What we won’t do in the AI era

Transcript

Thank you. Thank you. Thank you very much, Holly, and for this wonderful introduction. Long titles.

Right? So, but, yeah, my passion is supply chain. So the focus on this speech is really on the supply chain and what artificial intelligence and automation can do in the area of supply chain management. I think it’s no news if I tell you.

Right? So the number one tool which is being used by supply chain professionals are Excel spreadsheets. Right? So it’s not really artificial intelligence, and it’s not automation.

It’s still Excel spreadsheets which are being used day in, day out in order to make decisions. So the question is, what do we need to do and what we can do and what will be actually the impact artificial intelligence will have in the area of supply chain? A big question. I think many working in the area of supply chain will really have a lot of focus on this question because they really think that this is the first time in the last twenty, thirty years that there is a technology out there which really can make a big, big difference.

How we run the business, how we run actually the entire value chain, how we run the supply chain. When we think about what’s going on in the area of supply chain, it’s also no news that the disruptions in supply chains are the norm and not the exceptions. We all have experienced it. Right?

So through the pandemic, with all the, disruptions, but it’s not just a pandemic. All the geopolitical issues also disrupt supply chains continuously and constantly every single day. You open up the newspaper, you see actually something which definitely will disrupt the supply chain. Right now it’s the hurricane Milton in Florida.

And interestingly enough, right, on the pandemic and also with the hurricanes, the first product which always runs out of stock is toilet paper.

So it’s quite interesting, right, if you think about it, why? Nobody knows why. But it is actually, again, proven now through the hurricanes of Milton and also Helene before in Florida, the first product which runs out is toilet paper. So there seems to be something going on in this.

But in general, right, the disruptions are the norm, and this is something where all the supply chain professionals really need to live through every day. And I must say, right, so, all the supply chain people in the room or maybe outside or your colleagues you’re working with in the supply chain area, they’re doing a fantastic job because as an end consumer, you hardly actually see what’s going on in the background in order to make sure that still the products are on the shelf, that you still can buy them, that actually the delivery works, even if it’s a little bit delayed from time to time.

But they are doing a fantastic job with spreadsheets mainly, and also with some back end systems they are using and applications.

But this is actually really something which needs to be acknowledged.

But then the question is, right, so also what is the number one priority which the business is driving right now?

Throughout the pandemic, when the pandemic started, the focus switched a lot to risk management, right? So how can I manage risk in supply chain? How can I make supply chains more agile, more resilient, and create a resilient and sustainable supply chain? That was the number one task.

Since late this beginning this year, this first priority has changed again. It’s it’s all about cost. Right? So I work with companies around the world and across all industries, and the first question is always, how can we take cost out of the supply chain?

Why is that? Actually, because of the economical environment, but then the second aspect is also throughout the pandemic, companies started to increase inventory levels. Right? So in order to make sure that we are not running out, you actually ordered more.

There was some hoarding going on, so there is some inventory levels which are still in in the system, which needs to which need to be reduced. So therefore, the number one topic is all about cost in the supply chain. But that’s not everything. Right?

So the chief supply chain officers really look for ways how to improve productivity, how they can also improve the collaboration and the network with their business partners. It’s very important. Supply chain is never an in house aspect. Right?

It’s not just within the enterprise. There you have, of course, the big value chain within the enterprise, but then all these companies work with thousands of business partners, thousands of suppliers, thousands of carriers, outsourcing partners, and you need to collaborate and coordinate all of them and with all of them. And that’s why supply chain is actually such a complex and a task which really needs some new thinking. And therefore, artificial intelligence is is really the first time, technology, which allows us to really tackle all of these different variables at the same time.

Right? So there’s not just a big hope, but I can I hope actually in the next couple of minutes, I can make the point, that this is really true? We also know that chief supply chain officers are not only thinking about productivity and, the the net promoter scores or the collaboration, how they can improve the collaboration, communication with their business partners. But they need to think about a lot of things.

Right? So they need to think about how can they support new business models, new operating models, which are being introduced all the time. They need to think about, of course, the agility. How can I onboard new suppliers in a much more agile way?

How can I use carriers in a different way? They need to think about the, aspect of labor, shortages.

In supply chain, surprisingly enough, there are actually all the time labor shortages. There are still actually truck drivers missing, around the world. In in the US, I think there are three hundred thousand missing. In Europe, I don’t know the latest number, but this is actually very true.

Right? So there is still a shortage always, still in in in the supply chain area. So the topic of automation comes up also all the time. When you go to Japan with the aging population in Japan, there you see it’s to an extreme where people say we need to actually find ways how we can automate the whole processes and the whole aspect, not just on the physical supply chain, but also on the digital supply chain.

