3548: Logility and the AI Compass for Supply Chain Leaders

[00:00:03] Welcome back to the Tech Talks Daily podcast. I just wanted to give a quick thank you to each and every one of you who have been listening, sharing feedback, reaching out, asking questions. Because over the last year, I've been on the road at more than 20 different tech conferences all around the world. And increasingly so, one of my biggest goals this year is to meet more of you in person rather than just talking to your ear every day.

[00:00:30] So whether it is a coffee, a beer or just a quick hello between keynote sessions, I genuinely love to connect with as many of you as I can. So please check out my travel itinerary for the year over at techtalksnetwork.com. I update it on a regular basis as event invites come through, but there is already quite a few dropping at the moment. And while you're there, there are also 4,000 interviews, eight, nine different podcasts, all with yours truly.

[00:00:58] And if that is not enough for you, if you still want to spend more time on me, you can even work with me as well. There's a number of ways there. But hopefully we can make something work and talk one-on-one rather than me just talking directly to you every day without getting anything back. I generally want those conversations with you, not just the guests.

[00:01:17] And speaking of which, today's guest is someone who brings a rare mix of deep supply chain experience and clear thinking about where AI genuinely helps, where it can easily go wrong and so much more. And his work spans decades of planning, optimisation and applied AI across global organisations.

[00:01:38] And throughout his career, he's lived through multiple technology shifts and now focuses on how AI is helping teams understand decisions, not just blindly automating them. He thinks much bigger than that. So today we're going to talk about why traditional planning keeps breaking under pressure, how to digitise the why behind decisions, and why capturing tribal knowledge might be the most overlooked opportunity in AI today.

[00:02:05] Before I bring today's guest on, a quick thank you to my friends over at Denodo, who are passionate about logical data management for AI success. Because let's be honest, AI is evolving fast. But the elephant in the room is initiatives are still failing. Not because the models aren't good, but because the data foundation isn't ready. That's why organisations are increasingly turning to Denodo and logical data management.

[00:02:34] Visit Denodo.com and put logical data management to work today. So a massive warm welcome to the show. Can you tell everyone listening a little about who you are and what you do? Hi, very nice to be here, of course. My name is Pete Buck. I'm from Antwerp, Belgium. And I lived abroad a lot of my career during the .com. I lived in the US. I also worked in the UK.

[00:03:02] But I always worked in tech and supply chain in the areas of innovation. And it used to be the optimizers in the 90s. So that dates me a little bit. Around 2000, the .com and the cloud. And now since 2005, I've been working in AI. I'm a serial entrepreneur who combines and switch between corporate work and my own startups, of which the most recent was a fully AI-driven planning environment called Garvis, which was acquired by Logility.

[00:03:31] And Logility was acquired by Aptia. And now I'm together with Logility and Aptia trying to think how the future of planning could look like. And hence my book, The Compass for Supply Chain Leaders. Yeah, that's one of the reasons I invite you to join me today. Because in your new book, which is called AI Compass for Supply Chain Leaders, you describe planning as something that drifted into being overly numeric and disconnected from real-world decisions.

[00:04:01] So I'm curious, what did you see breaking down in traditional planning long before AI entered the conversation? I think it really properly entered about three years ago, even though it's much older than that, of course. But what did you see that was breaking down here? Yeah, it's an interesting question. Of course, every type of planning is defined by the area it has been invented. And effective planning in an organization means that every part of the organization

[00:04:30] is aligned to the same goal. However, until recent, expressing those goals was only possible in numbers, just technically. You could not gather all that information and effectively pass it on. So a 100 number in sales could mean an expectation of the growth of the current portfolio of 10 products. Or it could be that there was an expectation to get to the 100 if you created five new products.

[00:04:55] And so this needs constantly, and still needs, constant meetings to explain. And before 2000, a lot of decision-making was, when you think about Europe, but also the US, was decentralized. So local alignment was a lot easier for understanding the other divisions of the organizations. But if you think about the globalization of the global rollout of finance system, ERP and international consolidations, that decision-making,

[00:05:24] now in the form of sales and operation planning, became very, very complex and further and further away of where the real action was happening. And that was fine in 2010, between 2010 and 2020, when the world was evolving quite slowly and growing. But thanking COVID, boom. And then there were tariffs. And then there is disruption everywhere.

[00:05:48] So now, when something happens in a country, then before you get the information centrally, and the reason why, I have an example of a large life science organization, when there was a disruption, like in a country, like, for instance, Ireland, there was a contamination of a competitor. It takes 90 spreadsheets and almost six weeks, you know, to get that information centrally and be part of a decision or a reaction.

