Intro/Outro (00:00:03):
Welcome to Supply Chain. Now the voice of global supply chain Supply chain now focuses on the best in the business for our worldwide audience, the people, the technologies, the best practices, and today’s critical issues, the challenges and opportunities. Stay tuned to hear from Those Making Global Business happen right here on supply chain now.
Scott Luton (00:00:33):
Hey, good morning and good afternoon. Good evening, wherever you are, Scott Luton and Greg White with you here on Supply Chain. Now welcome to today’s show, Gregory. How we doing today?
Greg White (00:00:43):
I’m doing well. This is coming off
Scott Luton (00:00:45):
Smooth
Greg White (00:00:45):
As glass, don’t
Scott Luton (00:00:46):
You think? <laugh>? That’s right. As smooth as Greg White and that’s world class smoothness folks. Lemme tell you. All right. So Greg, man, we are tickled to be kicking off a new limited run feature series here today on supply chain. Now we’re partnering with Noodle ai, a dynamic organization driving powerful innovation in global supply chain. And today, Greg, as you know, we kick off episode one of this new series that’s entitled, making Better Supply Chain Betts with the Power of Probabilities. And today we’re gonna be focused on building a strong
Greg White (00:01:20):
Foundation.
Scott Luton (00:01:20):
Really deme, I can never say demystifying, demystifying. I gotta go real slow, real slow. Two miles an hour. We’re gonna be focused on demystifying artificial intelligence and talking about real opportunities and outcomes for AI in supply chain. Greg, are you, uh, excited as I am here today?
Greg White (00:01:39):
What can I say, Scott? I’m, uh, excited, enthused, intimidated. Rarely do we get to talk to people
Scott Luton (00:01:45):
This much smarter
Greg White (00:01:47):
Than we are. Usually they’re this
Scott Luton (00:01:48):
Much smarter than we are <laugh>, so,
Greg White (00:01:50):
Yeah. Yeah,
Scott Luton (00:01:51):
No, seriously,
Greg White (00:01:52):
I’m excited. I think, you
Scott Luton (00:01:53):
Know, Noodle’s not a new kid
Greg White (00:01:55):
On the block. They’ve been doing this AI thing since
Scott Luton (00:01:57):
Before
Greg White (00:01:58):
AI was cool.
Scott Luton (00:01:59):
Mm-hmm. So, yes. So with that said, let’s introduce our featured guests here today. I wanna welcome in Steve Pratt, founder and c e o with Noodle ai. Steve, how you doing?
Steve Pratt (00:02:10):
I’m doing great. Thanks Scott.
Scott Luton (00:02:12):
You bet. Really have enjoyed our pre-game already. And you’ve brought with you Professor Diego Kian with Northwestern University Diego, how you doing?
Diego Klabjan (00:02:22):
I’m good, thanks for asking. Great. Thanks for having me.
Scott Luton (00:02:25):
Well, great to have you. And, and Greg, we’ve gotta point out, we’ve got lots of sports allegiances, but Diego is a fellow Atlanta Braves fan, in addition to being a Cubs fan like Steve. Is that right, Greg? That’s what we learned in the pre ship.
Greg White (00:02:39):
Yeah, I heard him say Braves, so I’m gonna take that as fandom. <laugh>. Yes, <laugh>.
Scott Luton (00:02:44):
Alright, good, good, good. So Steve and Diego, great to have you here. Looking forward to a great baseball postseason and a great discussion here today. So I wanna start with this. So Steve, you’ve got an incredible background, man. We’d be here for days if we walked through it all. We really, really enjoyed our, our pre-show conversations I mentioned, but I wanna start with our team’s. Done a little intel, a little digging on your background. We understand you’re a parttime, grape And olive farmer, So you gotta tell
Steve Pratt (00:03:11):
Us more. It’s guilty as charge. Yes. I, so I have a ranch up in Napa County in St. Helena and got about 500 olive trees, several acres of grapes. And, and so I Actually Taught myself grape farming from YouTube, Videos, books, and, uh, a Yoda that lived down the street. In fact, we had our first harvest last Friday and it was four times what last year’s harvest was. Wow. And so we’re, which is like good news, bad news, <laugh>. ’cause it’s so much, it’s so much wine. Like we hadn’t planned for that. And so, but yeah, I, I think it’s connects me with nature, get to see the cycles of the seasons. I think. Well up there it’s sort of my, not quite my fortress of solitude, but something like that. <laugh>.
Scott Luton (00:04:06):
Well, Greg, if Steve is looking for, uh, regional fortress of sobriety, either <laugh>, yes. Right, right. Steve, if you’re looking for distribution representation in Southeast, hey, give us a call. We can help you get rid of any excess wide. Alright, so let’s switch gears here. And by the way, you’re paint a pi, pretty picture, 500 olive Trees there in California, I bet. No wonder it is a, a way to unplug and, and, and step away from all the cool things you’re do in the industry. Alright, so Diego, now this is a pretty unique piece of intel we gathered on you, Greg, our team says that Diego’s one of his biggest hobbies that you, that he wish he had more time to spend on mm-hmm. <affirmative> is coding. So in particular, Diego, we hear you love to explore and learn new computer programming languages. So I gotta ask you, when did this passion begin and what’s One of Your newest languages that you’ve tackled?
Diego Klabjan (00:05:00):
Yeah, so I, I liked computers and codings from my very early days from high school. That goes way back, what, 1980 roughly. So I started coding at that time, literally just for fun and I still love it. So I, as I said, I wish I, I have more time doing the actual coding. So my lady, sort of a computer language that I started exploring about a year ago, and I spent a few days working on it during the summer is rust, r u s t. Not sure if you’ve heard about it. So it’s a modern language that’s supposed to be called safe. Safe from the security perspective. So you kind of cannot do, you cannot create memory errors. Right. Okay. So that’s, that’s my passion. But now’s okay. On the less sort of professional side, I actually do enjoy a lot running marathons and I’m gonna run now on Sunday in the Chicago marathon. Oh, wow.
Steve Pratt (00:05:54):
Wow.
Scott Luton (00:05:55):
So do you have a target?
Diego Klabjan (00:05:57):
I’m Getting old and older by the year, so my times are declining every year, so I do have it. So I do have it. I’m not proud of it anymore.
Scott Luton (00:06:07):
You’re better, you’re better off Diego than probably 98% of the rest of us. So keep running
Diego Klabjan (00:06:13):
If you want a number. Three 30. So three and a half, three hours. Okay.
Scott Luton (00:06:17):
Three and a half hours. Wow. That’s pretty solved. I’m writing that down <laugh>. So lemme go back though for a second. And Diego, you’ll have to let us know how you finish. We’re gonna, we’re writing that down and we’ll do a check on you. Greg, you, you were nodding your head when he said rust. Are you familiar with that, that computer language, Greg?
Greg White (00:06:33):
Nope, Not at all. But I have people who are familiar with it also. I’ve heard Them say it a lot lately. <laugh>. Yeah, I, I can name a lot of Languages. I haven’t Been allowed to write in any languages, including English since 2011. So.
