Executive Summary:
If your supply chain planning still runs on monthly
forecast reviews and flat-file data dumps, you're
already behind - and this podcast explains
exactly why. Supply Chain Planning Reimagined
is a special series from The Enterprise Edge®
in partnership with SAP, and episode 1 -
The Vision Shift: Reimagining Supply Chain Planning - brings together Xavier Duprat, VP of Supply Chain Planning at Cascades (a packaging and hygiene giant running SAP IBP since 2016), and Joel Beal, CEO and Co-Founder of Alloy.ai, whose platform ingests real-time POS and inventory signals from over 450 retailers. Together, they dismantle the myth that better supply and demand forecasting algorithms are the answer - the real boon, they argue, is better data, delivered daily. You'll hear how Cascades overhauled its entire sales and operations planning (S&OP) process across 25 North American plants and
grew its business by 10%,
while simultaneously improving retail in-stock
rates, how AI agents are now catching
distribution glitches in retailer systems that
human teams would have missed entirely,
and why the best supply chain professionals
of the next decade won't be replaced by AI - they'll be the ones who learn to manage it. This isn't theoretical. It's a real supply chain planning transformation, already underway, with receipts. Stream it now, and be sure to LIKE, SHARE, and COMMENT! Trying to modernize your Supply Chain Planning? >> https://events.sap.com/tee-supply-chain-planning-reimagined/en_us/home.html
Transcript:
Mark Vigoroso (00:02.605)
Greetings all. Welcome to a special edition of the Enterprise Edge podcast. This is actually a special series that we're doing in partnership with SAP entitled Supply Chain Planning Reimagined. This is a two-part series and we will be bringing together perspectives from solution providers and end customers to unpack how supply chain leaders are envisioning the future of supply chain planning. Moving
from monthly forecast reviews to more continuous AI enabled planning that keeps pace with what's going on in the market. We'll also learn from real examples of AI agents that are making partially autonomous supply chain decisions. The governance frameworks that enable the trust and the path to enterprise scale deployment. So this is episode one. This is our opening episode and it's called The Vision Shift, Reimagining Supply Chain Planning.
I am pleased to be joined by two distinguished guests today. First, have Xavier Duprat. He is the Vice President of Supply Chain Planning and Fulfillment at Cascades. Founded in 1964, Cascades employs over about 10,000 people and offers sustainable, innovative, and value-added solutions for packaging, hygiene, and recovery needs. The company has run SAP Integrated Business Planning, or IBP, since 2016.
deploying S&OP or sales and operations planning across multiple business units and navigating dramatic demand shifts, including what happened during COVID-19. We are also pleased to welcome Joel Beal, CEO and co-founder of SAP partner Alloy AI, which is a purpose-built retail analytics platform that ingests real-time POS and channel inventory data from over 450 retailers and feeds it directly into SAP IBP.
enabling consumer brands to sense demand shifts and improve forecast accuracy by up to 35%. So welcome gentlemen, welcome Joel, welcome Xavier. Thanks for being here.
Joel Beal (02:09.038)
Thanks, Mark. Excited to be here.
Xavier Duprat (02:11.083)
Yeah, good day likewise. Very excited to be along you guys.
Mark Vigoroso (02:15.395)
Great, great. We are glad to have you. We're gonna get into a lively discussion. But then again, as it always goes on these podcasts, folks, if you've got comments, if you've got questions, if you've got similar experiences, please drop a comment in the comment field. Keep the dialogue going, keep the learning going. We will circle back with you, connect you with the right people and hopefully help move your business forward.
certainly in the area of supply chain planning, but really anywhere. All right, so let's talk a little bit guys. Let's get into the first tier of our conversation here around re-imagining supply chain planning. So Gartner data and other sources that I've seen shows that some of the best companies are still only hitting around 70 % forecast accuracy. And if you look at that number sort of longitudinally,
It hasn't changed a whole lot over recent years, at least en masse as a macro indicator, right? But the both of you, and we'll talk about the relationship that you all have and how you've partnered, but both of your organizations sort of made a bet that the problem really wasn't in how you forecast, but it's really the data that you feed into the forecast. And as I stated in the introduction, Alloy AI,
built a platform around real time point of sale and channel inventory signals from all these hundreds of retailers. Cascades overhauled its entire S&OP process across I think 25 some odd North American plants with SAP, IBP. And I'm doing a very broad brush summary on what kind of what has happened. But Xavier, maybe we'll start with you. What was the, you might call it a breaking point or the realization point where
where it became undeniable from your role and your remit that the old model of planning and forecasting wasn't just sort of underperforming, but it was structurally inadequate, out of date. Choose the word, but it just wasn't sufficient anymore. Was there a moment or was there a realization that you had and what led to that?
