The Next Frontier: Implementing AI in Your Supply Chain

Hello and good morning. Thank you all for being here. I'm Courtney. I'm with LEK's marketing team. And today, I'd like to welcome you to our webinar series, our final one, part of our Gen AI series here today.

This webinar is titled The Next Frontier, Implementing AI in Your Supply Chain.

A few housekeeping things before we start. One, this webinar is being recorded and will be available to those for on demand viewing at a later date. Second, if you have any questions, whether it's technical or for our presenters, please submit them in the q and a chat to the right of your screen.

Our team will address them and any technical questions that you may have throughout the presentation and Q and A will be held at the end.

Finally, today's presentation will be made available for those who are interested. Please contact us after the presentation is over, and we will make sure to reach out to you directly. Now without any further ado, I'd like to welcome LEK's head of data and analytics, Nick Barker. Nick, over to you.

Thank you very much, Courtney, and welcome, everyone. Good morning or good afternoon to those in in Europe.

So, yeah, this is our fifth and final webinar in our AI series focused on AI in the supply chain.

I'm sure many of you are thinking, hey, AI is not new to the supply chain. It's been around for decades. And that is completely right. But we also believe we're on the precipice of this next frontier of AI in the supply chain, as the webinar is called, with a lot of new technologies as well as emerging technologies really starting to mature and drive business value.

So excited to get into that. But starting off with a quick intro. So my name's Nick Barker. I head up the data analytics team in the US at L.

K. So I oversee our data science, data engineering, and advanced analytics teams.

Spend most of my time helping clients drive value from their data, often leveraging advanced analytics, data science techniques.

And as part of that, spend a lot of time around the supply chain, implementing demand forecasting, inventory, forecasting tools, network routing type optimization as well. So yeah, great to see you all and looking forward to diving in.

All right. Next, Matt Stanfield, a partner with LDK. I'm responsible for our supply chain operations and procurement capability.

And my primary areas of focus is about activating strategy. So when a company is trying to operate in a new market, drive growth, do they have the supply chain capability to enable it, as well as more established companies really helping them optimize how their supply chain is configured so that it delivers on the right customer experience at the right cost with the right service levels, and then a host of optimization elements underneath that. But it really is about delivering product on time and at an acceptable cost level.

Stuart Robinson, I'm a partner based in London, and I have the joint title of disruptive analytics lead. So within LEK, I'm responsible for thinking through how new and emerging analytic technologies affect business. Much of that is of course AI and especially in the last year or so generative AI, but I also consider aspects like quantum computing, which is also very important for the supply chain given its implications on optimization.

Excellent. So if we just jump into a quick overview of what we'll be covering today. So we're gonna start off with a kind of baseline intro to to AI and what we're talking about when we reference AI, as well as what we're talking about when we reference the supply chain. We're then gonna do a a a quick overview of some of the more established use cases in the supply chain, and perhaps where we're starting to see evolution. And then we're getting into the really exciting bit of what are some of the emerging use cases we're seeing today, and what do we expect over the next few years. And then we'll also be leaving a good amount of time at the end for Q and A. So feel free to use the questions box on the side, pop in any questions you have, and we'll get to them at the end.

All right, so without any further ado, I'll kick it over to Stuart to talk through this one.

Thanks Nick. And I'm not sure of those watching, if you've been to some of our webinars previously, but you may have seen us explain the journey of AI in a slightly different way there, as we've typically tried to illustrate how it has developed from the 50s. One of the things we're most conscious about when it comes to the supply chain is that implementing what's now bucketed under the broad term of artificial intelligence AI are a lot of techniques that the supply chain has been using in various different ways for a long period of time. So we really wanted here to call out all the different things that are now bracketed under this slightly broad banner, whether that's machine learning and natural language processing, those of you who are particularly interested in generative AI, the intersection of those two.

But we need to go beyond that when it thinks about the supply chain, we need to think about what's going on in speech recognition, how is computer vision assisting in predictive maintenance or other sorts of assistance there, and particularly robotics and the automation considerations within it. So a big broad waterfront where, as Matswab described to some extent, is not new, but is definitely changing the way supply chain works.

Right, great. And to talk a little bit about supply chain, just to make sure that we use a common lens. As I previously mentioned a little bit in the introduction, supply chain for us starts with one alignment with both the strategy of the business and the customer expectations. Otherwise, can end up building or optimizing a divorced supply chain for what you're trying to accomplish.

