Latest / Talk Commerce / CPQ Software and Deterministic AI for Complex Configuration with Tim Baynes of Compatio
Transcript
- Brent Peterson (00:01.56) Welcome to this episode of Talk Commerce. Today I have Tim Baynes He is the CEO and founder of Capatio. There we go, Tim. Go ahead and do an introduction for yourself. Much better than I just did. And tell us your day-to-day role and one of your passions in life. Tim (00:13.325) That's right. Tim (00:23.096) Yeah, no, thanks Brent. Thanks for having me on, really appreciate it. Yeah, so it's Compatio AI and we focus on solving problems related to selling, discovering, guiding, selling customers to complex products, could be configured products, products that have complexity in terms of how they go together. So what's one of things I like to do in life? I like to roast my own coffee and I do a lot of cycling. Well, all of the above. I do a little bit of mountain biking. I do gravel biking, which is really the dominant passion. I do some road biking as well. And my wife and I have a tandem that we love to get out on. I, you know, I got four different ways I can go get on a bicycle. Brent Peterson (00:55.36) Alright, road cycling like pedal biking. Brent Peterson (01:16.94) Well, you know, this is an AI thing, but the formula for getting for the amount of bikes you need in your garage is N equals N plus one. You probably already knew that. Tim (01:25.794) That's right. Yes. Yes, although my wife would probably say at this point n minus one is really the formula. Brent Peterson (01:32.974) My wife would never go on a tandem, so I salute you for that. All right, so Tim, before we get started though, I do need to do the free joke project. I'm gonna tell you a joke and you just give me a rating eight through 13. So here we go. Smoking will kill you, bacon will kill you, and yet smoking bacon will cure it. Tim (01:57.262) Hmm, okay, I'll give that one, I'm sorry Brent, I'm gonna have to give that one a nine. Brent Peterson (02:02.904) that's all right. Thank you. All right. No problem. Good. All right. So Tim, give us a little bit of your background. You know, I think we both talked a little bit about my background and that, you know, I was in the magenta space for a long time. So I'm very familiar with complex products, but tell us how you got to where you are now. Tim (02:25.451) Yeah, so thanks Brent. So I've been working with product configurators as we used to call them. We now call it CPQ, but I've been working with product configurators since about 1996. I've implemented, designed, architected product configurators, a la CPQ systems for I've been involved in roughly about 175ish projects over the years since 96 implemented them for companies all over the world, Jaguar, Johnson Controls, Cummins, PowerGen, Master Lock. So I've been doing that for a long time and working with KPMG in the 90s, then Oracle for about a decade. Gartner, was research director covering CPQ at Gartner. And then prior to launching Compatio, a company out of Denmark called Configure, very, very good company. And over those years, it became very apparent to me that there was a big challenge in the marketplace. Most configurators in the world or CPQ systems are targeted for manufacturers. They are build to order configurators. They allow you to specify a product that will be built for you. But the challenge that I kept seeing was that there are an awful lot of configured solutions in the world that consist of products from multiple manufacturers. If you get into HVAC systems or material handling systems or fluid handling systems or building security systems, all of these are systems that consist of products from multiple manufacturers. Some of those products may be individually billed to order, but you still got a custom solution or a Tim (04:37.58) complex solution with products that come together. so I became aware of this in probably 1999 or year 2000, but it took me a long time to figure out how I was going to address it. Compatio was designed to address that problem. We provide solutions that allow a customer or a sales agent to bring together products from multiple manufacturers at the point of sale, which might be a quote system, might be an e-commerce system like Magento, bring those products together and create a solution that is tailored for a particular buyer's needs. That's what Compatio is all about. at the heart of Compatio is this sense of product compatibility, what goes together. And then on top of that is a configurator that allows you to put together complex solutions made up of many products. So that's what Compatio is and that's what we're doing today. Brent Peterson (05:42.21) Yeah, and I have to apologize that I started off thinking it wasn't as complex as it is. It's obviously a lot more complex. And number two, I do write a lot of stuff in tandem with Gartner and I never really knew what CPQ was and I never looked it up. So learn something today two times over. So, you know, just going kind of coming back to this idea of on demand. just in time configuration or just in time point of sale, right? And this is more than just your, you're looking at more than just the average, the minimum would be like a t-shirt where you're configurating a different product, but you're putting together complex things like a computer or car or whatever, right? Tim (06:31.807) Yeah, so a lot of the e-commerce systems, they have set up configuration. It's really about managing the variants I've got a t-shirt. It can come in three sizes and five colors. Those are really variants. Brent Peterson (07:00.652) Yeah, I hear you. It just says your recording stopped. Tim (07:03.071) Can you hear me okay? Okay. Yeah, you may have to restart here. Sorry for messing up your recording here. I'm sure you