Xavier Duprat (04:37.165)
Well, the first of all, you're right, Mark, the transformation, you know, the S&OP just be the importance of S&OP as part of the, let's say the process of planning is the foundation, right? So for the past 15 years, we've been really working on S&OP process. And at that time, we were working with also legacy system. obviously,
SAP, IBP came on board and gave us the, let's say the platform to effectively manage S&OP and eventually link it with the SNOE and execution. But the most important signal is the demand. it's, let's say customer demand was always the fundamental piece. But over the years and recently with the partnership that we did with Jewel,
The ultimate goal is to extend and have the consumer demand, is for us, which has been the game changer for the past two years. And traditionally, you get that data from different system. You get it, sometimes it's flat files, sometimes it's just having the... And we started that whole collaboration meetings also with partners, retailers, distributors, but really the game changers to have that data on a daily basis again and to be synchronized.
with our partners. I guess, Joel, you're going to talk about it. But Alloy is mostly focused on retailers because retailers share their data. And it's not just daily point of sale data. It's the retailer sales plan, demand plan, supply plan. And then that intersects with our own models. And then we can have the really good discussion on forecasting.
Core term forecasting, yes, but most importantly also is the, let's say, mid long term forecast and retailers in general have that. And the fact that it's embedded in the systems, it brings a lot more value. It's quicker. Obviously there's one set of numbers. So we have the discussion, let's say much more efficiently than we were in the past and we're able to integrate, let's say the retailer forecast as part of our statistical forecast model. So
Xavier Duprat (07:02.647)
You know, from a foundation perspective, that's really the game changer for us is really to have the right level, right conversation with our partners, at least on the retail side.
Mark Vigoroso (07:14.551)
That's great. That's great. And speaking of which, Joel, maybe you could talk a little bit about the partnership that you have with Cascades and maybe what did you see with Cascades? What made them an ideal candidate to benefit from what you all have built over the last, I think, 10 years, right? I think you guys have been in operation for a decade. Is that right?
Joel Beal (07:35.107)
Yeah.
It's just around a decade now, hard to believe. So I think you touched on a couple things in your question of like, aren't we getting more accurate at forecasting? And I think there's been a lot of questions about this. when you look at what goes into a forecast, I mean, there's the methodology, which actually has changed quite a bit. I mean, a lot of it's been developed for decades. I mean, a lot of the core statistics, but.
Mark Vigoroso (07:40.738)
Yep. Yep.
Mark Vigoroso (07:49.751)
Yeah, yep.
Joel Beal (08:06.156)
Machine learning's definitely made some pretty significant gains, but we really are going at the problem saying those are kind of commodities. mean, everybody can get access to these models. They're pretty easy to apply. The real differentiator is what data do you have to put into it? And I think it's both the breadth of the data and the quality of the data. So I think when Xavier is saying, I'm sure they've been working with retail data far below.
far before we showed up on the scenes, you're trying to incorporate that, you can purchase syndicated data, retailers are sharing things. As he said, it's just coming in all sorts of different formats. And so it becomes, at that point, a cleansing problem. How do I normalize it? How do I incorporate this into my processes? Where I think a lot of this tends to break down and where probably why we're not seeing the improvements is just.
Joel Beal (09:01.28)
It's cumbersome, it's hard to do, it takes a lot of time. Time really matters here because if you can't get it done in a certain period, you're gonna say, yeah, I gotta do it monthly, because I can't do it weekly. I don't have the manpower and that ability. So I think that's, again, our philosophy is very much, we're always trying to push the models too, but it's how do we give you the cleanest, fastest kind of data set?
you know, as was mentioned in our world, this is what consumers are actually purchasing. So it's the best signal to tell you what that retailer is going to going to order. But getting to this is still very much a a people and a process and a change management. What's so amazing about Xavier and his organization is how they've embraced that. And I think you see some organizations that say, yeah, and
theory, everybody agrees in theory, this is a good idea. I've yet to meet a company that says, yeah, we shouldn't incorporate real time consumption demand. But there's a lot that say, it sounds like a lot of work. And I don't want to change processes that I've been running for decades. And you know, it works well enough. I actually found COVID broke a lot of those assumptions. But before then, we I was in so many conversations, people would be like, I've been doing this for 50 years, like, you know, or my company has it's kind of worked.
Mark Vigoroso (10:15.427)
Right. Right.
Joel Beal (10:29.024)
And so I think that's where you see the differences, know, folks like Xavier who say, we're gonna try something different. We're gonna keep pushing here. that's, you know, that's what we've seen in that organization and for companies that we've seen make the big steps.
Xavier Duprat (10:46.285)
And Joel, what you say, the IBP integration also is extremely powerful because you can, when we start first, start working together three years ago, it's just a fairly new relationship, right? So we started with just POS data standalone on the side, and we saw the benefits really quick on, let's say the fulfillment side.
But the ERP integration, which we did last year, is really, again, a game changer for forecasting because you're directly connected with all of your partners. So for example, I can give you a concrete example from this week is we saw right away the shift from one hour retailer, their forecast jumped by 40 % in July. And we saw it right away. In the past, would have taken probably
at least a couple more weeks, maybe we would learn that the supply plan change at the last minute because we get a, let's say a surplus of orders, but we were able to flag it automatically just because it's integrated with our system, with IBP. And we always cross match forecast from the retailer, our forecast, and also there's an alloy forecast. So we have three different feeds of forecast in one place.
