Arkansas, and live in Fayetteville, and we’re in Walmart’s backyard right here, so I’m watching all these battles take place between Walmart and Amazon, and all of the bots and agents and things like that. It’s a wild world. But that’s a great positive take on that.

[00:34:18] Mike, what say you?

[00:34:19] Mike: Yeah. I think our goal is to be where our customers are, right? It’s funny to hear Phil say he wants to be the easiest to do business with, and the problem with that is, no, I wanna be the easiest to do business with. So one of us has to be wrong, because we can’t both be the easiest.

[00:34:33] But I think it’s a good way to look at it. As a supplier, our goal is to support our customers in their growth, and one of the ways we can do that is to be where they wanna be and to reduce the friction in the transactions that they’re trying to do. The smoother things run for them, the more time they have to spend however they think is most valuable, whether that’s more time with their customer or more time with their family.

[00:34:55] Us making lives easy for our customers is a motivating factor that puts us in lots of different places and is one of the goals that we have to try and make our business smooth.

[00:35:06] Bobby: Yeah, that’s cool. Last question. It’s an easy one. Is there anything you’ve read, heard, or viewed lately that has impacted how you think? It doesn’t have to be business related. It can be anything. Phil, we’ll let you go first.

[00:35:17] Phil: All right, I’ll give you my most recent favourite one. It just talked about AI and how it’s changing businesses in the world. It was a great speaker, was talking about electrification and how that was such a big technology difference in industries during the time. Electricity came in, and some people put lights in, did the same thing with lights, and it added extra time, and it added extra capabilities. I think of that like the personal productivity function of AI now, right? It’s nice, it’s adding lights. The companies that won electrified their entire company, but then they also used that to rearrange their whole business flow. Old systems that had steam-powered churning axles that were moving machinery were structured around that basic thing. Electricity brought flexibility to those businesses, and the ones that took advantage of that flexibility were the ones that win. And I think AI is adding flexibility to businesses. It’s adding flexibility and capabilities that we don’t even know what they all are yet. But focusing on where it can make you more flexible and where it can make you better...

[00:36:18] That was super impactful for me to just continue to think about how we apply AI and how AI is going to continue to evolve in our industry.

[00:36:26] Bobby: I love the flexibility you mentioned. That’s fantastic. Mike, let’s close this out here.

[00:36:31] Mike: Yeah. Kind of along those same lines, there was a conference here in Seattle, and Malcolm Gladwell spoke there. He’s got a new book coming out, so it’s one of the reasons he was a speaker there, but I’ll paraphrase something he said when he was talking about AI and people’s reticence, the group of people who are reticent to adopt AI in terms of how they do their job.

[00:36:49] And his question, he got this from interviewing, like, an open-heart surgeon. So it’s a long story. I’m sure it’s in his book. But the question was: Is it reasonable to think that you will do your job the same way at the end of your career as you did at the beginning of your career? And if that’s reasonable to you, then you’re not really evolving.

[00:37:09] So I think about my career in IT, and over the years, I don’t do things the same way I did them before. And as I look at AI as a tool, it’s a step along the same progression I’ve been going through my entire career. I do things differently now than I did five years ago. Five years from now, I’ll do things differently.

[00:37:26] And his suggestion around it is that’s a way to frame AI for people who are in that mode where they’re not sure why, or that they want it to be part of their roles going forward.

[00:37:38] Bobby: Yeah, that’s fantastic. Two really good pieces of advice there to end on. Mike and Phil, thank you guys so much for taking the time to spend time with us, especially during this busy... It’s always busy, but we’re heading into fourth quarter. There’s always tech challenges. Thank you guys for doing this.

[00:37:53] Mike: Yeah, appreciate it. Thank you, Bobby.