The modern supply chain runs on data—31 billion transactions annually across the OpenText Business Network alone, representing roughly 10% of global GDP. Yet most organizations struggle to extract meaningful insights from this massive flow of information. At OpenText World 2025, supply chain leaders from Mars, Hershey's, and Lids gathered to explore how artificial intelligence is transforming EDI data from static transaction records into dynamic, predictive intelligence. This comprehensive keynote reveals how agentic AI, real-time anomaly detection, and supply chain orchestration are enabling companies to move from reactive problem-solving to proactive, autonomous operations.

Watch the keynote video or scroll down for the complete transcript.

Welcome to the AI-optimized supply chain revolution

Mark Morley (Product Marketing Team Lead, OpenText Business Network): Good afternoon, everyone. Thank you very much for attending this session. My name is Mark Morley. I look after the product marketing team within OpenText Business Network. We've got a jam-packed session this afternoon for you, including three fireside chats with some of our key customers. And I'm going to be joined in today's session—let's call this the three amigos. So we've got John Radko, Sushil Pancholi, and myself, who are going to be hosting this session with you this afternoon.

Now, we wanted to continue the theme of our keynotes that you've heard this morning and yesterday around data and the importance of data. And the theme I'm going to choose for this particular session is around the holiday season. Think about the data from a retail consumer goods perspective that companies are trying to gather and trying to derive insights from. And what we're going to be doing is looking at how data can be used to derive different types of insights. And as we go through the presentation, we'll highlight those.

We're going to go through, as I said, a couple of fireside chats, one with a VP of analytics and integration at Mars, but also with Hershey's and Lids. So three CPG companies, all have very different challenges at this time of year in terms of servicing their retail networks. And we're going to finish off with a demonstration of our Command Center. You've heard the words Command Center in the last keynote presentation. More from an OpenText point of view. But we wanted to show you a real application now in terms of how companies can leverage a Command Center to really monitor the health of their B2B platform and their environment. So more about that as we go through the demonstration later on.

The Data-Driven Holiday Season

Mark Morley: So in terms of the data-driven holiday season, we know that this is the busiest time of year for retailers and consumer goods companies in particular. But the insights that they derive at the moment in terms of their sales leading up to the holiday season, for Thanksgiving and for the holidays at the end of December, those insights are going to be used for next year's holiday season. So it's key to be able to retain those data insights and be able to derive information, actionable insights that companies can use to plan a better plan for next year's season.

So we're going to go through—we're going to talk about the importance of data and the context of retail supply chains, consumer goods companies. But when I was doing my research here, if you look at the total value of the end-of-year holiday season, it's $1.6 trillion. So that's a lot of revenue, potentially an opportunity that retailers and their respective supply chains are going after at this time of year.

Supply Chain Challenges

Mark Morley: However, there are challenges. We've seen disruptions occur in supply chains on a regular basis. I included the photograph there, the Evergreen container ship, because it's the most recognizable supply chain disruption over the last five years, especially in terms of bringing goods in from China to the Western markets.

But also cyber attacks, they're becoming more and more prevalent across today's global supply chains. And in the UK, one car manufacturer, Jaguar Land Rover, was impacted for almost a month where their production lines—they didn't produce any cars for a month. They also had the challenge that they couldn't actually register new cars because the cyber attack actually impacted their back-end retail systems as well. So security is job number one, certainly from an OpenText point of view, to help our customers secure their supply chains. We'll talk more about that a bit later on how we can help companies secure their supply chains. And we've got a couple of sessions directly after this, just in the next corridor, that will expand on how we can help companies secure their supply chains.

And then on the far right-hand side, global compliance. I think this is the bane in the life of many supply chain leaders, procurement leaders around the world, because not a day goes past where there is some form of new compliance standard. We know about electronic invoicing standards, for example, and mandates. We are fortunate to be able to support 50 different countries around the world from a compliance perspective. But also, you have other supply chain regulations, such as FSMA in North America for food traceability. There are digital product passports being introduced in Europe for electric vehicle battery traceability. They want the provenance of where those batteries are coming from. So there's a whole wealth of new compliance mandates that supply chain leaders have to embrace in order to comply with those regional government mandates.

Unwrapping the Gift of Data

Mark Morley: Key to this, again, staying with the theme of data, is unwrapping the gift of data. You'll see a slide later on that talks about the volume of transactions moving across our network. What if we could actually get deeper into those transaction flows and obtain insights that maybe companies haven't obtained before, and use that data from a predictive capability as well? So we're going to take a deeper dive into that in John's session in just a moment. But being able to unwrap, being able to mine that data and give companies the insights they need that they may never had before.

Now, where does that data come from? So in the funnel at the top of this graphic, whether it's logistics data, demand data, trading partner data, financial data, all of that information is feeding the business processes that you see at the bottom of the screen. So whether it's related to procure to pay, order to cash, whether it's reverse logistics, corporate to bank integration as well, we have access to a lot of data on our network. But being the ERP integration experts, we can also aggregate information from many different data feeds, whether it's from an ERP platform, not just SAP, but multiple ERP platforms, whether it's from a transport management system, a payment system, we can aggregate that information and derive insights on that information as well.