Scott Luton (00:06:50):
<laugh> well, along those lines, I’ll share a little bit of that with you. When I started college, I started in computer science, Diego and Steve, as Greg knows, and I, I had one, one semester of c plus plus, and that was all I could take. So that was not my passion. Like it is perhaps both of yours, but nevertheless, we gotta move on. We got a lot to, a lot to get into today that I think is gonna really inform, educate and in enlighten and de uh, demystifies. We talked about a lot of our audience because of course, you can’t have a single conversation these days without saying AI or gen ai or both, or you name it. So the key focus here today is demystifying ai and of course, in particular talking about how to apply generative AI to global supply chain. And we’re not talking about chat G p t, Greg, not talking about chat, G P T. We’ll talk, uh, more on that in a second. So let’s build a foundation. So let’s start with you Steve, if you would please, you know, clarify what gen AI entails and how it sets itself apart from all the conventional machine learning, artificial intelligence techniques. Tell us more, Steve.
Steve Pratt (00:07:59):
Right, so the, the breakthrough in generative AI is that it produces new content that never existed before. So It’s predecessor or the other main category of ai, which is predictive AI would take the things in his memory and would label them, or he would say like, find a picture of a dog and it would label this as a dog. Whereas generative AI would draw a picture of a dog. The very important construct that most people don’t understand at this point is that there are really three general categories of generative ai. There’s large language models, which is getting all the press, which is text and documents. Then there’s large image models which generate unique images from some prompt. And then there’s large graph models which are used Basically modeling data as a graph. And the connections among the nodes in that graph. So it’s large image model or large language models, large image models and large graph models. It’s very important to have that construct when you’re talking about generative ai.
Scott Luton (00:09:09):
Very helpful. L l m. Mm-hmm. <affirmative>. L I M L G M. We love our acronyms around here. Don’t look Greg <laugh>.
Greg White (00:09:16):
Um, yeah, that’s a really, that’s really good description, Steven. And under 30 seconds. So Albert Einstein approves <laugh>.
Scott Luton (00:09:23):
Yes. Love that. All right. Diego, what would you, how would you expound, uh, on that one? Talk about the foundational, what gen AI entails.
Diego Klabjan (00:09:34):
Right, so I would say, so generative ai, it’s really how do you create creative output that’s actually similar to your historical data. So in with technical terms or historical data, it’s also called training data, right? So I’m gonna be using the term training data quite often, right? So here’s a, a use a possible use case really to images, right? Because, uh, Steve mentioned three areas. One is images. So we suppose you have historical images of damaged products, and now you want to, so generative AI essentially creates a new, an image of a damaged product. And that image is not exactly the same as the images you currently have, but it resembles them, it’s a new image, right? And how would you benefit? So what’s the business value, right? So you could actually identify that such a pro, such a defect actually can happen of your product. And then you can go out there and examine sort of, uh, how to prevail such potential damages. So the entire area of generative ai, it’s roughly five years old. So it started with something that’s called GaN, so generative adversarial network. And then two, three years ago, a new technique called, uh, diffusion models came Out.
Diego Klabjan (00:10:48):
And now these two techniques are now combined together and they form sort of the basis together with something that’s called transformers, right? So they, so these three concepts now form the basis of, of generative ai. So why generative AI today? So first of all, generative AI morals, they require a lot of data. And with today’s advances in it and data collection and data storage, sort of, we have access to a lot of data. And the big push actually behind generative AI or the differentiating or stepping stone, okay? The, the big, the big step actually is the fact that we can now handle larger morals. And what does it mean larger morals. So it’s literally, you have to think from the brain perspective. So we have a lot of neuros in our brains. And so larger morals literally means morals that have more neurons, right? So the generative AI morals, they, they try to mimic, uh, human brains.
Diego Klabjan (00:11:52):
And so that sort of having, being capable of handling more neurons, sort of, that’s the, that was a big advancement behind generative AI in the last two to three, two to three years. Another term that you commonly hear that that’s related to generative AI is a term of foundational morals. And a financial model essentially is a model that, that is supposed to perform, or it does actually perform just one particular task, but it’s capable with some minor adjustments to do similar tasks in other domains, right? So an example here would be, so you have, you can have a model that generates marketing material, for example, it generates text together with images, right? So it’s a nice kind of marketing brochure or ette, something like that.
Greg White (00:12:41):
Yeah.
Diego Klabjan (00:12:41):
And originally sort of, you use historical data, let’s say in the C P G space, right? So the moral is definitely going to be able to create good marketing material for C C P G type products. But now with this adv recent advancement with a little bit of your own tailored data, for example, on cars, you can teach the moral how to create marketing material based on cars, right? Mm-hmm. So cars being quite different from c P G products, but yet you’re talking about marketing material, right? So the basic marketing material model train on or, or yeah. So trained on C P G data, that’s the foundational model, right? So, and that’s, so that’s generative AI behind, behind that, right? So that’s sort of, that’s my explanation of what is generative ai.
Scott Luton (00:13:33):
Thank you, Diego. I really appreciate that. I love the examples you Use as well. We’re gonna, we’re gonna, uh, dive in a little deeper On, on some of Those examplesAnd aspects here in a minute. But Greg, bringing you in, we heard Steve’s framework and Then Diego Expounded a, A bit More your thoughts, Greg,
Greg White (00:13:50):
you had me at Transformers <laugh>. No, I think one of the things that we have to recognize is that what we are being shown sort of educated on as far as generative AI is just a small example. We’re gonna talk about that, of what it can do, what it means that it can do this generative thing means that it can learn continuously, learn a, a generative adversarial network, essentially argues with itself to determine what is right. It says, Hey, does this look right? And it’s other self says, no, that’s not right. And in a much more complex way, I can’t believe I just said that in front of you two guys <laugh>. But, but it is substantially, that’s what it does until it eliminates all the possibilities of being wrong and goes, this must be right. And that learning capability is a huge advancement. And the fact that it, it can take, these transformers can take the data that it knows and generate data that is speculatively, this is what it could look like or should look like. Mm-hmm. <affirmative>, that’s hugely transformative. Mm-hmm. Yeah. So There is a ton of opportunity In using that, those kind of techniques out there.
Scott Luton (00:14:59):
Yep. Well said Greg And I wanna go back to something Diego said. Hopefully it’s mimicking y’all’s brain, Steve and Diego, and not some of the other <laugh>, not some of the other Yeah. Not our brains out there. Alright.
Steve Pratt (00:15:10):
That was a pretty good explanation.
Scott Luton (00:15:12):
I agree with you. I completely agree.
Greg White (00:15:14):
I feel like I may be doing something on the side a little bit. <laugh>,
Steve Pratt (00:15:16):
that’s Strange. Exactly.
Scott Luton (00:15:18):
I feel like in the last seven minutes I’ve earned a certification. It feels like Steve, Diego, and Greg. Okay, so let’s do this With every big powerful trend, tool, innovation, breakthrough, you name it, you get lots of common myths and misconceptions that, uh, you know, generate from a wide variety of folks that may not quite grasp it the way that y’all do. So Diego and Steve, if you wanna jump in as well, but Diego, any common, and when you think of misconceptions out there about gene AI or AI in general, what’s one or two thoughts that come to your mind? You?