And then if you think about the spectrum of the functions within supply chain, it incorporates both your demand and supply planning, any necessary flowing of logistics, and of course, the interconnectivity required, how you procure materials and services to then feed that into a make or manufacturing element, and then ultimately how do you get that to your customer via your distribution, either system or network all the way to your final mile and as necessary return logistics. And if you think about the customer interaction there, we've all used various user interfaces as we purchase goods and kind of manage that order from the time they've received it all the way to delivery.

That's what we want to make come alive. That's what we want to optimize, drive efficiencies and make sure that it is resilient or that it can withstand various impacts, either increases in demand or decreases in demand.

That's when we again, when we talk supply chain, it really covers that end to end perspective.

It's funny, one of things you started with Matt, around the importance of routing this in customer expectations and corporate strategy, as someone who spends my time thinking more around how analytical techniques are applied, that to me feels like where these technologies and these ways of working meet. Because it's easy to think about AI and supply chain as a sort of specific bit on every bit of this journey. But actually, if you root both of them in what you're trying to achieve and experience your customer should be having, then it's much easier to produce a holistic and seamless experience for people at each stage of this journey.

Yeah, Stuart, I think that's a great amplification. And the only part I'd add to that, which I think was consistent with your point there, is also the sustainability element, both from a resource consumption standpoint as well as who is managing these processes, the more you can get them, you know, automated and driven through AI so you're making advanced decisions, the better off your supply chain will migrate as your customer expectations are going to change over time.

Yeah, definitely.

All right. So excellent discussion, but I'm excited to get into some of the meat of today's call. So firstly, we wanna emphasize again, AI is is not new to the supply chain. It has been used in in various forms for decades, and we wanna just briefly cover some of the the more established use cases. So with demand forecasting, AI has obviously been a game changer to that space for many decades now.

With the onset of predictive analytics and machine learning, companies were able to digest rich historical data on sales over time, by product, by region, and use that to create very granular, precise forecasts in a way that was never possible before. And that comes with huge benefits for companies, whether it is reducing inventory and working capital expenditure, or whether that is making sure you're not having too little inventory and then missing out on potential sales. So huge, huge game changer there. I think one point that I would wanna call out though, which has been particularly pointed recently, is these sort of forecasting tools and AI more broadly is the predictions are only as good as the data that you're putting into it.

And this was really highlighted through the COVID era where suddenly companies were no longer able to rely so much on what happened last year or the year before that because we were entering a whole new world.

And so we worked with a number of companies around this, particularly in pharmaceutical space, in healthcare and consumer goods.

There's a lot of change there, that was where it really became important to not only rely on historical data from internally, but what external data or insights can you get from things? Maybe it's website views and how many consumers are interacting on your site. Is it things like social media, weather data, all that sort of stuff? So being able to integrate a broader set of data other than just historical we think is really key to coming up with accurate, reliable forecasting predictions.

It's it's interesting, Nick, because the example you're giving there as a risk of what happened in COVID and so forth, if if I may give an anecdote, which I think illustrates the importance of merging this technology with people and active decision making. So we worked for clients shortly, well, midway through the pandemic, who had managed to double their inventory in a matter of months because when the pandemic hit, they were in the aerospace sector, the aircraft were all grounded, the need for their services was decreasing, but they couldn't or they didn't quickly enough shut off the automation around ordering their parts and their supplies to make the equipment they made because that was running off history.

And it kept saying to them, even when they said cancel that order, it would immediately reorder going, no, we need this because this is the demand that's going. It took them a while and I suppose there are two lessons I have here. One is that you definitely need to have people working in tandem with machine, and the other is check your service level agreements and your contracts with your suppliers very carefully because it's useful if you can if you miss order that you can send things back, which in this cloud instance they couldn't.

Yes, and I would offer one other point there too. You've hit on a couple of what I would say pretty significant world events and the consequences of having disconnected elements within the planning function. But there's also elements such as, if you think about a retailer where weather patterns change and therefore you don't need as much rain gear in one geography or you need more. And then the same thing on introducing fall and winter clothing. These kinds of AI tools that are linked to or that can actually include weather as part of their forecasting will help a retailer replenish what they really expect to be needed and constantly check that up as the year, the month, the quarter progresses, which again, Stuart, back to your point, having the person and the machine being connected, but it's doing a lot of thinking twenty fourseven will make sure your supply chain is efficient and effective.