Right? So this is the request and the task. When we think about supply chain, we also need to think about it end to end. We need to consider really the end to end value chain.

Right? So from the engineering, r and d processes, to supply chain planning, including the procurement, including manufacturing and logistics aspect, All of this actually is considered supply chain. Right? So all of this needs to run-in harmony.

Richard, I think, made a very, very good point because this is the traditional way of how companies are organized. Right? With AI, there is a possibility that this traditional way of organization will actually change.

Why is this actually a problem? Because this actually creates silos. Right? Everyone who works in a company knows when you have an engineering department and a manufacturing department, the handover process is broken. Right? So the engineering changes are a mess.

You hand over something, manufacturing decides to produce something different, there’s no feedback loop to engineering, end to end. Right? So you have all of that going on, but it’s not just between these two organisations. You have the same between the supply chain planning and the logistics area and the procurement area.

You always have different systems which are being used. There are twenty plus different systems being used in order to run this value chain. Engineering decides which systems they are using, and then you have the procurement area, they say I use these types of systems. So what do we do in order to bridge the gap between these different systems?

Sometimes it’s integration, technology, IT, But in many cases, it’s manual. Right? So you really retype information or you have spreadsheets in in between in order to do the and create this information flow. So therefore, one of the aspects we need to focus on when we think about automation and also the use of artificial intelligence is the first dimension I call it, is really thinking about how we can help companies to break down the silos between these organizations.

The second dimension to think about is all about the aspect of how they can communicate and collaborate more easily with all their business partners.

It’s about seventy five percent of the b2b business is supported with EDI functions, but the EDI function is only covering a few elements of the information which which you need to exchange.

The most and the majority of information exchange is emails and attachments. There are thousands of emails to every single department, to the procurement, logistics department, especially, which are being sent every single day. There are some there are people sitting there reading through these emails, they need to read through the attachments, they need to make sense out of it, and then start processing it in the back end systems. So why can’t we use large language models, so artificial intelligence, reading through these emails automatically, reading through the attachments, with that understanding this unstructured information, convert it into structured information, and process it in the background right away and automatically?

And this is exactly actually one way how to speed up the processes, how to actually make processes flow much faster, much more accurate, and with that, really create a much, much better communication collaboration. So at the end, we accept the way that emails and attachments are existing, and we want to have this communication going on, but we leverage now artificial intelligence in order to really read through that, and with that, really improving the process dramatically. So communication collaboration, a big topic actually, and I come back to that again, where companies are focusing right now when it comes to applying artificial intelligence right now.

The third dimension, I kind of referred to that already when it comes to automation and the AI aspect, is really about these different systems which are being used. I mentioned there are twenty plus different systems for the supply chain, for this value chain, and it’s all about orchestrating these different systems. Right? So the orchestration across those with the process flow and so on is not easy, and it’s actually quite tricky.

And here also for this orchestration, we can use AI agents. Right? So when orchestration, we can use AI agents. Right?

So when it comes to, artificial intelligence, I want to say the next big thing, which most companies in the tech area, at least, are starting to talking about, So just read articles from SAP or from Salesforce or these type of companies. It’s agentic process automation. What agentic process automation means is actually they are intelligent agents, and these intelligent agents are orchestrating the processes through these different systems which are being used in an automatic fashion. Right?

So there’s process orchestration and agents which are being coordinated in an intelligent way. This is what, agentic process automation is standing for and also where I will show you, how this will look like and what is what this can apply and and use how this can be used in the supply chain context. So the question is now also. Right?

So if we think about AI and and automation, what can be the impact? And we know that there is the cost pressure. We also know that companies really want to improve productivity, in in the company. We did actually benchmarking, and we did all kinds of calculations and, external, internal calculations and went through the model.

Right? And the value is actually really there. Right? If you start orchestrating AI and automation across this different value chain, you can massively increase, actually, the productivity.

Why is that? In supply chain, you have many manual processes.

You have thousands of document based processes. So think about the freight orders. Think about the the freight documents, the import export documents, the bill of lading, the invoicing, the procurement orders, the sales orders. All of that are documents.

Right? So this is where supply chain is living on. So lots of documents, and you have a lot of repetitive processes there. Because all three elements are true, the improvement you can achieve when you digitize business processes in the supply chain area are actually really high.

Right? So there is a study which says that it can go up to twenty percent of productivity for the knowledge workers. Right? Why is there a difference between the knowledge workers?