And right away we can detect if there's an anomaly of something and not Larry that pops up. And in that case, we were able to right away identify it, get back to the retailer, ask question, why is this happening? Why do we see that? And in two months from now, are you doing some kind of activity promotional, et cetera? So, but my point is that an integration really makes you more proactive as a, as a partner. for me, as a partner with, with the retailers, that's, that's value right there. And there's tons of example.
This one on forecasting specific, I think is really interesting. But for those who are contemplating or working with alloy, think really the integration of the system drives the value also, because you can always do it standalone. the connection to have the data connected and to invisibility, that's super important for us. And we always keep pushing and deploying.
Xavier Duprat (13:06.473)
new development with the alloy team. We're always pushing on new ideas. How we can improve forecasts and stuff like that. it's really a strong partnership that we have together.
Mark Vigoroso (13:23.297)
No, it's great. It's inspiring guys. And it's, you know, it kind of leads into the next question of sort of joint innovation and problem solving and, you know, moving the ball forward. And I talk a lot about speed these days, right, with a lot of the companies I work with and how fast things seem to be evolving, certainly in relation to the AI era, but it just seems that the pace of
business and the requirements and expectations of customers and the the agility of competitors it seems as though the pace of business has picked up in a way that that seems to be unprecedented on the surface in many cases and so in that spirit you know there's there's this concept I'm not going to get too academic but there's this concept called complexity theory where there's
there's this idea of requisite variety. And that basically means that your control system has to be at least as complex as the system that it's managing. And I want to get much more detailed than that. bringing it into the S&OP space, traditional methodologies were you'd have monthly S&OP cycles where planners would have like 12 decision points, basically once a month decision point every year.
you know, effectively synchronizing supply and operations. you know, alloy is feeding and that all important data set of consumer purchase behavior into Xavier's, your planning processes and models. And you've had to, like you describe, you're synchronizing demand and supply constantly across
you know, your container board and tissue business and specialty packaging almost simultaneously. And I guess this is a long way around of asking in this sort of hyper velocity environment, how have you closed this gap between what you might say is more of a batch process where it's like once a month, kind of a true up process to more of a continuous process where it's more real time.
Mark Vigoroso (15:46.122)
Opportunistic might be the wrong word, but it's more real time at pace with the pace that the world is moving at. Xavier, maybe you could just tell me a little bit, if you understand the question, of how has that gap been closed? How have you closed that gap as an organization?
Xavier Duprat (15:57.132)
Yeah.
Xavier Duprat (16:00.857)
Yeah, so like you said, typical typically S&OP process 12 times a year every month, there's a cadence for sure. That's important. We maintain that. So for sure we have that cadence in place. We have the rigor. You know, it's kind of your budget that you kind of redo every month. So that's already in place. That's in place. Of course, it translates to weekly and daily execution. So I can tell you that
Especially now that we have on the, at least on the retail side on the, for the tissue sector, we have that kind of visibility. For sure. We are much more agile and proactive on a weekly daily basis. So for sure that enables that agility, I would say, because you know, planning is about managing different horizons that eventually collide and intersect. So the monthly, the monthly horizon.
comes into a weekly horizon. So super important to have those two tied together and really to have that sensing capability through alloy on a point of sale data. we're tracking over 20,000 store on a daily basis that again, to come back to our previous point that it's integrated in our IBP or ERP. Again, that's very powerful because we don't waste time to connect system together or retailers together.
We have that in front of our eyes on a daily basis. So we can very quickly identify and be more proactive on a daily or weekly basis. So, without going too much into detail, I mean, from a high level perspective, that's extremely powerful because you need, everybody needs that agility. cannot say you just run once a month and that's how I run my business. That's impossible. Like you say, there's too much, speed, the speed, the speed is faster than before the requirements from
partners, retailers, distributors is higher than before. There's more turbulence when you look from a macroeconomic, political perspective, climate, you know, there's a lot more events than we had in the past. So we need that agility. And ultimately we need to serve our customer and have, and that's another topic. But traditionally we weren't looking at OTIF on our side. on time and full, but now, especially since we partnered with Joel, the in-stock is the most important metric.
Xavier Duprat (18:28.717)
that we follow because the most important thing again is to add a product on the shelf for the consumers. That's what drives us and that's what drive our planning and execution today. And that's a big, big, a big step change that we did since we, since we partnered together, it opened our eye, you know, to that, the real metric that we need to follow because you can have a perfect OTF, but the shelf could be empty.
Mark Vigoroso (18:56.376)
Right.
Xavier Duprat (18:56.409)
So it's really important for us, especially in these times, to make sure we have all the tools to be proactive and make sure all the chefs, all the product are there at 100%.