Yeah, you're right. And it's funny, the example I give is more sort of warning against unnecessary cost, but I really like the way you're positioning it there around actually being able to best serve customers according to what demand is going to be, whether that's local, whether that's in response to seasons. And and actually over the long term, as as tastes change and requirements change, capturing that revenue opportunity quickly and making sure that you have the supply chain to deliver it is critical. So, yeah, I feel my example maybe is inadvertently given the negative side, but it's good to remember the positives as well.

Yeah, yeah. Very exciting space. And I don't want to spoil what's coming, but I think some of the innovations like generative AI mean, you know, the the world for the types of data that can be embedded into these systems is is is only gonna increase. So yeah.

Stuart, I know you have spent a lot of time around network and routing optimization over your career. Maybe you can talk a little bit to this middle point, what you've been seeing there, give a little bit of color.

Yeah, I mean clearly how one plans a network and optimises routes around it has been of interest to humanity for hundreds of years. Know, it's a well studied mathematical problem, and the economic consequences are greater now than they ever were before.

Saving even a minute or two on per delivery time can result in tens of millions of additional operating profit for those who are able to deliver it. I think there's a few things I observe in this space. The first is, and it relates to the point you were making about alternative data sources, is that as a mathematical problem and as an algorithm to optimise networks, that's a really hard mathematical problem. I mean, in algorithmic terms, it's known as NP complete, which broadly speaking means it takes a very long time to solve it well, unless a quantum computer comes along when it might be faster.

So the question you're asking here is either can you make small changes around the edge, can you find new data sources that allow you to solve the problem in new ways, that can be more material, or critically, when you have the opportunity to change the network, so you're not just constrained to trying to eke out small advantages on a fixed basis.

At that point in time, can you improve that network in some fashion? So if you're putting in a new distribution center as a logistics provider or retailer, where you put that will be one of the most impactful decisions for the cost of your supply chain that you ever make. Because once it's there, now you're talking about microscopic adjustments on current optimization problems.

I I think that for me is one thing I would want people to take away here is AI can help you with this in terms of once it's fixed, but really do apply AI and optimization techniques to think through critically where the best place to locate things and how the best way to design the network is, because that will create long term value in a way that shaving tens of seconds off your delivery time is important, but it's hard yards for small improvements.

Yes. Stuart, I've got a client in the life sciences space. And just as of very recent, one of the comments from their SVP of Supply Chain effectively was, I wish we would have had an integrated and database strategy for where we built our network because now we're kind of relegated to what we have from a physical asset perspective. And then around the margins of improvement that you mentioned, in some cases they're having to fly product, in other cases they're dealing with just inherent expiry of product that exists because it's just an inefficient supply chain. So I mean, that's a very real point that I think has not just short term, but long term consequences to both cost and your reach to the customer.

Yeah, I mean, I remember a last mile logistics provider we worked with who had managed to construct their network in such a way that for large population centers, this is around London in the UK, around the big population centers, they didn't just have the sort of nearest way of delivering, they had the second nearest and the third. So once this came to competitive pricing, the margin they could take because their cost to serve wasn't, know, their next nearest competitor was them again, and then it was them again. So anyone competing with them for this business, just their cost was so much more. They could cost undercut their competitors and still take a giant amount of margin because of how their network gave them those inherent benefits.

And and that's a big strategic question I think people need to keep in mind because those opportunities don't come up often that you are able to redesign that once you've put the capital to use. It's it's going to be a few years before you've got new capital to rethink it.

That's right. Yep. That's absolutely right.

Fascinating.

And then, Matt, I know one of our favorite topics to chat about is digital twins.

Maybe you can talk a little bit to what you've seen going on in digital twin space and what are some of the benefits of implementing that?

Yeah, so let me start with a couple of early points here. So when we say digital twin, we're really talking about a digital solution that replicates some or all of your supply chain, meaning it can operate from a capacity constraint, from a product mix, and then represents the costs and times to serve, you know, based on how you're currently configured. I can remember starting my career in supply chain a couple of decades ago where I started this with what if calculations or what if formulas in Excel, and it is so much more advanced today to where, you know, if you think back to the conversation we were just having on network, these digital twins allow you to do a couple of things.

Number one, you can kind of assess in parallel, how am I performing and run scenarios if my demand increases ten percent, twenty percent, thirty percent, either in aggregate or specific regions or products, what does that mean to my costs, my ability to provide it, and or in some cases, it might change my inventory strategy.