Diego Klabjan (00:15:51):
So one misconception that I wanna point out is more kind of from the human aspect perspective or say economics perspective, right? So a lot of people fear that generative AI is going to replace workforce. I don’t think so. It’s going to augment, uh, workforce, uh, is going to improve our productivity, right? So I like to explain this in terms of, so far we have human in the loop or humans in the loop, right? So interacting sort of frequently with information systems, generative AI actually, uh, takes that one level up, uh, to the human on the loop, right? Which essentially means we still have to monitor a system, but at a higher level with fewer potential interaction. But you still, we humans still have to be there. I also view this from the perspective of that generative ai, it should makes us better, not just from the productivity perspective, but also because it essentially, I view generative AI as a new competitor, right?
Diego Klabjan (00:16:54):
So some something that I have to compete with, something that I have to be better, right? So generative AI from that perspective, at least sort, that’s my personal goal or view, is that I want to be better than generative ai, right? So it, it brings me sort of, uh, additional motivation, right? Mm-hmm. So one and, and one thing another I think misconception about generative AI is people actually have too high of an opinion of generative ai. So when I started using generative AI about a year ago, yeah, I was impressed, right? So I was talking or not talking, alright? So I tried, for example, summarizing some extra long emails, uh, and those kind of things, and I was very impressed by that. But as I progressed, as I started using Genive AI more and more, I figured out that, so for more and more meaning, for more and more complex tasks, I figured out that there’s, there are definitely limitations, right?
Diego Klabjan (00:17:54):
Mm-hmm. So if I go back to my hobby of coding, right? So yeah, Genive AI is gonna produce very good basic code, but as soon as you try to come up with a more sophisticated data science machine learning code, it’s not gonna be able to. And in the last few months, I’m actually finding out that rather than using generative AI to give me a code template, that I have to then spend a lot of time fixing it and understanding and fixing sort of tons of bugs. I’m finding out that it, that it actually takes me less time if I just don’t use generative AI and start from a blank sheet of paper. Mm-hmm. Right? So in short, right? So people are afraid that generative AI is going to replace, uh, workforce, uh, I don’t think that’s gonna happen in any sort of foreseeable future, right? And
Diego Klabjan (00:18:45):
Second, I think people have too high of an opinion about or opinion, so they overvalue capabilities of generative ai. So it is powerful, so don’t get me wrong, it is powerful, but nevertheless, sort of when you’re talking about more complex, detailed stuff, uh, it’s still not quite there. And I kind of doubt it that okay? It’ll be There anytime soon.
Scott Luton (00:19:10):
All right. I appreciate that take. But you also said it’s a competitor, so maybe it’s gonna be motivating you to hit that three and a half hour <laugh> mark in the run coming up soon. Lemme switch gears for a second and then I’m gonna, I’m circle back to Greg and Diego. I wanna switch gears, Steve, and bring you back in when it comes to global supply chain management, right? Mm-hmm. <affirmative>, we love the focus on real results as a, and in this case, transformative use cases that you may be seeing as to how gen AI can make an impact. How would you, what are you seeing that regard and how would you characterize its potential impact? Steve?
Steve Pratt (00:19:44):
I think the combination of predictive AI and generative AI is going to completely change how supply chain professionals run supply chains, right? I think that, I think that generative ai, from a large language model perspective, that first category will have a MI minor, if any impact on supply chains. I th I think the people who are saying, I need to slap a text interface to my a p s is like that, that makes no sense to me. But the third category, right? Of large graph models is a perfect application for supply chains, right? In fact, we can talk about what we’ve done, but if you think of a supply chain is basically a graph, right? You have manufacturing nodes, distribution nodes, you have lanes that connect them, you have constraints, you have historical data that if you model a supply chain as a graph and, and run a large graph model on it, that you get astonishing results, right? So
Steve Pratt (00:20:47):
We’ve been, in fact, I can talk about this later, but we just received a, a patent for our generative probabilistic planning for supply chain optimization. So that was a really big deal for us. The reason that this is gonna transform supply chains is several fold. One is it’ll just give you better numbers. It’ll give you better recommendations on what your demand signal will be. Your supply signal will be your imbalances. The the constraints, the, the your fill rates, your inventory holding costs. And very, very importantly, that fundamental to, to generative ai and even predictive AI is they think probabilistically. It’s all about probability theory. And I think one of the fundamental flaws of existing advanced planning systems is there are what’s called deterministic. They give you one number, right? They give you the, your, your demand is going to be X and you can run different scenarios, but the output is always one number without any sense of it is how confident are you, it’s going to be close to that number, or do you have actually no idea what the answer is gonna be, but you had to give me a number.
Steve Pratt (00:21:59):
So I, I think moving to an AI perspective is moving from this next generation, moving from what we had sequential planning, then concurrent planning. I think that probabilistic planning is the next evolution, right? And that giving supply chain planners and understanding of the risks, you can tailor the risks. Where do you wanna take the risks? Do you wanna take more risk on fill rate? Do you wanna take more risk on inventory obsolescence? So I think those are really, really exciting things where we won’t be spending SS n o P meetings arguing about numbers. We’ll be making strategic decisions about mm-hmm. Would I rather risk a shortage in this customer? Or do I wanna risk my profits, my profitability number, right? Because of inventory outages and, and do that by skew or by product or by region or by time, right? So anyway, so I’m super excited about, we’ve in production with, with predictive AI now for four years at some major C B G brand hundred hundreds of C B G brands. Um, and yeah, we’re, so, I, I can talk more about the stuff we’re doing internally.
Scott Luton (00:23:10):
Well, it’s exciting. And we’re, and, and what’s one, that’s one next places that we’re gonna go is some of those ongoing projects that your team’s involved in. Congrats on the patent. Hey, before I go back to Diego, Greg, I wanna bring you in here. Probab, I know planning is next, is near and dear to your heart, probabilistic planning and how, as Steven put it, not just bigger strategic decisions, but better decisions. Greg, what’d you hear there? And what’s important do you think, for our listeners to take away?