Mark Vigoroso (19:08.149)
That's great. And Joel, I'm thinking, putting myself in your shoes a little bit and maybe you can help us understand like, I'm listening to Xavier and I'm thinking about these as, as requirements, these are evolving operational requirements that are customer centric, right? And his customer, right? And I'm curious, if you go back to the origin of the partnership was
Mark Vigoroso (19:39.492)
was there something unique or recognizable about the challenges that Cascades was having that made you confident that Alloy had the answer? That's sort of the first question is like, what made them a good fit for what the platform you guys had built, have built? And then second, how has that understanding evolved and perhaps even pushed the direction of your platform in terms of its capabilities? Have you?
Have you had to rise to the challenge or offer more continuous or more responsive data or different data sets or tell me a little bit about how your own business has evolved from the beginning of the relationship with Cascades.
Joel Beal (20:24.334)
Sure. So I think it starts with we work exclusively with basically consumer products companies that are selling through retail. Now we touch physical retail. We do e-commerce. We do professional channels. We really do all sorts of different distribution. But retail still, I would say, are bread and butter. And that's where we have the deepest experience. And so
We have a lot of confidence when we go in and I'm sure we met Xavier, I think a couple of years ago at the Gartner Supply Chain Conference. And, you know, when we understand what someone's distribution looks like and you know, Kascad has multiple business units and we don't touch some as much, but certainly on the consumer side, we can flip a switch at this point and get that visibility as he said, into 20,000 stores. I mean, it's something that we've done so many times. So from our level,
Xavier Duprat (21:00.387)
Mm-hmm.
Joel Beal (21:22.798)
that's pretty straightforward, right? We can do it quickly. you know, I think it was a couple of days after we actually got started with GasGod, they were seeing that first data coming through. But I think to the second point, look, we see a wide range of how people adopt getting that data. It's like all of a sudden you turn on a fire hose. And for most companies, again, they've been getting this data, they have access to it.
But all of a sudden they've got this normalized stream of 20,000, 50,000, whatever it is, number of stores and however many skews they have coming in every day. And very rarely are people looking at daily data before we get to them. Usually if they're doing anything manually, it's weekly at best. And that's where you see a difference. Some folks are gonna say, look, my goal is just to better understand consumer demand. I need to have better reporting, tell better stories internally. There's a ton of value in...
but of course, you know, our goal is to push that a lot further. And I think that is where cascades has been an incredible partner of saying, first of all, and most of our customers do this as well. I want to really integrate this in with my supply chain. This isn't just about getting better visibility for sales and better, you know, kind of retail partnerships. This is about how do I operate my business? you know, so it's really, you know, this demand driven supply chain.
and you know, specifically with cast God, they push us and these are our favorite customers. mean, we love working with them because we don't know everything. We are not the practitioners that they are. We are a technology provider. Now we work with a lot of companies. We learn a lot of best practices. but there's nothing better than someone saying, Hey, let me show you this specific problem. We are trying to do this. You know, we've got all the pieces. Can we work together?
to build that. And I think a lot of the best innovations that we've created, I wish I could say we came out of them, we came up with them, you know, out of the void or we were so smart, but it's really somebody saying, guys, why aren't you doing that? we're like, yeah, that's actually, that makes a lot of sense. So I think when you talk about that, and like I said, it just speaks to the organizations that are pushing it. Now, one more thing that came to mind as you were talking earlier and kind of the speed.
Mark Vigoroso (23:31.555)
Yeah. Yeah.
Xavier Duprat (23:34.169)
Mm.
Joel Beal (23:46.146)
going from monthly to weekly to daily, the first retailers, now e-commerce, it's real time, right? I'm seeing at the moment the order comes in, but talking about physical retail where there's more complexity, and it's an older industry really, you we have retailers now that are reporting, you know, hourly inventory levels and hourly point of sale. So this is gonna keep, I mean, it's gonna get real time there too eventually. And so it's kinda like,
How much are you pushing? Are you somebody who's sitting there being like...
I don't even want to be thinking about things on a daily level and to be like, you know, your competitors are going to be thinking about hourly and then real time as that data becomes more and more available. And speaking of lost sales and an out of stock, know, if I only have weekly data, I only see at the end of the week where I'm at. If I only have daily data, I only see at the end of the day, you know, do I have a, you know, do I have an old, um, so you really need to see the hourly, the real time to be like, Oh, there was a gap of two hours.
Xavier Duprat (24:39.763)
Mm. Mm.
Joel Beal (24:50.28)
And in fast moving consumer goods, like as gods in that matters, right? I want to see any gap in the shelf. And so, you there's a lot of value to getting this more real time data. Obviously supply chains take time to move products around. But I think, you know, we're going to see more and more differentiation across companies of those that are kind of running after this problem and excited about it. And those that are lagging and I think are going to lose as a result.
Xavier Duprat (25:17.529)
Can I...