Further, as you start thinking about deploying a digital twin from a strategic perspective, you can start doing some of that re envisioning that Stuart was referencing. You know, if you don't have the network that you want today, how do you build the network that you want that really satisfies that future? So it ties back to your new three, five, seven year strategy and what the expected evolution of customer expectations are going to be. It allows you to simulate that within a safe environment, so to speak, so that you can see what would happen before you really start spending big capital dollars and making big bets on physical assets. And so, this gives the organization an opportunity to test a number of permutations before you really start committing. And then also, of course, it lets you react to changes in customer demands, all with giving you some ability to be proactive.

Yeah, and I see the logo Kraft Heinz, and it's obviously a very sophisticated implementation, but with that, it also sounds somewhat costly. Right? Do you have any guidance for the the types of companies that a digital twin is is suitable for, or am I thinking that about that in the wrong way?

Well, it's a great line of delineation that I would say both exists and doesn't exist.

From a Kraft Heinz perspective, they built a very bespoke digital twin, so truly bottoms up that's specific to their business and then it runs using Microsoft Azure's capability to keep it alive and run calculations.

However, as you start thinking not from a you know, twenty dollars thirty million plus company and start migrating down, I've helped companies with as little as fifty million dollars in revenue use some form of digital twin, but you start looking for more off the shelf products and platforms. In many cases, they're software of service, you know, kind of environments. You lose a little bit of the specificity and functionality, but it gives you the same amount of capability to make decisions on your network. You know, not to the same level of nth degree of accuracy, but certainly enough so that you can you cannot just kind of stand by and watch, you can continue to be proactive. The last point I'd make here is if you're a single node network from a supply chain perspective, you don't need it. When you start getting into multiple nodes, that's where it's going to become far more necessary.

Yep.

Excellent. Well, I know we could spend all session talking here, but we must move ahead. So before I jump into this slide, I actually want to bring up a poll for the audience.

So this poll is just to gauge current adoption of AI solutions.

So you should see it popping up or on the right, and it's how would you describe your organization's current usage of AI across its supply chain? With options for is it cutting edge, ahead of the curve, playing catch up, or need to do more.

Hopefully you should see that. And so while we wait for some of the answers to come in, I'll just introduce this slide. So L. K. Also ran a survey last summer. And for this survey we interviewed, I think, close to one thousand business leaders across different sectors and different functions asking about how they expected their adoption of AI and piloting of AI solutions to change. So we asked, by twenty twenty five, how do you expect your adoption to change from twenty twenty three?

So they are the results that you're seeing on the left.

Stuart or Matt, any reaction to the answers you see in here? I definitely have some thoughts, but curious to what you think.

Just one comment from my side and Stuart, kind of goes back to a point you made early on is AI has been or supply chain has been a test bed and implementation base for AI for quite some time.

And it's really exciting to see that there continues to be value for further investment for more AI in the supply chain as ways to unlock value.

But I guess in this case, Stuart or Nick, I'd ask the question, what do you see as the biggest unlocks through this next wave of AI investments in supply chain? And we've talked about cost and data availability, but what's really next on the horizon?

Yeah, think it's a great question. Just want to think this data potentially hides for supply chain is how much of that ten percent well, more than that increase is people playing catch up versus people who are at the cutting edge. I don't I don't know. Some of this might just be the wake up call that happened.

I mean, for me, I think the big the big things that are being unlocked I see in supply chain at the moment is more generally generative AI allows for an engagement with unstructured data and interaction with customers in a way that was all cost prohibitive previously. And I think that has the potential to be transformational and is currently on the way to it. I don't I don't think I'm seeing anyone who's totally cracked that yet, but there were definitely people playing at the at the forefront. I mean, the bits that I'm always aware really unlock this is some is organizational expertise at being able to use advanced analytical techniques effectively.

And I think it's really easy to forget that you can put all the data scientists you want in a room, but unless they understand how your business works, it's not going to produce value. And I think the other unlock at a more technical level new algorithms. In the same way generative AI relies on the algorithm that drove chat GPT and the explosion we've seen in the last year was something called a transformer that Google Brain discovered in twenty seventeen. It took five years for people to see an application to that, and I think we all see the same sorts of things now.

Algorithms discovered last year will start to creep out and into regular practice before the end of this decade.

I can't tell you which of those will be transformational, but I'm pretty confident some will exist.