can edit it. Brent Peterson (07:10.178) No, it's all right. If you think you come back in, should, it'll bring us back and we'll splice it together. It says your media, Tim's recording a stop due to media disconnect. Tim (07:23.048) Do I need to do anything for my side? Brent Peterson (07:26.542) It looks like you're still recording now. Tim (07:28.2) Do I need to do something? It says it is recording. Yeah, my left earbud went out. I try to trade them back and forth because I'm on calls all day. OK, so let's go ahead and I'll start and respond to your question here. Yeah, so many of the e-commerce systems have a concept of configurability, which is really about just Brent Peterson (07:39.202) Yeah. Tim (07:56.462) managing variants. I've got a product like a t-shirt that may come in several colors and several sizes. We're going to guide the user, allow them to select the color and the size, and you're going to arrive at the right variant. That variant probably is already defined and it's sitting there as a skew in the system. Configuration gets much more complex than that and in terms of Magento can happen in a couple of ways. Number one, you might have variants that aren't expressed yet, meaning that you've got options that aren't set up as SKUs yet. Maybe that t-shirt can come in, maybe it's custom colors and we can have thousands of colors and three to five sizes. We're not gonna want to go represent that as a bunch of SKUs just yet. you can do with a of a higher horsepower CPQ or configurator system is to allow those variants to be generated on the fly. And you can have any size you want and any color you want and user will pick size and color and then we're going to go generate a SKU to handle that that will represent that particular variant. That's one path that we can go down. But you can also get into a more complex scenario, again, like sticking with an e-commerce system for a minute, like a Magento, where multiple products need to be brought together to configure a solution for the customer. Let's take the example of a bike, because we were talking about cycling before we got on the recorded interview here. you may have a frame from Specialized and you want to build out that bike. It may combine various drivetrain components, a derailleur and a crank set and a rear sprocket and a chain and shifters. You need to put handlebars on it. You need a seat post and a seat. You need wheels. You need tires. You need all of that stuff that goes into allowing you to build out the full Tim (10:24.497) bicycle, all of that stuff has to go together. You can't put any crank set on any frame. It just doesn't work. So there's a compatibility there. So Compatio is designed to solve those kinds of problems where you could not only handle variants, which is kind of the one of the less complex cases that we solve, but you can also combine products in multiple ways. which is accurate. We ensure that if you put that bike together, that sprocket, that cassette that you ordered for the rear of the bike is going to go with the chain, is going to go with the derailleur, is going to go with the crankset. It's all going to go together in a full configuration. So that's configuration kind of at the next level and then there's a kind of a more complex scenario on top of that. which is where it's build to order, where you're going to go configure a product specifically for your needs. And that product's going to be manufactured for you furniture, I want this love seat for my living room. And I wanted in this material and this size with these cushions and and that's going to be built and shipped to me in six or eight weeks. That's build to order configuration. And then the most complex scenario is where you've got a mix of both. We've got a solution, multiple products that go together, but some of those, one or more of those products are actually build to order configurable. That's a very complex scenario like You get into equipment like an HVAC system for an office tower. You've got water chillers and air handlers and variable valves that are in the ceiling. All of those are configurable products and it's all got to come together. So we kind of, we really handle everything from variant management all the way up to complex systems. Brent Peterson (12:29.516) When I look at a builder, like if I were to go onto a car site and they all have a build order, that's really assembled order, right? That's basically, they've pre-configured everything. Yeah. Tim (12:43.311) Yeah, yeah. Build to order, assemble to Yeah, build to order, assemble to order. Yeah, that's correct. You go configure a Mustang, you're gonna, yeah, that car is going to be built for you. That's a build to order, assemble to order scenario. Brent Peterson (12:51.904) And so are you moving, and I'm just making an assumption that most of the CPQ was in the enterprise space. Are you trying to move it down into sort of mid market or tell us your ICP on this type of product? Tim (13:09.467) Well, yeah, so you're right. Many of the CPQ systems that are out there in the world, or at least historically over the last couple of decades, they've been targeted to single manufacturers and are largely about build to order, assemble to order. They're going to allow you to specify a product that's going to be built for you. And these are mostly B2B products. There are B2C products out there. Master Lock sells some, can configure a lock. Mustang, Auto, that's a B2C application. Configure the Mustang or the Jaguar or the Mercedes or the Porsche for yourself. Those are all build to order scenarios. sorry, I forgot your question. We're gonna have to edit that one. Brent Peterson (13:59.298) No, I was just saying, are you moving it down from enterprise to mid market? Tim (14:05.774) Yeah, sorry. Okay. Okay, let