Mark Vigoroso (25:18.359)
Yeah, for sure. Yeah, go ahead, Javier. Yep, go ahead.
Xavier Duprat (25:21.687)
I mean, it's such a great conversation. I just want to, I'm reflecting at the same time, Joel, what you're mentioning. I think the important aspect we didn't touch on is the integration goes both ways, right? So it's not just from retailers to IBP or Cascade, but it's the other way around. And really, again, a key differentiator is that Alloy is for us is the perfect platform.
to speak to our partners because we let's say we have 10 different retailers. We want to speak their language, right? So being able to bring again, our planning data, inventory data, bring that into alloy and then instantly go from one retailer to another in terms of, because they have unique, let's say languages, they have unique different calendars, different way they call their product. So, but.
Fundamentally, everybody talks about in-stock loss sales. So again, to be able to pivot like that quickly. Internally, we have our language. We look at things globally. But then we can pivot and be customer facing. So we have planning teams. We have customer fulfillment teams. So there's different use, to working together in that partnership.
And also eventually we're going to touch base about it probably, but the, when we start introducing AI agents, eventually also it will be interesting on both sides. So internally and externally with retailers, when we speak their language, because IBP is not really made, you know, to speak to customer. It's, it's going to be complicated. It's, it's not, it's not the place to do that, but we think that always the good place to be able to, to enable that next.
that next step sooner than later.
Mark Vigoroso (27:17.985)
No, it's a great it's a great segue, Xavier. In fact, the AI topic is unavoidable, right? It's it's a has to be discussed. And that's where we're going to head next. Because it's almost like the you know, like I said before, timetables have shifted, playbooks no longer apply, at least in some cases. The ceiling, the performance ceiling is now for anything, any business function in terms of productivity and throughput and things like that is now
Xavier Duprat (27:24.289)
Exactly.
Mark Vigoroso (27:47.98)
sort of being challenged by the new capability that AI brings us, whether it's generative or conversational or agentic AI or any of the other flavors. And there are companies out there like McKinsey, well-respected companies that are estimating that in supply chain specifically, AI can reduce forecasting errors incrementally by up to like 50%. And they can cut, you know,
AI tools, generically speaking, could cut lost sales by 65%. These are massive numbers that presumably are based on McKinsey's own research, know, alloy itself, alloy AI itself, you we talked about in the intro, you know, are already claiming and I think realizing 35 % improvement in demand planning accuracy, right? So, you know, I think in all of that, right, you know,
Cascade, you're on a great journey here that's not over, right? I'm not sure it'll ever be over, right? It's a continuous iterative process in partnership with folks like SAP and Alloy. And I think one of the things that you're able to do is you're responding way faster to volatility in demand than perhaps historically was the case. Maybe COVID is a great example of that, right? When demand shifted.
dramatically. And so I guess the question again with that backdrop for maybe we'll go to Xavier to begin with is based on where we are now, based on where you and your team is at now, where is the incremental value to be had within your planning processes, presumably on the shoulders of things like agents and things like that.
Where is that value to be had? Have you begun to uncover those additional value creation opportunities? What are some of those remaining obstacles perhaps that you're still facing as you continue to move this ship forward and continue to refine the precision, the accuracy, the productivity, the throughput, the responsiveness, all the things that have always been the aspiration of S &OP as a discipline.
Xavier Duprat (30:07.705)
Mm hmm. Well, for sure. First of all, COVID, it's still what, five years ago? So that's kind of old stuff. So for sure, everybody heard about the toilet paper panic buy and stuff like that. But we still see those not to that scale. let's say last year we had a port port strike in the US. We saw a panic buy. We see panic buy this year, earlier this year when we had the freezing and snowstorm in the south of the US. So we still see those events and they're gonna keep happening for sure. Not to the level that we have hopefully in five years ago, but these happen and you're right. I mean, the idea is to be, we sense those events, we're able to be a lot more proactive. And in those circumstances, you often need to do, let's say allocation, move product around.
And we know where the product needs to go. That's the difference versus, let's say, five years ago. We just follow what the retailer wants. There's too much demand for the supply. And then we end up not sending the product to the right place. So the difference today is that we know where to send the product. And we're able to manage and overcome, generally speaking, those events that are extraordinary. And for sure, we saw in the first year of implementation,
Mark Vigoroso (31:29.761)
Yep. Yep.
Xavier Duprat (31:35.054)
We saw the lost sale. First of all, we had no idea of the magnitude of lost sale that we had. So that's the first thing. And then just in the first year after a year, we're able to cut it by half, is, and it's quite a significant amount in dollars. So the value, obviously for us, the number one value is the lost sale, the end stock. So it opened our eyes to end stock, which was not stock at all in the business three years ago. So in stock was.
very fragmented, was reserved for the, let's say the commercial team. was account by account, but now in stock is fully visible to everybody. So every morning, our CEO, executive vice president, everybody looks at this metric on a daily basis. It's the utmost, most important metric. So it drives everything we do. So again, that's the most, that's the foundation. So by shifting to the in stock,
For sure you work on lost sale. For sure you work on improving efficiency, not just for us, but let's say with our partner. So shared efficiency on working capital. And we see some heavy reduction in working capital on both sides. So let's say a third on the retailer and two thirds for us. And not only when you look at the finished product, but also on the raw material that go through that chain. So for sure the value.
also is on working capital. And of course, in terms of, let's say the size of the team, I need to run the business for sure it's lower than before because I don't have to have people that manually grab data, put it together, waste time. know, I like does that for us. They're the expert, have over I think 450 integration. So.