That's an interesting point you raise on the human capability side, because if you know, going back to the setting corporate strategy and expectations, the more you deploy AI, the more informed and clear you have to be on what you're really trying to achieve. I mean, I think that's a really critical distinction that you can throw as many bots and algos that you want within supply chain or just your entire enterprise, but if you don't communicate and use it properly, it's going to give you either suboptimal results or in some cases even negative results.

Yeah. Totally.

Yeah. The change management piece of implementing AI is often larger, takes longer, harder than actually implementing an algorithm. I think, yeah, we see that time and time again.

I noticed our poll results are in. I think they're interesting, also potentially not that surprising. So I'll just read them out. Just as a reminder, the question was how would you describe your organization's current usage of AI across the supply chain? And so for those that would say they're cutting edge, that was only eight percent of answers.

Ahead of the curve, only four percent.

Playing catch up, thirty seven percent and then need to do more was fifty percent. So the vast majority would say they're either playing catch up or need to do more, which I think goes to the point that we see all this stuff in the news about the new cutting edge algorithm, big tech is creating these huge strides. But in reality ninety nine point nine percent of the companies look very different to those that we see headlines about.

And the majority are still figuring these things out. It's definitely what we still see with clients.

Stuart or Matt, does any those results surprise either of you?

I think the observations you're making, Nick, I'd agree with.

I'm honest, I think it surprised me a little bit, partially because I've been on few of these other webinars in other topics around AI, and we've normally found it a bit more bell shaped, so more people in the middle and less at the bottom than the, at the top. So the fact this is more skewed towards the bottom has is something I need I would need to think through a bit, but part of me wonders if it reflects the fact that it is a more mature as a a set of expertise, it's more mature and more used to employing analytical techniques, and with that knowledge of what it really takes, I think is maybe producing a more truthful set of answers as to what people aspire to.

Whereas I know in other spaces where we put up similar polls, I think there's perhaps more naivety than I think I'm seeing here in terms of where they're really positioned. So Yeah. I'm I'm I'm quite curious. It is different to other ones we've seen.

I can definitely think of ways to explain it, but I'm it is more skewed towards the could do better, need to work a bit harder end than I think I've seen in other situations.

So Stuart, you're kind of saying the more I understand what it is, the more I understand what I don't understand.

Well, exactly. That's what it's making me think about. Because if I take a previous webinar we were doing around R and D and AI and product development, that had a sort of more classic bell shape. And that one, I was actually more surprised how many people felt they were ahead of the curve in that space.

And I'm seeing a contrast to that and it's making me think through actually the sort of product developments and R and D side.

AI's implementation in that is probably more nascent than in supply chain. If you think kind of like, let's take drug discovery, AI and drug discovery is less than ten years old. AI and supply chain is forty years old if you sort of take the broad definition of AI that is now in common usage. So I may be wrong, but that's one of the thoughts that's occurred to me is, do we have a wiser set of people more aware of just how much better things could be?

All right, so I'm particularly excited about this slide. Here we want to spend a bit of time highlighting some of the emerging use cases we are seeing in the market with different companies adopting AI across the supply chain.

So I'm gonna hand it over to Matt and Stuart. Matt, I know you have spent a lot of time in and around warehouses advising on logistics.

You want to talk a little bit about what you're seeing on the left hand side in the robotic space at the moment?

Yeah, absolutely. And so let's start with warehouses have a bit of an unsexy moniker, But, you know, think about these as getting, you know, groceries, getting pharmaceuticals for lifesaving procedures, and frankly, getting your daily necessities fulfilled. In some cases, hourly, quarter hourly, in other cases on a daily basis. And now this concept has been around for centuries, right?

Pile up all your inventory in one location and dispatch it from there. And because of that, many warehouses, you know, have a lifespan of thirty, forty years, they were not designed to incorporate some of the capabilities that we're starting to develop, and specifically around the robotics piece.

However, what we're starting to see really with the not the advent, but more the evolution of AI is the robots are now starting to be able to work in some of these older designed environments and plan for what's the optimal way both to operate inside this box to minimize travel time, to make sure that we have chain of custody of product in the case of pharmaceuticals, and also cohabitate with humans safely. And so, this has been a pretty significant game changer from both a cost perspective, a service attainment standpoint, as well as it's really opened the aperture for clients being willing to give it a try.

It's now not I have to design a system that can use automation to I can now deploy automation and robotics inside my existing warehouse. So it's kind of brought that hurdle rate down to the point that it's a much more deployable and implementable solution. And that really gets back to AI has been that unlock. Robotics have been around for many, many decades.

Now, they can operate together safely alongside of humans.