So white color, blue color. When you think about in supply chain, again, one phenomenon which is there, the physical supply chain has been automated quite a bit. Go in any shop floor or in a warehouse, you see actually the robotics driving around, you see the autonomous vehicles driving around, you see the conveyor belts, all of that. The physical supply chain is actually in a very good state when it comes to automation.

The digital supply chain is not. Right? So this is the funny thing, actually, that the physical is is actually really good. The the digital is not. So therefore, the potential is we need to bring the digital world closer to the physical supply chain, and with that, really increase the productivity.

How can you do that? Right? So there are, I want to say, four steps how companies really adopt this type of automation and artificial intelligence programs. The first one, and this is kind of the traditional one, is really much more on the robotics process automation.

What this means is really there are manual tasks, and you build software software robots which kind of emulate the person. Right? So instead of the person typing in all this data, the robot is typing in this data. That’s kind of what companies have established already when it comes to automation.

There’s not yet AI. Right? So this is really the robotics process automation aspect. The next thing that companies really started to work on already and have applied artificial intelligence successfully is in the communication and collaboration aspect.

This is where it comes to these emails, this unstructured information. This is where it comes to the attachments. Right? There is logic which allows to read through the emails.

The artificial intelligence algorithm understands the content, the context, the sentiment.

Typically, there are several aspects in an email which can be triggered in the background. Right? So think about you get an email, the ship cannot go through the Suez Canal because of the attacks at the Red Sea, so they need to go via the Horn of Africa. Right?

So you get an email now which says it takes eight days eight days longer, the cost go up, and the carbon footprint go up. So three different processes which need to be triggered in the back end based on this information. Then you typically have an attachment which which details up out the the eight days delay, then also what the cost numbers are, and also the carbon footprint information. And you take this information together, and you process it in the background.

You can also use generative AI in order to respond to this email. Right? So they can actually really use here a full loop type of process. And this is, I want to say, really the the focus what companies are doing right now with AI and where many companies have done it already successfully, and it has a huge impact in how communication, collaboration can be done.

The next level is really specific AI solutions. And here at Peak, right, a perfect example, Peak really stands for that and really provides perfect solutions here for these specialized AI solutions. I think, Richard mentioned the dynamic price quotation, artificial intelligence solution. There’s the inventory reorder point.

There are forecasting solutions end in end. Right? So there are quite a number of these solutions which are based on massive amounts of data, which are there because many companies invested in data lakes already and which can be used in order to train these models, and peak is perfect in order to do that. Right?

And really generates a lot of value. Right? If you think about the impact when you have a really, really good inventory management automation solution, what this has in order to drive and reduce cost is huge. Right?

So it’s just fantastic when you can do that.

The next level is then really the agentic process automation. Right? So with the agentic process automation, what that means, now I can orchestrate not only this one function of I want to do inventory optimization, but I use the entire end to end process. I can include all the different systems which are involved in order to provide the data and also process the data in the background.

And I use agents which intelligent intelligently understand what’s going on and how I can actually what what the impact will be with any disruption and coordinates that actually with the inventory optimization algorithm in an automatic fashion. So that will be the biggest impact, but this is where we, I think, want to say where we are very much in the starting point, right? So this is not yet really in real life being used broadly. Right?

So there are some examples, first examples, but this is very much the starting point. What everyone does is robotics process automation, where many companies have started to do it, is the intelligent document processing.

We have yeah. I want to say a few, maybe a little bit more than a few, started really to be successful is on the specialized AI solutions, and where the starting point is really the last area. But the value we can generate is really big. How can this be done? It’s really done much more with automation programs with a clear vision in mind. Right? I give you an example from a huge consumer products company in the US.

They want to introduce an autonomous demand planning process. Autonomous demand planning means let me first talk about the state they are in. Right? So they have ten thousand salespeople, they have five thousand supply chain planners. These ten thousand salespeople create every single one of them creates a forecast every every week.

Then you have the supply chain planners. They also consolidate the forecast, but also create forecasts, and they use statistical methods in order to improve the forecast. So they come up with a process every week which involves fifteen thousand people.

It takes not just fifteen thousand hours, actually, you can multiply that by two or three. And then also, you have statistical methods in addition to that. The forecast accuracy improvement over the last ten years was maybe zero point three percent. So with all this effort you put in there, the forecast is still typically biased and it is off by about twelve percent.

Right? So that’s kind of in the consumer products industry pretty much, the standard which didn’t change. So this company said, we want to not do that anymore. Right?

So we just discontinued this process. And it is a huge issue. Right? So they actually fight with the sales teams because the sales team says, well, well, we still want to continue the forecast.

But they say, no. We want to have actually AI coming up with the forecast itself. Right? So that’s it.