You know, they're always up to date with the recent change when retailer update their platform. So I don't have to worry about that. So I can only focus on what drives value, you know, which we talk about. So for sure, it's a big, big evolution and the future I think is going to be the most exciting because we already did the, let's say the, the hard work, let's say the integration of the ERP is not something you do in days. So we're ready for the next phase and
Mark Vigoroso (33:38.092)
Mm-hmm.
Xavier Duprat (33:55.482)
And I let you Joel talk about it, but the roadmap that they have, think is really interesting in terms of pushing that AI component to the next level. But we're all for it. And we want to be out of the game, right? So in the industry, because for sure, I think others are going to follow or are probably following, but we really want to be there for our partners or retailers and be the partner of choice. So out of the game.
The in-stock that's really the most important.
Mark Vigoroso (34:29.933)
Yeah, so absolutely. And maybe Joel, you can lend some insight to this. But as you look at the future, near term future, it's hard to predict these days. But, you know, I don't know, 12 to 18 months out, with a lot of claims being made around what the autonomous supply chain looks like, that's one word, or what the hands free supply chain looks like, right?
effectively making some pretty aspirational claims about how much execution work can actually be done by things like AI agents, right, as opposed to just guiding human decision making, right, that's sort of the fundamental premise of agentic AI is that it actually does the work based on, you know, highly informed
positions, if you will, that they develop based on data sets and analysis and whatnot. So I guess the question is, Joel, as you look forward with your partnership with Cascade, what is the role of AI going forward, maybe, obviously, across your entire customer base, in delivering more value to folks like Cascade and ultimately continuing to raise that bar when it comes to innovation?
Joel Beal (35:53.41)
Yeah, so let me give a very concrete example because I think AI is both over and under hyped at the same time. And I believe like I think it is everywhere you go right now, every company is touting AI everybody's talking about in their earnings calls. And and of course, I think you you dig a little deeper and nine times out of 10. It's like, okay, this is marketing speak or this is someone throwing a word around. And yet when you really see what's, I mean, we aren't even scratching the surface of what is possible with the technology. I mean, it is, it's insane. Like in it and terrifying and exciting at the same time of what this is going to mean. But let me give a very practical example for us as a company that, you know, we're an AI company, we're developing this and what we see. We are rolling out as we speak to our first customers, a,
Mark Vigoroso (36:26.455)
Mm-hmm.
Joel Beal (36:47.746)
You know, we've been making recommendations. So if we talk about lost sales, it starts with understanding that you have lost sales, as Xavier said, right? And making that a true North, which we're seeing more and more companies do, right? It's the easiest money I have, like just avoid the sale a customer wants to make or avoid losing the sale a customer wants to make by making sure it's there, right? Nothing rocket science, but easy to say, hard to do. Once you get the data together, the natural next thing is, well, how do I avoid this?
Joel Beal (37:16.594)
And some out of stocks are unavoidable. Sometimes there is, you know, we work with a lot of health and beauty brands. Something goes TikTok famous. Like you can't necessarily predict that. Okay. No forecasting algorithm is. So you're going to be on your back foot, but oftentimes you're looking where, where do I have a problem that I could have anticipated? Like this wasn't, this was because we didn't execute QQ correctly. and so, you know, we built this recommendation engine that will now.
Joel Beal (37:46.318)
turn that, hey, these are your lost sales. These are the ones you can avoid. And these are the orders that retailers should be making to you to avoid those out of stocks. And here's all the data to support that. So we've been doing that now for a couple of years. We have lots of great examples, including with Kazgaz and many other customers, millions of dollars of incremental orders that you get, but it still requires a person to log into Alloy, see the recommendation.
export those results, write the email saying, hey, I think you should do X and I see hundreds of these examples from our customers. And then maybe they get a response from that retailer saying, well, what about this? Or I don't have enough budget for that. only have, you know, all that. Well, now with AI, and again, we are rolling this out as we speak, we can do that whole process for you, right? We can write that initial email. We can send it.
Joel Beal (38:41.634)
we can wait for the response from the retailer. When they come back and say, well, wait a minute, give me clarification. We have all the data to give them the clarification. We don't need a person to do that. So I think that's where you're seeing a lot of the exciting when you say, hey, how do we change from recommendations to actual action? You know, there's a very human element. By the way, the retailers are gonna do the same thing, I guarantee you. Like they're gonna be building their own agents.