I jump into the second and in fact, maybe the third one I'll touch on is they're both very much generative AI focused, which when it comes to forecasting and network planning, mentioned, there's a lot of science that has already gone into this space. I think what has been unlocked recently has been the alternative data sources, as I think you touched on when you talked about demand forecasting, Nick, earlier.

People are working very hard here. There's a lot of thinking about it. If I'm being honest, I haven't yet seen established definitive ways to improve. I'm very optimistic they exist.

But I think this is one where humanity as a whole is still working out how to use this technology to bring less structured data sources into our forecasting and into our route planning.

The third one, however, I think is closer. If I say like the second one, I'd say certainly before the end of the decade, maybe within, you know, two years, three years, we could could be seeing material improvements from this. This third one I think is very, very close, which is using generative AI in automating many discussion so I've made negotiation points with contracts and in standardizing across contracts. And I certainly know law firms across the world. I think just before Christmas, Allan Overy announced they were releasing an AI driven contract negotiation tool.

That is coming live now. It's a question of adoption and people learning how to use it, but the technology is there.

I think the other observation I'd make on this is if you think someone like Walmart, Walmart will have hundreds of thousands of contracts.

At the moment, their procurement teams cannot analyze all of those terms and conditions and create a data set that allows them to sort of formally use that in negotiation or in setting up Ts and Cs and getting their procurement right.

Generative AI unlocks that. And one of the things to think through here is what that's going to do to downstream suppliers. So if you're a supplier into a large organization, there's always been an information asymmetry. You've always been at a bit of a disadvantage. The risk is that could get a lot more soon because you can now look across far more terms. You can standardize far more in your, you know, far more supply relationships in a way that is favorable to you. So it is worth thinking through how to respond to that, because I think this is going to be a fairly near term trend.

Yeah, and Stuart, I just add a couple of quick points on the generative AI around certainly around sourcing and procurement, which is, if you're dealing with significant tail, whether you're in a retail environment or otherwise, this is a great solution to go challenge that, right, without deploying a tremendous amount of human resources. And then the second thing is from my time as an industry operator, we spent thirty percent to forty percent of our human resources doing administrivia. And this kind of capability would shift those human resources far more strategic. And I think that's just the full value of that is certainly Yeah.

Be wholeheartedly agree.

Excellent. And then just to round out this slide, do you want to hit on the supply chain visibility, Max? I know there's been a lot of exciting kind of new things coming to market there.

Yeah, and let me tie it a little bit back to the digital twins. So just keep that discussion point in mind.

But the predictive analytics, imagine with deployment of Internet of Things within your facility that can consistently manage how well your OEE or how well your equipment is performing, What sort of schedule attainment is happening? And then in the background, what's going on with both demand and supply signals? How has that changed your ability next month, next quarter and maybe even next year for you to ensure that you can provide the right products in the right locations at the right time with the right cost. This is the engine that will take real time performance, tie it back to your broader supply chain construct and then help you start flagging areas that you may run afoul based on how it's operating. And so, this means you don't have to stop, do an assessment looking back in time and then running a calculation looking forward.

This will be a real time warning system that will constantly update based on performance levels, sales changes, etcetera. And so, this is going to be a tremendously powerful tool to let people run their supply chain, their operations far more efficiently and also more resilient.

I wouldn't agree more.

And then to to round out our bit of kind of presentation, you know, it wouldn't be an AI webinar without plugging in something from ChatGPT. So here, I I asked g ChatGPT to to create a visual of what the future of the supply chain looked like.

You know?

Not sure I agree with everything, but I I just find it terrifying.

It looks like an invading army sort of commanded by these operators.

Yeah. At first when I did it, I was like, oh, yeah. This is kind of cool. And then as you look further into certain bits, you're like, oh, that's actually, yeah, very terrifying. Yeah.

Should we open it up to questions? See Courtney, you've reappeared to help moderate on our behalf.

Hello. It's been really interesting listening to you guys speak to this.

A question that I have to kick things off while our audience submits their own is, what are you most excited about for the future of supply chain?

I don't know if Stuart, you wanna go first?

What am I excited about? I truly believe that technology and particularly analytical technology has historically provided a lot of value and will do so. And I'm really excited as to how people learn how to use the learnings from generative AI, reinforcement learning, and sort of structured games, I think, could be very exciting in terms of improvements there as and when quantum computers come around. That's gonna have a huge impact on how these things happen. That's gonna make supply chains cheaper, more effective. That's going to make goods cheaper and better for customers, and that that whole loop, I think, is very exciting.