And we want to have the supply chain planning team focusing on forecast reduced from five thousand to five hundred.

Right? Think about it. Now you take out from fifteen thousand, you go to five hundred people who are dealing with the forecast every week. Right?

So what a difference this will make, and the forecast accuracy with artificial intelligence is actually a little bit better than it was before. It’s not worse. Right? So that’s kind of the point.

They say, well, think about what the savings of these fourteen thousand five hundred people in terms of time, they can work on completely different tasks, more value added tasks, instead of thinking what is the forecast for the next time. Right? So this is an example, very difficult to implement. Right?

So it’s very difficult because you have a lot of resistance from the teams which are involved, and especially also on the sales side. But this is an example of where autonomous demand planning can go. Right? And you can go now through all these different areas and can think about this this similar kind of processes when it comes to purchasing, when it comes to the purchase and invoice processes, when it comes to the transportation, transportation routing processes.

So you can actually really think about in the company where you could apply these type of of processes and really have these massive changes or impacts actually in the company.

The use cases which are being used and which are being adopted when you go through this value chain, and the number one area which is adopting these use cases is actually the procurement area. Right? So when you go through the different functions and organizations in the company, procurement tends to be the number one area which adopts automation and artificial intelligence.

The conversion from purchase orders and purchase requisitions into purchase orders where you have the internal checks and also the external checks, so where you say, I need to see is the supplier actually financially viable, or is is the owner of the supplier’s company on a blacklist? And all these type of checks, you can run actually automatically with an agent even proactively. Right? So you don’t need to wait until the purchase order check is done, but the agent can always and continuously check, and with that really update the information.

So purchasing, procurement, sourcing, really the areas which are adopting this first. Second is actually logistics. And logistics, the interesting thing here is it’s so document based. There you have all these freight documents, the import export documents, you have the bill of lading, all these kind of aspects.

So a perfect example with all these documents to really digitize them and process them in a very different way, getting more visibility in the supply chain, in the track and trace process, if you have the bill of lading correct, if this is actually always checked also against the invoices which you need to have and pay and so on. Right? So the second area, quality management is the third one in in production, which is, also typically based on seven seven different systems.

And you also have the information from your supplier about, the quality, information which you need to consider. So third is is actually production. But this is where you can see actually in the company where potentials are and also where the adoption is first, where it’s second, and also what the value drivers are for these areas. When it comes to, the AI aspects, I mentioned already that artificial intelligence has an impact already right now. Right? And, again, it’s very much on the intelligent document processing area, which typically includes all the documents, the content generation, but also, the content processing. So you have the impact in the order management area, you have the supplier communication, these emails I mentioned before, and also with the freight documents.

The second area, of course, and again, I want to highlight here, peak is actually the perfect example here. It’s really when it comes to specialized models for dynamic price automation for inventory, reward points and so on. Right? That’s the second area. The third one is really then the organic process automation, where we adjusted the starting point, and I also mentioned already a few examples here with the autonomous demand planning being one which crosses against many different systems where you can leverage these intelligent agents.

A typical pathway, right, is again along this line, the same line, where you first establish actually an automation platform, where you think about what kind of business processes are there, what should be the first process to adopt. The second aspect is then to adopt really the the document aspects because here these large language models, all of them are really good and can read these emails, can understand it, can understand the context, the sentiment, then applying the AI logic, the AI specialized solutions and and Adjentic.

So from an IT architecture perspective, this is nothing new. Right? So this is actually something which is not that difficult to establish. There is an application tier available and application layer available in every company.

Typically, you have an ERP environment with SAP or Oracle. Then you have the data lakes. Right? So with the information tier, where you typically have Google or Microsoft Azure or AWS, Oracle maybe from time to time.

And then you actually need a layer which is orchestrating across. This is where UiPath comes in. Right? So the company I am working for where we orchestrate information.

We are not replicating. We are orchestrating this information using the technology, applying this type of technology, which allows you to automate these processes with AI, in an end to end view. Right? So that’s kind of how, this this is coming together.

So to summarize, everything. Right? So when it comes to the supply chain area, this is the one area which is really disrupted, and the disruptions are staying, so disruptions are the norm. So that’s, I think, the first statement.

The second aspect is in in the supply chain area, we need to think about end to end. We can really think about processes which cross the internal organization, which connect the external organization, and cross all the systems which are being used. Right? So that’s a very important element.

You have really three dimensions coming together, and artificial intelligence and also automation really can make a big difference. I want to say most companies are really very much at the starting point when it comes to applying AI in a broad context, but doing that with AI and automation together has huge value, and we have shown that, and we have seen that in many different examples. So that was my little speech today. Thank you for listening.