[00:06:15] So that's the problem we're looking at, the decentralization of this, and not knowing why anymore, decentralized. And you're someone that spent years working with organizations under real commercial pressure. And although there's a lot of talk around AI at the moment, I've been to like 25 different tech conferences around the world in the last 12 months, one reoccurring theme I keep hearing is, without good data, there is no AI.

[00:06:43] So I'm curious, when leaders come to you and say, hey, my data is already good enough, where do you usually see the blind spots once that volatility hits? Because it's more important than ever now, isn't it? It is. It is. And it's basically, it almost sounds like not a complex question, but it basically touches on everything. It's interesting, because as you will see in different questions,

[00:07:12] there are a lot of facets to AI. It's not just one AI. You need to capture data. And they're having these data in the ERP systems are basic. So they are like the basis on which you can start using the AI. But then it becomes like a sliding scale, where you say, okay, the more data I have, the better value I can create. But the AI can do a lot more than just solve based on data.

[00:07:42] It can also pass data in organizations. So that's an aspect that is less understood about the AI. It can help be part of the processes to bring information that usually resides in one part of the organization to the other. It can be very simple, like helping me in my decision understand what data are in another part of organization,

[00:08:04] type of BI, but it can also combine information from your team, or it can be more systemic. And so you have to see where in the journey you are. And of course, ideally, when you plan what you thought strategically or what you thought tactically, how that relates to the reality, what's happening right now.

[00:08:33] So the more you can create this decision canvas against which you can actually look at what's happening right now and compare it, am I off, am I not off? And how does it impact my decision making? The more you can create that in a digital format, the better decisions you can make and the faster you can. And one thing I think we all took for granted over five years ago was supply chains.

[00:08:58] But since COVID, they keep reappearing in the headlines from pandemic disruption originally to tariffs and geopolitical tension continuously rising. From your perspective, what does AI genuinely or where does AI genuinely help leaders anticipate disruption better? And where can it create a false sense of confidence if it's misunderstood? It feels like quite a delicate balance there. Yeah, yeah.

[00:09:26] And it's again, there is more than one AI. So yes. Yeah. And the focus in a lot of thinking and that's maybe where I'll start with. There's a lot of different AI for different situation. And the false sense of confidence comes mostly from trying to automate things when there is not enough internal automation, understanding of these things. So trying to automate.

[00:09:50] So my five penny first advice is never use AI automation on what you do not understand yourself. So that's the basic rule. Don't give it away what you don't understand. And then what you need to try to do, there are like three points where you have to think of. There is the understanding. It's like their models can help language levels like ChatGPT.

[00:10:19] There are solves that bring you automation. They're more opaque. You know, they're more black box. But they can help you schedule. They can help you network optimization. And then there are agents. Agents are like the most recent and the biggest evolutions. They are specific pieces of intelligence that can do things, you know, and they can understand. They're almost like co-workers. So for each situation, you have to try to combine all these elements together in function of,

[00:10:49] do I need to make an as accurate possible answer as possible? Do I need an answer that I can explain? So the explainability. Or do I need an answer that's fair? Think about, yeah, I will schedule without problem, you know, production shift, midnight, Christmas, you know. But that's not fair, you know, because you don't want people to happen.

[00:11:13] Or in allocation to customers, you know, it's not very good in understanding these fine elements. Oh, I know this customer and I've promised it and it's the fifth time. So that combination you have to make. And that's an element you need to bring into your assessment for which decisions do I need what? What I don't understand, don't automate, supervise. If it's easy and it's just gathering information, then please automate, you know, make it.

[00:11:41] If it's something that has an impact that's sensitive, then please supervise, you know, because it needs a fairness of it. That's very important. So, and that misalignment, of course, you know, creates the issues about understanding AI. And it's a new world, you know, so it's normal that people need to learn about it.

[00:12:04] And you also make a very strong case for explainable and collaborative AI rather than black box automation. So how does generative AI change conversations between planners, finance teams, and operation leaders on a day-to-day basis? What do you see here? How is it changing those conversations?

[00:12:54] Yes. What do you see here?

[00:13:25] into the knowledge of all different departments. So if sales makes a decision, then the planning department now can understand what sales has done. If there was a promotion, an increase in the forecast, and this was based by a promotion, maybe it's accessible. It's understandable that this is acceptable. You don't need to have a meeting anymore to do it. But the same for a credit check, for instance, that if sales can prove, for instance,

[00:13:54] that this credit check is done and done, but with the help of AI or an inventory check, and everybody else can see this. So you don't need to come together or you don't need to disrupt those decisions anymore. And the simple fact that you now have support in finding out support on the decisions you can, will revolutionize, and bring the control to the control that can make those decisions a lot better.