And I think I'm also interested to see how it transforms the expertise needed to be effective in the supply chain because one of the things that particularly ChatGPT did was it gave people who weren't data scientists the ability to wield effectively at quite advanced analytical and data science driven techniques at their fingertips interacting with it naturally.

That I think is going to be really interesting for what that means for the experts in this space and how they can drive new improvements forward. Because they've now got a whole series of tools at their fingertips, which even three, four years ago could only be dreamt of.

I'm excited to see what happens there.

And I would just kind of follow on. I think you and I share a bit of similar excitement. And for me, it's when do you get to a fully integrated supply chain solution versus functional optimization? Because today, we've got incredibly powerful planning tools that aren't necessarily connected to my manufacturing four point zero controls inside of facility.

And then, and probably more importantly, far less connected to the supplier based on their level of sophistication. And once you get into that same ecosystem using some of these advanced technologies, everybody's supply chain will lurch forward and also give the customer another entry point into influencing what's really important to them. And so unlocking those two pieces, suppliers and customers, across your integrated supply chain, I think will really just significantly change our overall experience level and be able to dial up what's important and dial down what's really less critical.

Yep. And I can't help but throwing it in. You said, what am I excited about? It's not necessarily gonna move the needle the most, but it's hard not to be excited about some of the stuff that's going on with the autonomous vehicles, the the delivery drones, the the humanoid robots. It's that like, I'm not trying to say, you know, this sci fi apocalypse picture that we have on the left, but seeing how some of that plays out, I've seen integrating ChatGPT now into some of the Tesla humanoid robots is a little scary, but also it's pretty interesting in the way that it's gonna completely free up some of the current labor to do higher value things.

And I know autonomous vehicles have been talked about for decades, but now that we're actually starting to see them adding value, seeing how that plays out is gonna be interesting to watch too.

Great. Thank you, guys. So we do have a couple questions from our audience. I'm gonna start with this one. Regarding use cases that you covered, What are the differentiators, excuse me, from firms that have leveraged AI better than others?

What do you want to take?

I'll have a quick go and it goes back to the human point.

The firms I have seen who are most effective at using AI are the ones who thought most critically about how to integrate that with their people.

Whether that's training managers to think about those aspects, whether it's interfaces between their scientists and their data scientists or their analysts or their operational research specialists and their operations team. People who've managed to glue that together and made sort of people and machine work in tandem, I've seen have been far more effective than those who have necessarily the most cutting edge technology, but have not thought about that organisational element.

That for me would be the main differentiator.

And the only add on I'd put there Stuart, past retail client implemented an advanced replenishment logic, AI driven, I'd argue is one of the best in the industry.

The human never trusted the machine. And so machine would make recommendations that intuitively didn't make sense and then they would quite literally manually adjust replenishment.

So they didn't unlock the efficiencies, they over clogged their supply chain with things that weren't spilling through and it's just an unfortunate really protraction of implementation or adoption process. But I think adding that part, have to trust it at some point, or you'll just get the value.

Yeah, mean, another similar anecdote I could add organization I saw who was using a very effective planning tool that was highlighting bottlenecks in their supply chain and problems they had. The problem was it was highlighting too many because they'd set the whole thing up badly. So they turned it off because that was proving distracting. It didn't it didn't solve the issue. It didn't resolve any of the problems that it was it was moving. It was it became just too too too overwhelming and too much of them to deal with. And if you don't as you say if people aren't thinking through how to do it or people aren't trusting and knowing when to go with their own judgment and when not to, it's not gonna be effective at all.

That's great. That leads us to our next question.

By when do we expect AI to be an integral part of most companies, SC management and even strategic planning?

I'm happy to say some words and would love to hear the others' thoughts. But I would say AI is already an integral part of most companies' supply chain management and strategic planning, for sure.

I think it varies, basically. AI is not one thing. It's not one feature. There's different types. So we talked about before how things like demand forecasting, network and routing optimization have been around for decades now. And they are like very well established actually across all sectors.

And so maybe you may not necessarily think of those as AI, but they are. Where I think you'll start to see a bit more differentiation though is on some of these edge cases. It may be something like your supply chain visibility that we talked about. It may be integrating some of the generative AI kind of new innovations is where it's gonna be a lot less commonplace. And that's where it's gonna come down more to the maturity of those technologies and those solutions, the maturity of the company, the kinda quality of the data that they're keeping in house, and ultimately the value that those solutions will bring to that specific company. Because the value is going to vary a lot from sector to sector, company to company.