Greg White (00:23:37):
Well, you know, one of the things I heard is this notion of trade-offs, right? Which is essentially what supply chain is all about. It’s you’re trading One risk for another. You’re trading, trading speed for reliability, for ethics, for cost, and various and sundry other things. But those are substantially the pillars that you’re basing it on. And there are, I mean, there have been plenty of deterministic models that assume, I’m gonna use some more statistics, stochastic scenarios, which basically means randomness that we can’t predict Or Manage. We have to consider that it could happen and provision for it as if it will happen. But if we can identify the likelihood of those events or the causes, even better, the causes of those events of disruption, let’s say, of your supply chain that put the speed or the reliability at risk, if we can do that from learning about past disruptions and imparting that to the model, then that changes the amount of inventory that needs to be held in supply chain, which goes directly to the economic Steve was talking about. Yeah. And
Greg White (00:24:42):
There are so many of those trade-offs, and, you know, a lot of what Steve is talking about is new to CPGs or manufacturers. It’s not new to retailers because they’ve had to carry the weight of the supply chain for virtually the entirety of the existence of the supply chain, because CPGs and brands and manufacturers have foisted the risk of supply chain off onto the retailers who carry it in warehouses or distributor distributors, right? Who carry extra inventory and warehouses who carry safety stock, presentation, stock, all these additional stocks. Yeah. And if we can somehow shave that down into a, a level throughout the supply chain, um, like retailers have done for decades, then I think, I think there is a ton of risk and also a ton of costs to be taken out at exactly the same time. Mm. The other thing that this allows us to uncover is these other risks, right? Let’s call ’em existential risks. Am I using a slave trader as a vendor? Mm-hmm. I mean, there are all kinds of things you can learn about your supply chain that are more than just the planning aspect of it. It’s the entirety of the ecosystem.
Steve Pratt (00:25:54):
Hmm. Yeah. I mean, one way we think about this is the, is that this is, you know, it’s basically supply chain professionals are, you can think of them as professional gamblers, right? They make betts, they’re constantly making betts, right? Mm-hmm. Every, every day it’s thousands of betts. And a lot of times the the there are systems do not even give them the probability. They don’t give them the odds, right? So they, they don’t, right? It should say like, Hey, if you’ve got, uh, two kings, you don’t hit Right? Right. Blackjack, right? Like, but it’s like the game is constantly changing and it doesn’t give you probabilities. And I feel like it’s like we all get together as a noodle team is like, it’s like sometimes we feel like it’s like gamblers anonymous. The first thing we have to like admit that we have a, that it’s that our, we have to help our customers understand the odds and where are the risks, AnThen they can play the game. Mm-hmm.
Steve Pratt (00:26:49):
Right? But we Have to inform them on how to Make better Betts.
Scott Luton (00:26:54):
And I love the gambling analogy, and I think one of the things that Comes to my mind As you’re using it is it’s almost like we’re moving it from roulette, which is complete chance to at least blackjack where there’s some sort of a system, right? As you were talking about. Alright, so Diego, I mean, it’s more than that, Scott.
Greg White (00:27:10):
It’s counting Cards at blackjack. There we go. Is what we’re doing? Yeah. <laugh>, I mean, that’s really where we’re going. I wasn’t gonna say that in case legally counting cards, right?
Steve Pratt (00:27:19):
Yeah. Right.
Diego Klabjan (00:27:21):
No, exactly. One, one difference. So Greg mentioned to has, uh, I think you use the term optimization, right? So one big difference between what was in use five, 10 years ago, which was actually to has optimization, was that the subject matter expert in supply chain sort of, and risk management, they created a few scenarios, right? So let’s say three, four scenarios, and then they build the entire analysis around that scenario. So what Moodle is doing today, it’s actually not just two, three scenarios, it’s, we’re talking about millions of different scenarios and then reason on top of those region scenarios and create istic estimates taking into account of all possible interactions in the supply chain.
Scott Luton (00:28:03):
Mm-hmm. <affirmative>, Diego, thank you for adding that. And I wanna give you a chance before I ask Steve about chat, G P t I can’t wait to get, uh, his response and, and all of y’all to, uh, to weigh into. But Diego, anything else you wanna add? You were kind of touching on use cases and, and the impact of gen AI and current impact, potential impact earlier, but anything else you wanna add before I move forward, Diego, in terms of the impact that this is gonna make?
Diego Klabjan (00:28:28):
So in terms of supply chain sort of, and, and this is sort of what Steve already alluded to, generative AI is definitely able to capture all possible, or not all possible, so many, many possible interactions, right? So, and Steve mentioned sort of graphs and networks, right? So that capture interaction, right? So they’re capable of, of more, much more complex interactions and in greater detail than what we were able to do five, 10 years ago, right? So clearly you have to have the knowledge of the underlying generative ai, so scientific knowledge of the underlying generative AI to, to trend or to use that in, in a supply in supply chain networks, right? So, and, and Steve did, Steven did a fantastic job of assembling an excellent group of data scientists and, and together with management to develop these capabilities of using generative AI for supply chain method.
Scott Luton (00:29:19):
Wonderful. Uh, I appreciate you adding that. Alright, so before I move on to any of the casinos that may be listening, the Greg, we’re just using analogies just for Illustrate purposes, right? Greg? They’ve
Greg White (00:29:31):
Already got my picture of <laugh>. Yeah,<laugh> the eye in the sky Is looking for me.
Scott Luton (00:29:36):
Oh, no.
Greg White (00:29:36):
Fortunately I don’t play blackjack, so.
Scott Luton (00:29:38):
<laugh>, right? Yeah. All right. Steve, Diego, Greg, let’s move. Steve, I wanna pick your brain on something and everybody try not to roll their eyes, but I think there’s a lot of listeners that are experimenting with chat G P T, it is, uh, easy to use, uh, you know, kind of a, I think of democratization when I see platforms like that, regardless of everyone’s opinion on the accuracy and all those concerns they have. But Steve, question for you. Is chat G p t good or bad for AI’s perception ministry?
Steve Pratt (00:30:10):
Um, of course the answer is yes, <laugh>, yes. Uh, so it’s the, i I think the, the good part about about chat G B T and large language models in general, whether it’s Bard or Llama or there other, there are other large language models, is that it’s, it’s made it easily accessible to individuals. You know, you can sit at home or sit on your phone and you can play around with it, and you see the power of it, of a neural network generating new text in an astonishing way. I’m still astonished by it. I just wrote, you know, I just told it to like, write a limerick about the use of graph neural nets and reinforcement learning, right? And it wrote this amazing Limerick, which was completely digestible, or you can say a love letter to somebody in the st in the style of Mark Twain, right? And it does it. So I think that’s a hugely positive thing. I I think the negative part of it is, first of all, it’s blotted out the sun. Hmm. Right?
Steve Pratt (00:31:10):
And any other, and so, and, and everything sort of gets sucked into that vortex where really cool other things like mid journey, which is for generating images, which is a large image model, right? Everyone’s saying, well, that’s chat G B T. It’s like, no, it’s completely, it’s completely different chat. G b is it, right now it’s just text. But mid journey, which generates new images is really cool. In fact, for some of the, the screen actors Guild strike that’s going right, right now is about the use of AI in images, large image models, because they were able to like de-age Harrison Ford and Raiders of the Lost Arc through generative large image models, right? And that was, they’re really worried that large image models are gonna like, overcome this. And the large graph models, I mean, like the cool stuff we’re doing, or the, uh, alpha fold, which is an example from Google, they’re using for protein folding, right? Right.