[00:14:21] So we will go away from those silos because the weak points in that decision-making will be supported by AI. And the secondary effect also, because we have these language models, is that we will be able to harvest the knowledge of the manufacturers, the people who really work in the day-to-day with products. So think about suppliers that deliver bad quality.

[00:14:47] Before it gets to the top, the people who are really working with the materials will be able to keep that information using language models. So this access of information will become possible for everybody. Quick thank you to the sponsor supporting all of the shows on the Tech Talks network. Their support helps me bring you over 60 interviews each month with leaders building global teams. And this month, I've partnered with Alcor.

[00:15:16] And if expanding engineering operations beyond your home market can be overwhelming, you're not alone. Because if you've ever wrestled with local laws, slow response times, and partners who treat each country as separate rather than part of a wider strategy, you might want to check out Alcor. They approach expansion completely different. They specialize in building tech teams across Eastern Europe and Latin America, and they combine employer of record services with recruiting.

[00:15:46] So you get one singular coordinated process. They help you choose the right jurisdiction based on your needs, run proper evaluation of candidates, and onboard teams quickly. And their model is also refreshingly transparent. Most of your contribution will go straight to your engineers, and their fee shrinks as your team grows. And there is no cost to exit if you move the team in-house at a later date.

[00:16:13] And I think that kind of clarity is why so many high-growth companies in Silicon Valley are working with them right now. So you can find out more details at alcor.com slash podcast, or simply use the link in the show notes. And I do think there's still a lot of anxiety that AI will de-skill planning roles or replace experienced professionals. But based on what you're seeing in the field, how are the best teams redefining human judgment rather than losing it?

[00:16:43] Because I think it is exaggerated sometimes that how AI will replace everyone. I think critical thinking, human judgment and experience, et cetera, that will always be paramount and working alongside AI. But that's just me thinking out loud here. What are you actually seeing? No, it touches to the human element. In the end, you know, AI should support the humans, and not the other way around.

[00:17:11] We're not there to automate everything. But there will be change. Please, I don't. But oversimplifying, it's something like it's a real revolution. It's like moving from horses to cars. It's not the same, of course. But, of course, that disrupted the whole industry of horses, stables. But you will need people to manage the highways, the gas stations, and the cars, too. And so the same for AI. You know, it will reshift.

[00:17:41] The human brain works automatically with acquired skills, but it still needs a lot of energy to make difficult decisions. So what AI will help, you know, is to give more time and more information for people to make those better decisions and to create that environment.

[00:18:07] And in the current environment, there is very little time for that because a lot of energy is putting in gathering that information, manually preparing things, and making basically rather simple decisions because so much time goes into the process.

[00:18:25] And with the future of AI, we will have this more deeper and more value-added decision instead of pure automation. And we will try to focus now on not so much on executing what we planned, but what are the depths, I would say that the depths that we have, you know, to be the best we can be.

[00:18:54] And this will be a big win, you know, for everybody. Of course, the problems, and there are still problems to be solved, like mostly these automated tasks would be things which you would train juniors or like less complex jobs. And so we will have to find a way to engage this in another way than before more in executional tasks, more in customer relations tasks, I think.

[00:19:25] Yeah, it is a concern that a lot of the younger people coming out of education, those entry-level roles that they would be taking might no longer be there, and how we better bridge that gap is incredibly important, isn't it? And, of course, many organizations talk about being AI first but struggle to move beyond pilots. We saw a lot of samples or examples of this throughout 2025, a lack of ROI and a lack of progress.

[00:19:54] But what is the most common leadership mistake that you see when companies try to introduce AI into supply chain planning? You've probably seen quite a few mistakes and missteps, but anything stand out there? Any common mistakes? Yeah, absolutely. Absolutely. Absolutely. It's like the polarization. It's either too small or it's too big. If it's too small, it doesn't bring enough value. It doesn't move the needle. Management is not interested.

[00:20:23] And they see it as a side project without importance. And there's no real ownership. So that's one. So then the other thing is like it's too big, you know? Okay. Let's put everything in there. Then it's too big of transformation. It takes too long or too many hurdles in the middle. So the most important thing is to create value and is to where does AI really bring work?