And honestly, Nick, I don't know that I'd add anything else to that. I think that's spot on.

Wonderful.

Let's take two more questions.

Let's go with this one. Are there any AI driven programs today that you would recommend small manufacturing companies use to help manage their supply chain?

Let me start on that one and it gets me back to one of my favorite topics, which is really around the digital twin. If I'm a small manufacturing company, while I may not have a heavily complex distribution network, I am going to be very reliant on competitive input costs. My lead times are going to be heavily constrained based on where I source to balance cost versus availability. And then the third thing is, as we've all seen through COVID and then most recently the hostilities in the Middle East, all of that has a very, very real impact on lead times and shipping lanes.

And so if I'm a smaller manufacturer, I'm going to want to be able to model continuously what sort of impacts that's going to have on lead time and cost to serve, as I'm relying on supplies as well as ultimately logistics. And then I think as an add on to that, and then at what point do I need to start growing either from a brick and mortar perspective or a distribution node, having that digital twin in place will constantly let you challenge each of those components and let you make decisions faster, should make you a better caretaker of the business.

Thanks.

Let's go to our last question and I'm putting up on the screen your contact information. So anyone's interested in speaking further with us, please let us know. Our emails are below on that screen and we'll reach out with the presentation as well. But last question, any advice on how to ready the supply chain organization for AI, tech skills, leadership skills, cultural shifts? Any last thoughts on that sort of advice?

Yeah, I think it's such a good question. I fear that I'm not going to do it just as quickly.

But certainly, think just more widely, not just in supply chain, but everywhere. I am a big believer that getting that cultural shift and getting that thinking in place is one of the main unlocks of way to unlock value from AI. So I think it's exactly the right way to think about it.

Generally, my observation has been that trying to train people in tech skills is a bit of a bottomless pit. If I take the specific items included there, that there's always more tech that can be taught to people. And what you really want to achieve is a sense of ownership from the operations of what the technology is doing. And that can be achieved in a few ways.

That can be achieved by training people on some of the technical skills. That is true. But sometimes it's about actually relatively basic understanding how a programme or how an algorithm makes recommendations to engender confidence within the individual people in the ops team, that it's not just a black box to them and building that confidence. The other way is having people who can translate.

So you don't just have a team of people of say, operational researchers, you have some sort of internal client relationship between operational research and more wide operations to help act as that glue. Because I think ownership and belief that the system is working with them and not scary and outside of it is one of the biggest cultural shifts.

Beyond that, it would start to become quite specific to the to the organization itself. But I I think, you know, to the point you were making that fear is one of the biggest obstacles, and people often fear what they don't understand. So the more that it can feel transparent and people feel like they are masters of the tool rather than the tool being masters of them, then you're much better set up to allow for happy coexistence of machines and people.

Yeah. And a couple of things I would throw in on this point is I think get buy in from the top as much as possible. Implementing these types of solutions requires a lot of change.

And if you don't have all the people involved believing it and pulling in the same direction, it's not gonna achieve value and then it's gonna get forgotten about and things aren't gonna get integrated.

And the other one is start small, which isn't always exactly what you wanna hear. But being able to prove out value with smaller use cases is a really good way get buy in from senior stakeholders, get more leverage, get more leniency to do more things, and keep everyone kind of bought in. If you go for one big transformation, tons of dollars doesn't work, that's the best way to get a whole Get the whole thing going.

I was gonna say don't start your pitch with I'm gonna go implement AI to improve the business.

Yeah.

Because that definitionally, you're you you you're so general. There's so many solutions to get there. You gotta to Nick's point of start small. You gotta decide what problem you're trying to solve and then map what the right solution to do that is.

And then just kind of build on that. And you heard us talking earlier about some of our hopes and dreams for the supply chain of the future is more integration, more broad capability, that's still very much to come.

If you haven't gotten to the first level of demand planning or supply planning with AI driven, that's where to start, as an example.

Wonderful.

I feel like we could go on for another sixty minutes talking about next steps and advice, but let's have our audience reach out to you guys directly.

Looking forward to that. And I just wanna thank you all for taking time today and joining us and sharing your insights and looking forward to the next series of GenAI webinars from LEK. Thank you all.

Thank you all.

Thank you. Thanks, everyone.

Thanks. Talk soon.

Bye.