Steve Pratt (00:32:07):
Like, understanding how proteins fold is like one of the fundamental block like blockages of understanding discovery of new drugs and like alpha folds could revolutionize medicine. And it’s, so I, I think, I don’t like the fact that it’s blotting out the chat GPS blotting out the sun. And it especially kills me when they say like, it, large language models are the key to supply chain, right? Because it’s, it they’re not. Right. Right. It, I mean, it, it might help in some edge cases for accessing text data, but I think the other one misconception about chat G P T that people need to understand is that training a large language model is extraordinarily difficult, Which
Steve Pratt (00:32:53):
Congrats to all the people who’ve been able to do that, but you have to have a cutoff date for the information. So the cutoff date mm-hmm. For chat G P T is September, 2021. So if you ask it anything that’s happened in the planet since September, 2021, it has no idea. Right. Because the data that went into it was a cutoff there. So it, it, it kills me when it’s
Scott Luton (00:33:14):
A bit of a problem.
Steve Pratt (00:33:15):
It’s a little no. If you wanna opine about the Revolutionary War, it doesn’t matter. Right?
Scott Luton (00:33:21):
Right.
Steve Pratt (00:33:21):
But if you’re saying like, what’s the recent trend in whatever? Right?
Scott Luton (00:33:26):
Right.
Steve Pratt (00:33:26):
Absolute one idea. And, you know, ask it, ask it who the president is of the United States <laugh>.
Scott Luton (00:33:32):
All right. So Right. For the sake of time, I’ve gotta move forward. We’ve got a lot. Yeah. We got a lot more to, we wanna get into with Steve and Diego and of course Greg. Uh, and you know, Steve, you’ve been referencing kind of in passing some of the things that you and the Noodle AI team have been up to. So tell us a couple things here. First off, how your organization approaches specifically Gen ai and then, you know, talk about some of those ongoing projects that Noodle is leading that leverages that.
Steve Pratt (00:33:57):
Yeah, so right now all of our products are based on predictive ai, right? So we’re using predictive AI for, like I said, we’ve been in production at hundreds of very largest brands that you would all recognize for multiple years. We use predictive AI to find unknown hidden patterns within data, incredibly useful for predicting demand that much, much better than any other approach and supply and imbalances, and converting that to a probabilistic understanding of risk. So generative AI is something that we have finished the data science. So the data science is completely done, it’s now patent patent pending. We’ve run it on multiple customers’ data and it’s getting jaw dropping results, right? So we’re putting in more and more make sure all the real world constraints are in there. Make sure transportation lane capacity, shipment, shipment frequency, every possible constraint you possibly can put in. And so the data science has done, one of the things we’ve learned is noodle ai, is that creating an enterprise software application in ai is that the software engineering around that is as difficult, if not more difficult than the data science itself.
Steve Pratt (00:35:20):
So on generative supply chain planning, we’re done, we’re done with the data science, we know it works. It’s like we revolutionary, but getting it so it’s scalable, reliable, right? It’s, it, the data pipelines won’t go down that it can stay in tune all of that stuff, which is several, several generations ahead of sort of standard software development like DevOps, right? You go to DevOps and the ML ops and then, but having a learning algorithm in the middle of an en of an enterprise application is like really hard to do. I mean, we’ve been, there’s so many edge cases we’ve been working on this for, for three, four years of how do you make it so that it’s fault tolerant and it won’t go down in this mission critical part. So we’re super excited about gen ai. It’s scheduled the, uh, the, or the initial releases not, we’re already at Alpha, we wanna be like ga like a few customers we’ll do in Q one, right?
Steve Pratt (00:36:21):
And then, and then we’re already running in the background at one of our customers and we’ll be two very soon, but we’re super, super excited about it. I think it’s gonna be, it’s, I I I hope that the processes change accordingly, right? Mm-hmm. Like the way we run S N O P and SS n o E in most companies is a lot of it’s to get consensus. So that, okay, we all agreed to these numbers so nobody can complain, right? And, and so I, I think that let’s get away from arguing about the numbers and let’s get, let’s argue about strategic stuff, right? About what do we do? What, what is our intention? Where do we wanna, where do we wanna place our betts? Where do we wanna, where do we want to go strategically, right? In the company?
Scott Luton (00:37:09):
Love that, Steve. And I bet when I hear statements like that really, really truly modernize conversations and where there’s Powerful alignment Around data, we can get stuff done. It usually takes us weeks, if not months, and get it done in a matter of hours, right? Imagine what that will open up. Greg, I’m come to you after we get Diego to comment. Diego, any, anything you wanna add or comment on in terms of, as Steve laid out Noodles approach to Gene ai?
Diego Klabjan (00:37:39):
Sure. So we thought, so industry thinks that we pretty much have nailed down ML ops, uh, right? So, but now this generative AI actually sort of creates a new, uh, challenge when it comes to ops, right? So in other words, when it comes to deploying and maintaining and the actual pipeline in production of generative AI type type models, right? So what Steve just said sort of is, is actually known in, in, in the business community, right? So that no one in the sense that it’s, it’s a new challenge of how do you integrate now large language models or not, like, I’m sorry, say generative AI models in a production setting. And in this sort of, there reports there from consultants, et cetera, stating that there are actually very few companies that today use generative AI in, in deployments, right? So in live, live deployments. So we essentially went from something that, that we think we nearly down so lops now to step up, which is how do we put this generative AI in production? Hmm.
Scott Luton (00:38:46):
Uh, all right. So Greg, when you hear the approach, you hear a Patent pending, You hear it’s Being AppliedTo one client soon To be, uh, two Customers, or one, one customer, soon to be two customers. And it sounds like in a matter of days, If not weeks, another aspects of their Approach. Uh, Greg, what comes to your mind?
Greg White (00:39:02):
Well, a thing that astounds me, not just from what Steve and Diego have said, but just what I’ve seen as the investor is the rapidity with which generative ai, generative AI is generally what we’re talking about, has accelerated how much the technology itself has evolved over the last 20 minutes or so. I’ve had a few thoughts here, but, but I think the thing that’s really very impressive and that we have seen and heard Scott firsthand from a prominent retailer in the US is they took a process that did take 400 people and four months and condensed it to 11 people and a weekend. HmmSo
Greg White (00:39:48):
I, I believe that it will replace people and it will replace them in what we currently consider very, very high value jobs, consulting jobs. The writers had to write the writers that, from their writer’s strike, had to write a, an element into their contract that they can’t be replaced by AI if they weren’t worried about re being replaced by ai. And I think we’ve seen Mark Twain can be replaced by ai. I’m sure some of these TV writers can be replaced by ai. So I think there is so much that it can do that. And we talked about this on a show earlier, Scott, you know, I think we have to recognize that there is so much that it can do, and because it’s constantly learning, it will continue to accelerate its effectiveness. Hmm. And
Greg White (00:40:37):
I don’t think we need to be worried about that though, because it’s going to take jobs that are not what humans are best fit for anyway. Humans will still have to intervene if it writes a, a Mark Twain love letter and make sure that it’s saying the right things and it has the right tone and that sort of thing. But it will get you off the ground. And there are people that do that today. Hmm. Right. So People will move up the chain, just like people move from driving, from driving spikes on, on the railroad, John Henry, right? The old John Henry story. Yeah. People
Greg White (00:41:10):
Moved. We didn’t, we don’t have fewer people working in railroads after that. We have many, many more working in railroads after that. And you know, the jobs will move around because of that. But I think the key thing to understand here is that we can’t even conceive, and those, even those of us, like Steve and Diego and I, not me to the level these guys do, but who are thinking about what AI can and should do and trying to apply it to problems every day. I’m astounded at what AI can do And Every day. I mean, I’ve been asking, and I know Steve, you probably have too, as a founder, I’ve been asking the, well, could it do this question? Mm-hmm. <affirmative> and every day the answer is, let’s see, and a few weeks later, the answer is quote the great Stephen Pratt. Well, of course the answer is yes. <laugh>
Diego Klabjan (00:42:04):
<laugh>. So let me, can I ask the following, Carol, it’s not rhetorical philosophical question, right? So philosophical, it goes along the lines of replacing workforce, right? So suppose you go on Amazon and you try to buy a book And You see a note there next to the book that you are interested in, and the note says completely a hundred percent written by generative ai. Would you change the perception of that book? Would you buy it?