[00:20:52] And we go back to what I said before is that before everything was in silos. So go where the silos meet or where you need to do control on each other in the sense that sales inventory, orders, financial control, think about examples like give sales insight. Do I have the inventory? What are the lead times? What is the added value of the dollars?

[00:21:18] Now things you can all bring together with AI and make somebody responsible for the value across the different silos. So not because otherwise you get this fight. And of course, like you said, people feel uneasy about technology running around in areas where they put a lot of effort to learn it.

[00:21:46] And so you need some leadership in order to make sure people don't rate the systems or try to sabotage the elements, but that they are partners, become partners in that selection. So that's important. Yeah. 100% with you. And in your book, you also emphasize keeping humans in control while letting AI handle some of the complexity.

[00:22:10] So I'm curious, what does that balance actually look like in practice when decisions need to be made quickly and the data is incomplete and maybe they're not mature enough for that yet? What are you seeing happening in the workplace? I've seen that many keynotes over the last 12 months on that utopia that we're heading towards. But what's the reality like? Yeah. Yeah. So the reality is, of course, in the beginning of this, because the biggest problem is the process change. It's not so much the AI.

[00:22:36] The AI is usually going very fast. It sits in a kind of an exponential model in the sense that it doesn't move too much, but it will explode. And that's something I can say because it brings value and it's real. And once it works, it will work for everybody. That's another thing. It's digital.

[00:23:03] And so people are now making, okay, I think EEP can have a competitive advantage if I have AI, but that will erode. You know, it's like cars and horses. It's like everybody will have cars, you know, which will be. So that will go away. So there are experiments everywhere. And there's something a little bit from what we talked in our last question, say, in between or too small or too big.

[00:23:25] But the most progress is happening in the planning environments where sales and operation planning are now connected to different horizon. More data are brought in. People try to synchronize these elements across like order consumption, like order to cache processes where they bring in.

[00:23:50] So there is a lot of progress being made. But what I think we're waiting for is like an organization that really says, okay, I go for a big change and I look at my processes more fundamentally and improve it works. And then I think everybody will just follow very, very fast. But I just want to say something about the incomplete data.

[00:24:16] And the most important thing, I think, in AI is that is the sliding scale. We start from supportive to really being decision making. But the biggest evolution we will see, I think, is in the capability to capture tribal knowledge. So not the traditional data sets, but how have you done this in the past, you know?

[00:24:46] And what are the guidelines? And that's basically very, I just advise everybody, you know, to try this out with a new generation of agents. You can, what you call, prompt them. You know, you can say to the agents, you behave like this, you know? You don't just accept information on the first hand. You make it funnier, you make it, all things are possible, but you check twice and you remain within these rules. And these rules are the tribal knowledge.

[00:25:15] And that I think is going to change a lot to make these corner cases workable, which are the exceptions that would otherwise kill the project. And if we look back over the last five years, our world has changed so much in that short amount of time. And we've gone from what remote working at scale, hybrid working, AI everywhere. And I think predicting the future is incredibly hard.

[00:25:41] But if I did ask you to look into my virtual crystal ball and look another five years ahead, how do you think AI will inevitably change what strong leadership looks like in supply chain, especially for executives who maybe grew up in those spreadsheet-driven planning cultures? Any big changes you see happening over the next year? Yeah, yeah, yeah. So, of course, you know, like now, of course, strong leadership will remain, you know, points. But strong will define differently.

[00:26:10] Now it has a little bit of a, I would say, watch out for what I say, but more like a narcissistic element to it. I take a decision. I know, I know what's good. The new world will be a lot more data-driven because the obvious strong man's wrong approach to things will be obvious, you know, for the whole organization.

[00:26:35] So you will need to be able to understand, you know, the data sets, how they compile, and what does it mean for the organization, and how do I play the tactical game. So it will be a more sophisticated game, less pure leadership. And I say, and it's like that, you know, it will be like everybody will have access to the data, and everybody will understand.

[00:27:04] It will be more like how clever can I be, you know, that I understand that I can explain the direction we're going. That's what I think. And, of course, it will benefit to the whole world, because there will be less waste, there will be less stupid decisions, you know. Well, I've loved chatting with you today about your book and how you see things evolving, and also your experience and insights.

[00:27:33] And I'm curious, though, you're the guy that's wrote a book, you're an author, but what do you read? I always ask my guests, is that a book that you'd recommend? Well, add it to an Amazon wishlist that maybe they check out. And obviously, I'm going to add your book, but is that a book that you'd recommend, that you've read, that you enjoy? What would that be and why? Yeah, so if you want to think about the AI, you know, then think about the coming wave from Mustafa Suleiman. That's a real comprehensive document.