me try. Yeah, so most of the applications, most of the CPQ applications are out there are, are implemented for specific manufacturers in our build to order scenarios. What we've seen is that there are there's a huge number of configuration challenges in the world that combine products from multiple manufacturers. And that's where we focus. So that means that we're dealing our ICP, they're either large distributors with large catalogs that are selling complex products that must go together, industrial automation, electrical, HVAC, building security. Those are not usually build to order components that's off the shelf components, but they are being brought together and there is still configuration. They all have to go together. That's where we primarily play. We do do build to order. So we do sell to manufacturers that have build to order products, but the dominant play for us is really at the distributor level. You've got big catalogs of products that have to go together. or manufacturers that are now big enough that they have many brands of products, portfolio of brands, and they want to make sure that they're cross-selling across all of those brands. So we're seeing that scenario pop up a lot where you've got companies that are grown by merger acquisition or it's a PE funded roll-up. They're trying to create, they're trying to acquire brands and they want to create some synergy across those brands. Well, those brands, there may be compatibility there. I can sell this custom door, but now I need molding from my other division to go around the door. How do we make sure that's all compatible? That's what Compatio does really, really well. We can cross sell across many products and facilitate a fully compatible solution that totally satisfies the buyer's needs. Tim (16:29.321) even where some of those products are configurable in and of themselves. Brent Peterson (16:30.062) If we were to say, I don't want to say the olden days, but let's just say in the 80s, all this was just done on paper or was some of this done with mainframes or this is, mean, an AI has helped accelerate this process, right? Tim (16:50.295) Yeah, so it's funny because configurators really are AI. They were AI in the 80s. They were one of the most successful types of what we used to call expert systems. And yes, in the early ones, like the one, the first really commercially viable configurator that rolled out was by a company that you may have heard of called DEC. and it was a configurator for their vac systems as I remember. But yeah, and it ran on a mainframe or at least a mini. And yeah, now fast forward to today and yes, what we do now is quite different in some ways, but at the heart of the problem, Configurators still have to be grounded in symbolic logic. Even a high-end LLM like Claude or ChatGPT or Genesis or Gemini, they still have a tendency to make stuff up when they don't really know the answer. You can't do that with configuration. You have to get it right. If you're selling an industrial automation system, where you've got a motor that's controlled by a variable frequency drive and that's got some fuses in it. And these are big fuses. This is big voltage, big amps. You can't get it wrong. You can't make it up. If you're gonna put a fuse in that, you gotta get the right fuse. And that's not something you're gonna wanna go ask ChatGPT about. And so where Compatio sits is we utilize that core symbolic logic that really came out of these expert systems from the 80s and 90s, but yet we combine it with modern AI that allows us to generate new solutions while verifying that they're correct and or match solutions to make the optimum recommendation for solutions. Tim (19:07.65) to a buyer specific needs and that where that optimization can be based on a number of factors. might be optimizing for value. We might be optimizing for margin by the seller. We might be optimizing for quality. So we can kind of combine this state of the art AI based match it to your needs, but ensure that it's actually correct at its core. And that's what we mean at Compatio by we call it deterministic AI. or what our term is real intelligence because it is somewhat artificial but it's also grounded in human expertise that knows whether things are correct or not. Brent Peterson (19:47.342) Yeah, would you say the difference is in the past we'd have learning models, but they would be blank and they wouldn't have the ability to make stuff up? Like I can't put a copper penny in and replace the fuse if it doesn't exist. It wouldn't make something up like that where now we have this sort of added ability to LLMs where they've added a new dimension to, hey, I'm gonna get you an answer almost no matter what. Tim (20:06.196) Yeah. Tim (20:17.429) Yeah, yeah. Well, I mean, in the past, they didn't really learn at all, right? It was purely symbolic. mean, a human, an expert would provide the knowledge and then a programmer would sit there and encode that knowledge into symbolic rules and everything was purely deterministic. And the problem with kind of what I think of now is kind of old fashioned CPQ. is that it's purely deterministic and therefore it's not very good at making recommendations of optimal solutions if you stay on that course, that very deterministic course, and it can't learn. And so what we can do now is feed analytics back into the process and Configurators can allow you to represent sometimes billions of possible configurations. That's what they do. Which of those billions of configurations is the right one for you? Well, there could be symbolic logic that guides you through a path. Pick this option, that option, that option, and then you come up with the configuration. But maybe some people in your region, in your industry, with your