Scott Luton (00:42:31):
Interesting question, Greg.
Greg White (00:42:33):
Yeah, absolutely. I would. Okay. But that doesn’t change the fact that it will still replace
Diego Klabjan (00:42:39):
Well, ’cause see, so my answer is actually I would probably not buy it.
Scott Luton (00:42:43):
Okay, Steve?
Greg White (00:42:44):
Oh, sorry. I’m sorry. Yes. Sorry. Your question was, would I buy it? I would think about it before I bought it. Absolutely. I thought your question was, would I think about it before I bought it? I would absolutely think about it before I would buy it, but we don’t have to debate it, Diego. I mean, chances are good. Yeah. But neither one of us are right.
Diego Klabjan (00:43:01):
<laugh>, my, my point is that at least sort of, I would not buy it. Right? So, and that kind of, yeah, I’m just one person, right? Yeah. So, but this does imply that at least sort of as far as I’m, I mean based on my taste or philosophy, whatever you wanna call it, generative ai, it’s not going to replace rider
Greg White (00:43:19):
It. Okay. Maybe not writers. So, right. <laugh>
Scott Luton (00:43:23):
It is fascinating. We would need a couple more hours, I believe, and really have a full conversation around some of the viewpoints here. But this is what I want.
Greg White (00:43:30):
This could turn into a sports show real quick, <laugh>. That’s
Scott Luton (00:43:33):
Right. With hot takes and all that good stuff. I don’t,
Greg White (00:43:35):
I think Patrick Mahomes is overrated.
Scott Luton (00:43:38):
<laugh>, but you know, it’s interesting when, I’m not sure if it was Steve or Diego, it could have been Greg, about the continuous learning aspect of ai. It is interesting to think about. Yeah. It’s, we humans try to get our, I don’t know about y’all my five hours of sleep at night, six or seven or whatever it is. It’s just working and learning and working and learning. And it’s just, it’s a fascinating thing to think about, right? Certainly one of the, and it never
Greg White (00:44:01):
Forgets. Yes.
Scott Luton (00:44:02):
Correct. And once
Greg White (00:44:02):
It learns something, it always uses it consistently.
Scott Luton (00:44:06):
Exactly. It
Greg White (00:44:06):
Never is misdirected by emotion. Mm-hmm.
Scott Luton (00:44:09):
Right.
Greg White (00:44:09):
Or fatigue or,
Scott Luton (00:44:11):
or your favorite Sports team winning or losing, you know, all that good stuff. All right. So as we start to come down to home stretch, great conversation here, Steve and Diego, I wanna ask y’all a question, and I’m gonna try to, it, it, it’s along, it’s in the vein of what we’ve been asking, but I really wanna challenge y’all to think of something that maybe you haven’t shared here today. So, you know, I don’t know about y’all, but we hear it a lot. We get a lot of feedback from when we do these shows that Greg was referencing the course. Today’s, I still don’t get it. Why Gen ai, right? What’s the big deal, right? And think of those folks that we have plenty of folks doing big things out in the industry that aren’t necessarily technologists. But to that end, if you had to really boil it down to just a couple of thoughts that you hadn’t shared yet, what couple are the key primary advantages of adopting gen AI techniques? And let’s start with you, Diego.
Diego Klabjan (00:45:04):
Well, so <laugh>, I know answer is if I can make money, I would use generative ai, right? So if it can improve my business, clearly there’s a cost side as well, right? So we all know that fine tuning generative AI models is not cheap, right? But just from the business perspective, value. So if I can gain, if I can run my business more efficiently, if I can create a better product, et cetera, I’m all for it. I mean, use, use generative ai. And from that perspective, actually, you’re right. So alignment, safety, those kind of things, okay? We’re not here, we’re talking about strictly business perspective, right? So there’s potentially no issue for alignment, safety, those kind of things, right? So, but as long as I can increase my revenue or, or run away, I’m all for it.
Scott Luton (00:45:51):
Well said Diego. And I think that a lot of folks could agree with that. Steve, I’ll come to you next, but before I do, we’ll see if we can apply Gen AI to maybe some better baseball. Um, empowering. Do you think that’s in the cards, Steve? But anyway, getting aside, Steve, what would be your response to those to say?
Steve Pratt (00:46:07):
How about just actual
Scott Luton (00:46:08):
Intelligence to umpiring <laugh>, right? Yes,
Diego Klabjan (00:46:10):
Yes, yes. We’re gonna The brave, the Braves Don Yes. The Braves don’t need better Empires. Yeah. Bad.
Steve Pratt (00:46:20):
Her, her Hernandez ai. Yeah.
Scott Luton (00:46:23):
Oh, there you go, man. We’re gonna upset the 0.1% of our listening audience that are maybe active umpires. I don’t know, Steve, who cares. But you think about, again, why gen ai? What’s a couple key thoughts that come to your mind beyond what we’ve already talked about?
Steve Pratt (00:46:37):
Yeah. Well, I, I, I’ll just relate it to supply chain professionals. I think that there is a need to give much better information to supply chain professionals. I, I think there’s a lot of frustration and what is really an almost unachievable job, right? We had one customer before we came in, they were getting 580,000 inventory alerts on Monday. The team of hundreds of people would sort through those alerts. They’d try to figure out what’s real, what’s not real. By Wednesday, Thursday, Friday, they need to make thousands and thousands of critical decisions that are gonna infect billions of dollars of inventory. And it just, it’s a thankless job. It’s a frustrating job because there’s so little information. Ultimately it comes down to telephones and texting and Excel spreadsheets and just sort of this heroics at the end. And I think that, I think it can make that job a lot better, right? And more you can get better results. It’ll be more using what human, the human brain is amazing at judgment and thinking. It’s really bad at calculating and estimating probabilities. And, and so let the computer do what the computer does best. Let the human brain do what the human brain does best. And I, so I think that that to me, that’s a really, really exciting use of, you know, both predictive and generative AI is to, is to help the supply chain professional.