Joel Beal (39:08.994)
they're going to be interacting back with their suppliers and you're going to end up with two agents going back and forth with their respective information. But that is a very real thing that we are seeing customers use today. So this is not in the future. This isn't not, maybe it'll work. And of course, again, I think we're scratching the surface of what's going to be coming in the coming months.
Mark Vigoroso (39:34.734)
Yeah, we are. although you mentioned, gosh, mean, a segue again, Joel, you talked about the human factor, right? There are humans, the sensational headlines like to say that it's the rise of the machines, right? But the humans are a critical component of what's happening right now. If for no other reason, then there's this massive thing called change management that...
organizational change management that is staring everybody in the face. And I guess this is one for especially for you Xavier about the human factor and what has happened and what you anticipate to happen at Cascade with regards to change management. You know, I'm thinking about forecasters, supply planners, demand planners that are, you know, career statisticians maybe or perhaps they've gone through training certification, education.
years, decades, right, very good at their job. And, you know, they're managing production schedules, right. And perhaps they're doing that on spreadsheets. And, you know, it's, it's sort of the classic fear that, you know, AI is going to take my job type of thing, or automation is going to take my job. And, and, and in many cases, as we're seeing in the headlines now, there are tens of thousands of jobs that are being, you could say, redeployed.
you know, Oracle and UKG and big tech companies that are, that are making significant personnel moves. But I'm just curious how you and your team and your colleagues, Xavier are thinking about this, this human reality as, things evolve so quickly, technologically and operationally, what's happening organizationally, what's happening culturally. and what does that day to day planning job look like today?
versus yesterday and versus tomorrow, if you follow.
Xavier Duprat (41:34.274)
Yeah, yeah, yeah, for sure. And Joel, I think you're absolutely spot on. know, today, let's say I have people with master degree spending a lot of time clicking in systems. So like you like Joel said today, we go in the alloy system. We still have to do, let's say a dozen of clicks to get the info we want. Like you say, write the email. So all of that, let's say for someone for our team is not really added value, right?
So today that happens also in, in IBP, in APO, in the planning system. So we have a lot of people doing a lot of clicking around, sending information by email spreadsheets. that's, and everybody's excited to, to go again to the next era, which will be, you know, the system does that for you and you're there to train. You're there to, you're more, say acting as a business expert rather than a
I would say, let's say a clerical, you know, doing clerical tasks. So I think our job are going to be much more interesting because there's going to be, you know, less, less time, a lot less time, I hope spend on doing this thing. And by the way, there's also a lot of people communicating with each other for very basic information. we're really, we're really hoping that
the agent AI is gonna empower our jobs. And obviously the skill and the talent pool that we have, obviously is gonna be people with more expertise and let's say to the level of let's say bachelor, master's degree. So for sure it's gonna head towards that, but it's nothing that is scary for us. we're not, we already,
started, let's say that we were aware that that transformation was coming. So we're super heavy on people. we're really, again, always focusing on having the right talent that's going to be able to leverage the technology. Let's say we have people working with the alloy team on a daily basis. We're pushing the system. We're pushing the limits. So like Joel said, let's think you said we're demanding. But we have people that are experts that can do that and that's really important to have this kind of skill set for us.
Mark Vigoroso (44:05.431)
Yeah, it's really exciting. I mean, it's terrifying and exciting, think Joel, you said, or something like that it is, because I'm starting to see people sharing org charts that are delineating between human labor and machine labor, right? They're basically showing workflows and tasks that are now the responsibility of humans as opposed to agents, right? And there's...
the agents have names, I mean, not like Bob and Sally, but they have names that are indicative of the task that they complete or multiple tasks, right? And so to your point, some of those tasks that have historically been done, maybe begrudgingly or out of necessity by people that might be overqualified, they're now delegated, right? To agents, right? Fit for purpose agents, right? And so I think I'm, you know, obviously I'm seeing this across the board and
Mark Vigoroso (45:01.537)
you know, I think and we're gonna run out of time here, guys, but I think we could probably go forever. Good thing this is a multi episode podcast. So we'll be back. We'll be back more later. But I, I wanted to I wanted to kind of dive in. Xavier, you gave some great examples already around like the toilet paper example. The panic by example. You know, there are things now that I would only imagine that that your team is now capable of doing that couldn't even have been imagined.
Mark Vigoroso (45:31.887)
years ago, right? Whether it was call it pre pre transformation pre IBP pre alloy, where where you were living with a status quo, that now would be unacceptable, right? And I'm curious if there are any sort of standout examples that might serve as, as as instructive for the 1000s of people that are listening, when it comes to like, what what's a ... Are there sort of quintessential example where you were able to be, were able to either through a metric like OTIFF or on the shelf or some other kind of systemic improvement that you were able to make that resulted in customer retention or expansion or something that just kind of proves out the story that you've been telling us so far? Anything that comes to mind?