application, have already chosen this kind of an app, this kind of a configuration, and it's going to fit your world very effectively. We can actually discern that and recommend that very early in the process to save the user a lot of time, but also help to provide analytics back around what sells well and what doesn't and allow the manufacturer or the vendor to sort of trim their solutions set to stuff that's really making money. So there's lot of different ways we can go with this, but combining machine learning techniques with that symbolic logic can create some very powerful, very effective solutions, and that's what we do. Brent Peterson (22:21.55) is the biggest wins that you're seeing, not just in cost reduction, but also just in the speed in which things can happen. Tim (22:34.538) Yeah, actually where we're seeing the best fraction has to do with guided selling, which is the process of asking the user for it could be starting with different personas of user for right off the bat. Where are they? What region? What language? What's the role? Because these are B2B scenarios typically. then then mapping who they are against what are they trying to do guided selling will ask them what do you want to do? I'm buying a tractor. Well, what are you going to do with that tractor? you going to you know, you're going to farm you know, large fields of you know, 500 plus acres in in southern Missouri. What are your crops? I want to do corn. I want to do tobacco or whatever. So we could map their requirements to the right kind of solution and then allow them to go configure it. That's what guided selling does. So that's where we're seeing most of the traction right now is around using our combination of both configurability with recommendations. We've got, we started out as a recommendation company, so we've got very powerful recommendation capability. If you combine recommendation sort of state-of-the-art recommendation with state-of-the-art configuration, you end up with a very powerful guided selling tool and that's what we offer. Brent Peterson (24:09.14) One thing that I can remember this transformation on B2B going to just getting off of their ERP or phones onto the web, this digital transformation, do you think that this is also kind of bringing a lot of legacy, larger manufacturing or distribution businesses into the digital age where now they're enabled to really do a lot of these complex things without the... with almost letting the end user do it or having the guided selling portion do it. Tim (24:44.415) Yeah, absolutely. we're really, we're seeing three major trends. One is B2B digital commerce is finally catching up to B2C digital commerce in that things are becoming a lot more self-service. So B2B companies, manufacturers and distributors that sell complex products now want to make that process of buying those products much more self-service. So that's a trend, sort of digitalization. The other trend that we're seeing is frankly a bit of a crisis having to do with product expertise. And we're hearing this from our customers constantly that the folks that really know their products well enough to sell them And these, again, we're talking complex products where you got to know a lot to sell these things. HVAC, industrial automation, material handling. These are complex domains where you have to be very knowledgeable to recommend the right kinds of products. The challenge is that across all these industries, the folks that know enough are now they're 30 years into their careers and they're retiring. And it's very challenging to go find the right people to replace them and keep them on board long enough to train them to be that good. It takes 10 years. And again, so we're hearing constantly, it's, as I said, a bit of a crisis. so you have, and then the third, factor that we're seeing is just the rate of technology change, how quickly products are changing. So you've got a desire to digitalize, get your products out so they can be bought on a self-service basis by your customers, whomever they are, combined with we don't have the expertise to staff that, combined with products are turning over rapidly, Compatio really provides the solution to all of that. Tim (27:05.733) We can encode expertise. We can learn configuration rules from data. We can place this out in self-service channels like e-commerce sites. And our system is designed to deal with high velocity of product turnover, which the older configurators don't do. We're a lot better. at dealing with catalogs that are very, very dynamic. So all of that together is what we're responding to a lot of trends all at once. And they have to do with expertise, digitalization and product advancement or rapid product technology acceleration. Brent Peterson (27:42.926) Oh yeah, Tim, so we've, we're rapidly running out of time here. We have a few minutes left as I close out, I give everybody a chance to do a shameless plug about anything they'd like. What would you like to plug today? Tim (27:58.771) Well, well, so can I plug world peace? you know, yeah, really happy to see that. And, but what we do is we help companies sell a lot more effectively in digital commerce channels. We solve problems that the traditional configurators or CPQ systems don't solve. Brent Peterson (27:59.444) Absolutely, go for it. Tim (28:28.275) And we do it in a way that's very intuitive to help users discover what they need to buy, put together solutions that suit their needs, and book the orders. So that's what we think we do better than anybody. So I'll put in that plug in addition to World Peace. Brent Peterson (28:42.158) That's awesome, thank you. Tim Baynes is the founder CEO of Compatio. Thank you so much for being here today. Tim (28:52.976) Yeah, thanks Brent. Thanks for the time.