Scott Luton (00:48:09):
Steve. Well said. And if you heard Greg there, he said preach it, because there’s a lot of kindred spirits there. And I, I would add one more thing before I come to Greg. You’re talking about Better info. I would also argue they Need more time. They Need more time, they need More successful Practical tools that Can give them, Give more Time and some, and some peace of mind, frankly. Greg, your thoughts when you hear Steve and Diego talk about why gen the ai?
Greg White (00:48:31):
Yeah, well, lemme just start real quickly with what it won’t replace. It won’t replace the ability to make life and death decisions or high, high stakes decisions with very little no or inaccurate data. It will not replace that, it won’t replace any kind of knowledge that is required of the last two years or whatever. I’m sure they’ll shrink that timeline. Right? It also is gonna find it very difficult to, at least for now, to contemplate new concepts, right? For someone to visioner something that’s never happened before that is not based on a foundational premise or foundational data or a foundational information or something that just doesn’t exist today. So that is humans doing human things to Steve and Diego’s points both. What I think it will do is it will free us to do those things rather than those mundane, all those high stakes, but easily manageable with a base of knowledge, with enough data to support a base of knowledge that, that today take an entire week, like what Steve just described, that can be done literally overnight and say, Hey, we did this for you while you were out for the weekend.
Greg White (00:49:47):
Or at least we did these things for you. Here are some things that you need to continue to review. We didn’t have enough data. That sort of thing. The, that is what technology has always done and we’ve just reached such a level, high level of efficiency in terms of so many interactions with technology that this is just another condensing of timeframe around those kinds of challenges.
Scott Luton (00:50:11):
Hmm.
Greg White (00:50:12):
So I think that’s what it will do, is it will continue to allow humans to do less of those things that, I mean, you know, this whole back and forth, God, I was living it while you were saying it, Steve, this whole, Well why did that happen? How did you hurt the company yesterday? We gotta dig into this and figure out why that happened. And then by Thursday we’re back to work. Mm-hmm. <affirmative>, now we can come in and go to work on Monday and know that there are very few things that can’t be handled and we don’t require all this back and forth because all the data that supports the argument that we should do this differently or we should do this the same or this will never happen again. It’s all, it all exists in the models and is presented.
Scott Luton (00:50:56):
Yeah. Love that. Uh, you know, as you’re describing that and each of y’all are talking maybe a good or maybe a poor example, I don’t know, it’ll let humans focus on making, crafting that culinary delight, whatever big fancy recipe and focus on the creation. And maybe AI will take care of setting the table and washing the dishes. Wouldn’t that be something, maybe we’ll be there one day. Alright, so as we start to wrap, one last question before we make sure folks know how to connect with everybody. And if we can, what’s your, in a nutshell, simple straight to the point response here. And that question is one of my favorite questions to ask. Uh, we’ll start with you Steve. How would you suggest our listeners get started with Gen ai?
Steve Pratt (00:51:41):
I actually think playing around with Jet G P T in large language models is a good place to start because it’s accessible, it’s something you can use immediately. I think playing with mid journey for generating new images, you can say, I don’t know why you’d wanna do this, but like draw a picture of a dog juggling on the moon, right? It’ll, it’ll do it, right? So whatever you want, draw a new logo for my company or whatever, right? And large graph models are less accessible to a consumer, like an individual consumer. It’s, I’m not sure that there are ways of playing around with that, but I think, you know, obviously I would encourage people to call Noodle ai. I mean, you could, we could show you what we’re doing, we could show you what’s in production and predictive ai and that’s absolutely rock solid, that generative AI stuff, which we can show you the things that are working in Noodle Labs and that on customer data on a, in a production basis. And so, yeah, I, I would say, you know, just humans are sometimes afraid of things that are new and they, and that are, is kind of scary. And so I think just let’s increase your exposure to it, understand what it is, and you can understand how it can be used for good and the potential misuses of it. And I think that’s what I would say.
Scott Luton (00:52:59):
Excellent. And to your point, school, a generation of technologists were largely created with approachable video games in the eighties and nineties. Kind kind of little bit to your point. Alright, so Diego, same question in a nutshell. How can folks get started? How would you suggest they get started?
Diego Klabjan (00:53:16):
Oh, so if one is interested on the technical side, then, so what Steve said, so G p t,It’s a good start or it’s an excellent Starting point and if somebody wants to Go deeper, so something like link chain and hugging face and those kind of aspects, uh, not aspects, sorry, tools and another way. So there, there’s, there are now also cazillion of websites that offer generative ai. Uh, uh, they, let me just say they showcase generative ai for example, creating presentations, summarizing documents, uh, responding to emails, et cetera, et cetera. But if one is also for managers, I would actually say that the good starting point would be just Google search about potential use cases, hurdles to actually use generative ai. So I mentioned before like pipelines and those kind of things and cost of training so that there’s a lot of information available on the web, but I, I find attending conferences are also extremely valuable, right? So if one from the management perspective is interested in generative ai, so should it be used in my company then conferences and talking to others and hearing about use cases, that’s also, it’s my recommendation.
Scott Luton (00:54:28):
Awesome.
Diego Klabjan (00:54:28):
And where to find me at a marathon on Sunday?
Greg White (00:54:31):
Yes. Three
Diego Klabjan (00:54:32):
And a half hours.
Scott Luton (00:54:33):
We’re writing it down. We’re Gonna check in. Alright, so Greg, before we Make sure Folks can connect with Steve, uh, and Diego, anything you wanna add in terms of how folks get started?
Greg White (00:54:44):
Yeah, I agree with Steve Chat. G P t I mean, it’s a party trick, but it, that’s the way we get people to learn these things is make it simple and consumable and, and I think you’ll find what you can do. You’ll find the limitation and you don’t, it doesn’t know anything past September, 2021. I found that really quick. It’s frustrating, but it is what it is. That’ll probably give all of us a lot of solace to Diego’s point about whether this could replace us. We know it’s not gonna replace us in the next two years. Right. And, and yeah, I mean if you are in business, and particularly if you’ve got, uh, the planning challenge, I can assure you I had a planning company. I’ve seen the competitors out there. There is no one using AI the right way that I have, I’ve spoken to, seen on the marketplace except noodle. Hmm.
Greg White (00:55:33):
Everyone is using it. The way that I see people using it is to select old fashioned forecasting models out of a best fit forecast model. And that’s about as good as it gets if you are doing something not just predictive, but also probabilistic and using gen AI to do it. Or if you need to be doing that, I would give Steve a call. First of all, just talking to him, you’re gonna learn something and I I I know also get some also that there’s no one else out there doing it. One more thing. Yeah. Is can I just, well should I save, I should probably save this for the close.
Scott Luton (00:56:08):
Yeah, let’s save. It. We’ll save it.
Greg White (00:56:10):
Let’s save it for the close.
Scott Luton (00:56:11):
We’ll. Keep an inventory For just a second. Yeah, we’ll fulfill The order and Just say, yeah,
Greg White (00:56:14):
Don’t fill it all now. Got it. <laugh>. Okay. Alright, lemme gimme note, save this later.