[00:28:00] The interaction between man and technology is about thinking fast, thinking slow from Khan. I don't know his first name, but it's really like, it's a hard book to read. So maybe you can, it's a summary, but even a couple of, it gives you, you know, that although we think we are very rational, we are not, you know, we're very, very influenceable.

[00:28:27] And we have a lot of biases as human beings. So understand those, you know, can help you to position also AI. I think in general, if you have a goal, then everything you see on your road becomes something that can support you. So if you can read about AI, not try to understand everything, but in function of what you want to do with it, you know,

[00:28:49] then it helps a lot because then you see a lot of nudgets either in newspapers or in weekly magazines or there's a lot of people are writing about it. But think, okay, what can it do for me, you know, that I think is helping. And of course, because I'm very specific on how I could change to planning, it helps me to sift out and try it out. You know, there is a lot of AI available. Do it, JGPT.

[00:29:16] We ask it stupid things, you know, to do for you. It's amazing what it can do. It's amazing. Well, I'll add a link to the book so people can take a look at that. And today we've covered so much in 30 minutes, talking about your hugely successful career, the experience and insights you've gained along the way, your company being acquired by Legility, author of a book. But I'm going to take you right back to the beginning as we come full circle.

[00:29:43] Now, if you look back on your career, was there somebody that saw something in you, invested a little time, someone you're grateful towards? Because very often we all got somebody. But who would it be for you? Yeah. For me, I would like to give this tribute to Robert Byrne. He was the guy who started with the first generation of demand sensing in AI. But Robert had a really deep understanding, you know, and you need to go to the bottom sometimes.

[00:30:12] Somebody needs to have patience to take you to the bottom. And then you can say, okay, you know, now I learned it and now I can start putting elements together. Because the worst that can happen is that you think something is possible and you don't have the basic elements, you know, to really understood or thought it through, you know. So that what you say, it's not so bad because you lose, you look bad.

[00:30:39] It's more about all the effort you've done based on false assumptions. So I'm really grateful that he took me really to the bottom of things and, yeah, and helped me understand supply chain in depth, you know. Love it. And a quick shout out to them there. I think it's so important to recognize the people that we're all standing on the shoulders of. It's so important. He's probably blissfully unaware on the scale of the impact and gratitude that you have towards him too.

[00:31:07] So thank you so much for sharing your story today. One last question before you go. So anyone interested in your work, your book, where would you like to point everyone listening? Yeah. So, of course, my book is available on Amazon, the AI Compass for Supply Chain Leaders. I do also a masterclass that it's free. So I basically every two months next year, I'll talk about, okay, what I think is new.

[00:31:35] And it's like, it's a kindergarten, you know, there's so much, or a candy shop, what do you call it? You know, there's so much new things happening every week that the series can go on, I think, for another year. And then, of course, you know, I post on LinkedIn from time to time, you know, happy to connect to anybody who, the ones. Yeah. Brilliant. Well, I'll add links to everything you mentioned there. Make it easy for people to stay in touch. Maybe join one of those masterclasses as well.

[00:32:04] I think that's a great opportunity. But more than anything, just thank you for sharing your story today. Really inspiring. Thank you. Okay. Thank you for having me. It was a pleasure. Huge thanks to my guest for such a thoughtful and grounded conversation. I really appreciated his perspective on using AI to support human judgment rather than replacing it. And it's his honesty about why so many AI initiatives struggle once they hit real operational complexity that I think really hit home.

[00:32:33] So I'll add links to the show notes so you can learn more about him, his work, his book, and dig a little bit deeper into the ideas we discussed today. And as always, I'd love to hear your thoughts. And finally, I also want to come back to something that I mentioned at the start of this episode. I'm spending a lot of time on the road again this year, speaking at events, attending conferences across regions, recording podcasts, taking the show on the road. So I'd really love to meet more listeners face to face. That's my big resolution of the year.

[00:33:01] So if you're attending any of the same events or if you fancy grabbing a coffee or maybe even a beer while I'm nearby, take a look at my itinerary over at techtalksnetwork.com. There are so many different ways we can work together too. Let's see what we can line up in 2026. But that's it for today. So let me know your thoughts, especially around how you see AI genuinely helping decision making in your organisation and where do you still think it needs a firmer human hand.

[00:33:31] And I'm going to leave you thinking about that one before I return back tomorrow. Speak with you then. Bye for now.