Xavier Duprat (46:26.713)
Well, I already mentioned the, you know, the last cell improvement that happened pretty quickly, but at the same time, our business grew by 10 % and in stock improved by two, 3%. And it might not seem a lot to have two, 3 % in stock, but it requires a different level of depth and knowledge of your retailers partner supply chain. One example, let's say it's not, and Joel, you're right.
Xavier Duprat (46:55.127)
The engine in alloy is about, say one of the big, big benefit is the replenishment recommendation. So you touch that. So there might not be sufficient level of orders from the retailer. So we're able to detect that to this queue to the DC level. that that's one thing, but we're also to give another example, we're able to detect also if there's a flaw or sometime a glitch in the retailer system, let's say there's no distribution happening between a DC to us to stores to a cluster of stores. So
We're also able to detect that, which in the past, in the past, we would never be able to. So we're kind of a second pair of eyes for the retailer because they might have not spotted it. They might have something else to do. Their, their expert might be on vacation and then you have someone else covering and that person doesn't have all that knowledge. So we're able to just jump in at the same time and have, again, have that pair of eyes watching and because we flag those stuff now.
We never did in the past. And again, on a daily basis, we look at the in-stock for everybody, for all these queues. So just to have that metric in our face, sorry for the expression, but in our face on a daily basis drives totally different behavior from the entire team. Not just the supply chain team, commercial, operation, because we onboard everybody with this metric. So operation team is well aware of what it means and they follow it as well. And that's, I think, the
biggest shift that we did and the most valuable one for us and ultimately for our retailers. So I think that's another good example that we could share with the audience.
Mark Vigoroso (48:33.155)
That's great. That's great. Joel, I'm gonna I'm gonna give you the last word in the spirit of what's to come right forward looking and perhaps in the spirit of giving advice to people who are listening, looking for something actionable when you think about the role that AI is going to play in supply chain planning specifically, you know, Joel and team you get you guys have a lot of experience in retail and consumer goods. Do you have any statements any sort of like
Futurist perspectives on on where agentic is going to take us in the supply chain space specifically with regards to planning Any any it doesn't have to be anything I'm gonna hold you to but any anything you expect to be seeing in the few next few say 12 to 18 months that you'd say Folks that are in roles like Xavier should be preparing for expecting
Joel Beal (57:48.578)
Well, think Xavier actually summarized it really well. First of all, there's a lot more that's exciting here. Again, it is a little scary because our jobs are all changing, all of ours. My job's changing. But he just commented on how we're getting out of the clerical work. I think that's the term that Xavier used. And that to me is exciting. We all do. I do lots of clerical type of work, right? I'm responding to emails. I'm having to...
Joel Beal (58:17.528)
click a bunch of buttons, they add up, they take time. And I think that to me is what's exciting is I think our roles are everybody's going to become kind of a manager of AI. I get to give it ideas and we're still going to be very important in that. Anybody who's spent time developing with AI, it is so much about what are the instruction you're getting? How do you iterate with it? And then how do you review the outputs? And then usually continue to iterate. So I think, you know, our roles are evolving. Everyone needs to kind of Embrace that I think the mistake people make is they start to say, I don't want to use this because I'm concerned about what it means. And candidly, then somebody else who uses AI is going to take your job. AI is not going to take your job. It's going to be somebody who uses it effectively. And so that's, that really should be the perspective, but we are, you know, I think my, recommendations to people right now is just start using it and trying it. I mean, I use it obviously a lot professionally. use it personally.
Joel Beal (59:17.95)
I I built my own running coach because I love to run and, and, and it's fun to iterate with it and tell it every day. Okay. Well, let's change how we're doing this calculation or I don't like this logic. And the more you use it, the more I think you see both the incredible strengths it has and it will blow you away. And also how dumb it can be. I mean, there are times that it really doesn't work. And I think that gives you greater confidence of, again, I'm going to have an important role going forward. And, but I'm going to have.
hopefully more fun doing my job, because I can spend my time doing the things that I'm really good at. I'm uniquely good at that it's gonna be hard for a computer to replicate.
Mark Vigoroso (59:57.513)
Yeah, it's very wise. It's very wise. Well, great, great guys. I appreciate your time. You know, it's really inspiring with the journey that the two companies represented here are on together. I mean, just the power of the partnership, the willingness to take some calculated risks and to benefit from the results. I just think it's inspiring. So gentlemen, I thank you for your generosity and your candor with this conversation. Joel and Xavier, thank you so much.
There will be another episode coming folks. We will have episode two in the early part of next month where we would continue this conversation and continue to dive into hopefully some insights that will be useful to folks that are not only in retail and consumer goods but perhaps in other industries where of course AI is also making an impact in the supply chain functions that we're also familiar with. So with that, thank you all. Again, any comments, questions, reactions, please drop them in the comment section.
I want to thank SAP for helping to bring this podcast to life. This is Mark Vigoroso, founder and CEO of the Enterprise Edge. We'll catch you next time on the next episode. Take care, everybody.