Scott Luton (00:56:20):
Steve and Diego. Steve, that is high praise. Trust me. I’ve been doing this for quite some time with Greg and when you get a compliment like that is high praise and genuine praise. So let’s do this. So Diego, first off, how can folks, other than the Chicago Marathon, if folks want reach out and compare notes. You talk about the Cubs and the Braves or all this cool stuff you’re doing and helping others do, and the now generation do it at Northwestern, how can folks connect with you
Diego Klabjan (00:56:47):
The usual, traditional way, sort of email X, what else? Discord. Yeah, that’s, so email is, is the best one. And as I said, X is Okay Twitter, right? So I’m talking about Twitter.
Scott Luton (00:57:00):
Yeah, yeah. You know, when are we gonna be able to drop that and just say X and stop saying formerly Twitter, right? All the, we’re still stuck in that cycle, but we’ll find Jill X and of course we’ll include those things in the show notes, Diego, to make it really easy for folks to connect with you. So, uh, Diego, really appreciate you being here today.
Diego Klabjan (00:57:19):
My pleasure.
Scott Luton (00:57:21):
Alright, Steve, really have enjoyed your perspective here today. You and Diego are quite the one-two punch. Big thanks to you and the Noodle AI team that’s on the move as, as Greg was sharing, really appreciate y’all partnering with us on this. I think it’s gonna be a very informative, demystifying, we’re breaking the mold of some assumptions that I think a lot of folks have made based on what they’re reading or talking about. So really appreciate that. And folks, to our listeners, stay tuned. In the next episode, we’re gonna go deeper with more practical use cases for ministry to really help you connect the technology with the outcomes. So stay tuned. So Steve, how can folks connect with you and the Noodle AI team?
Steve Pratt (00:58:03):
So you can find us on online@www.noodle.ai, so not too complicated. You can reach me@steveatnoodle.ai and LinkedIn, although LinkedIn’s a little overwhelming, right? So it’s <laugh>, right? So my, I think I have tens of thousands of unread messages on LinkedIn, so I, and they make it really hard to delete messages, so it’s, so anyway, I would do, email is probably the, the best way. Or if you’d like to see a demo, you can go on Our website. It’s really cool stuff, so.
Scott Luton (00:58:40):
Excellent. And we’ll put that demo link in the show notes. And by the way, I’m being reminded the email address, Steve, that you’re bold to put out there, Steve, not Steven, steve@noodle.ai is what I’m being told Steve. Yes. Does that work?
Steve Pratt (00:58:54):
S st. S t v E at Noodle ai.
Scott Luton (00:58:57):
Wonderful. All right, good, good, good. Well really have enjoyed it. Uh, big thanks to both of our guests. Greg, I’m getting your key takeaway. We’re got order ready to go, but wanna thank Steve Pratt, founder and c e o with Noodle ai. Steve, thanks for being here today.
Steve Pratt (00:59:11):
Yeah, so absolute pleasure,
Scott Luton (00:59:15):
Abso, I, I feel smarter really after speaking with the three of y’all, but especially I agree, Steve Diego, thank you for
Greg White (00:59:21):
Bringing all this knowledge to us guys. I appreciate it.
Scott Luton (00:59:24):
Absolutely. And Professor Diego Klain with Northwestern University Diego, great, great to have you here today.
Greg White (00:59:31):
Yeah, thank
Scott Luton (00:59:32):
You very much. Appreciate it. Thanks
Greg White (00:59:34):
Diego. Alright,
Scott Luton (00:59:35):
So Greg, a lot more coming. Big thanks to our cloud partners over at Noodle ai. We look forward to continuing the episodes of this limited run featured series making Better Supply Chain Betts With the Power ofProbabilities. So, Greg, For our wrap, What is, if You had To distill it all down Right, distill it All down into a bottle of Steve’s finest from the Vineyard,
Greg White (00:59:58):
never did determine whether he makes wine yet or if he’s just growing grapes.
Scott Luton (01:00:03):
Yes. What, What is, you Had to pick <laugh>, you had To pick one thing, Greg, that folks gotta pay attention to from this conversation with Steve and Diego. What would that be?
Greg White (01:00:12):
Yeah, well, it’s coming. Um, I mean, AI is gonna do a lot of things. I don’t know if anyone, uh, remembers an old Disney movie where people are sitting in big old cushy chairs and robots bring them everything that they want. Don’t be lazy because the more replaceable you are, the more likely you are to be replaced. Mm.
Greg White (01:00:38):
I mean, and it’s gonna start from the commodity jobs on up. So find a niche. I can tell you, I can tell you two people that are not gonna get replaced by AI and that’s Steve and Diego <laugh>. If you ha no, seriously if you have that kind of knowledge or if you, if you master something like this, right? If you really and truly are a, a master of some kind of skill, then you won’t be replaced. If you are an automaton, um, or a, or a production Worker,
Greg White (01:01:10):
It, it’s very likely, uh, just my opinion, Diego and I are gonna have a fist fight after this. But no, I mean, it’s very likely that you will be replaced. So be great if you haven’t read Good to great. Read it the Hedgehog concept, be good, great at something and make your living doing it. So, um, you know, it is all about economics. Both Diego and Steve both said that if you have an economic need that’s not being met, I would say by a technology that exists today, especially as regards planning, then take a look at this thing because there are massive limitations. Me and AI are writing a book together.
Scott Luton (01:01:52):
Mm-hmm. <affirmative>,
Greg White (01:01:53):
Um, I’m just kidding, Diego <laugh>. But I mean, you know, I am, I do have this whole list of supply chain rules and one of them is that the technology that exists today is not sufficient to, is not sufficient to tackle this. I think Noodle is onto something here and I think that they can do things that other technologies can’t. So take a look at it.
Scott Luton (01:02:14):
Mm-hmm. Wonderful. Greg, I appre really appreciate that, Greg. I’m gonna take a look at it. Also, by the way, <laugh>, great place to finish and to our listeners, hey, adding to that keep learning, keep learning. Yeah. Be bold, lean into new places that are new to you. Yeah, it may be scary and all, but that’s part of the secret sauce of this journey We’re all in. So to all our listeners, hopefully we’ve enjoyed this episode as as much as we have. Again, I feel smarter after the last hour or so. Be sure to connect with our speakers and their organizations. Be sure to find supply chain now, wherever you get your podcast. Subscribe to ’em, miss anything. And on behalf of all the team here at Supply Chain now, hey, remember, it’s Steve’s not words. Take something that Steve Diego, or Greg said here today. Put it into action. That’s, uh, the big must need, footsteps gotta be taken. So hey Scott Luden challenging you to do good, to give forward and to be the change. And we’ll see you next time, right back here on Supply Chain now. Thanks everybody.
Intro/Outro (01:03:13):
Thanks for being a part of our supply chain now, community. Check out all of our programming@supplychainnow.com and make sure you subscribe to Supply Chain now, anywhere you listen to podcasts. And follow us on Facebook, LinkedIn, Twitter, and Instagram. See you next time on Supply Chain. Now.