Agentic AI: Understanding the game-changer in supply chains

Mark Morley: But then we address the big elephant in the room, agentic AI. Now, when you think about the data that's on our network, imagine being able to apply agentic AI. And I would say imagine because you actually saw it live this morning with the error analysis. So that's just one perfect example whereby we are looking for agentic use cases for the data that's moving across our network. And really, it's about modernizing the EDI process. We're applying agentic AI methods. We're building out a new data platform that you'll hear about a bit later in this presentation to specifically help companies embrace this new technology.

So being able to make decisions based on contextual data, being able to adapt workflows dynamically, depending on the situation being experienced in the supply chain, being able to collaborate with agents. You heard about the agent-agent integration earlier on. So what about an agent from business network here at OpenText being able to talk to an agent on the SAP side, for example, whether it's Joule, so Joule and Aviator talking together and being able to exchange information? And then also being able to learn from the outcomes. How can we continuously improve the processes based on the output, based on the data being exchanged? And of course, the digital knowledge workers, they'll enable, eventually, those autonomous business systems.

Why EDI is the DNA of modern supply chains

Mark Morley: Now, we talked a moment ago about the agents, but as many of you know, EDI is a technology that's been around for more than 50 years. It's probably one of the longest-standing technologies in the IT market today. And here we are now talking about agentic AI combined with EDI data. I've always said to many of the customers that I present to that EDI is really the DNA of today's supply chains. It's something you can't rip out. It's going to be there for a long, long time. And so using these agents is going to allow companies to leverage that data in a much more powerful way.

The three or four examples shown on this slide with Trading Grid Aviator on the left-hand side—if you're a new user to Trading Grid, being able to have a conversation with your integration platform is key. We have a new enhancement to our IoT platform, being able to have a conversation with your devices to get the insights that you need in terms of what's happening, what's the temperature, what's the location of a shipment, for example. So we're trying to find different ways to embrace this agentic approach, and also generative AI across the various solutions that we have within business network.

And as you saw with the presentation this morning from Honda, that ability to identify errors in transaction flows really quickly and then being able to act on those as soon as possible is going to be key. And as you heard, the time that was saved in that particular example this morning.

How AI improves supply chain optimization: Key statistics

Mark Morley: I always like the benefits slide, the statistics slide. So what I found here in terms of the latest thinking from a manufacturing perspective: 49% of manufacturing companies said that agentic AI will improve supply chain optimization, orchestration, which is a topic I'll come on to later, and also end-to-end supply chain visibility.

51% of supply chain organizations are building autonomous agents already to detect and act on disruptions. We were talking about error detection this morning. But what about working with a partner that provides data feeds that interrogate your supply chain network and discovers some form of disruption, whether it's a container ship getting stuck, whether it's a weather event, being able to aggregate that type of information as well.

And then finally, 76% of supply chain execs said that agentic AI can improve supplier relationship management. And this is a topic that we're going to be covering in our next session in terms of improving supply chain collaboration. So even applying agentic AI there to improve trading partner onboarding, day-to-day collaboration, that is going to be an important area to focus on as well.

So from that, I'm going to hand across to my esteemed colleague John Radko, Senior Vice President Of Engineering at OpenText, to talk a bit more detail about where we're going with our AI strategy.

Building the OpenText AI data platform for supply chains

John Radko (OpenText Business Network): Let me add my welcome to Mark's. We really appreciate your being here. We appreciate your business. The chance to meet and talk to clients is so important. Over the next 10 minutes or so, I want to give you a sense of what we're doing in terms of our AI data platform within the business network. We can't cover everything, but I think what we can do is give you a direction and give you an idea of the kinds of things we're thinking about and considering.

First, I want to talk about a little bit where we've come from AI and why we view this current generation of AI as different. The first thing to realize is AI is not new. Artificial intelligence is already baked into many of our processes, certainly into manufacturing, into systems management, and largely in the form of machine learning. The key difference between predictive AI and generative AI is what I would call the barrier to entry.

Machine learning—and we do a lot of machine learning at OpenText. If anyone uses our capture capabilities, optical character recognition and image recognition involves a lot of machine learning. This is some of the most sophisticated software development we do. We use PhDs. There's a lot of math involved, all kinds of sophistication. Machine learning tends to be reserved for problems that have an extremely high payback because it's expensive to put solutions in place, similar to custom development that way.

Generative AI is the polar opposite. I would guess that many of you are using generative AI multiple times a day for even the smallest of tasks, because the barrier to entry is quite low, and currently the cost to us as consumers of it is quite low. So that is a big difference right there. It's a very approachable technology.

Understanding Agentic AI and why it matters for your supply chain

John Radko: Now let's talk a little bit about agentic. Agentic leverages that generative ability and then enables a degree of reasoning and the ability to use tools. So let's think about this for a minute. If we compare machine learning to generative AI, in generative AI you can just talk to the AI system as though it was another person, and it will use natural language, figure out what you wanted—what you want. You can converse back and forth to explain it. That's a really low barrier to entry. So that's one barrier reduced.

Agentic solves the scale barrier. What I mean by scale is many of what—many of the things we do in supply chain today, we know how we could do better if we had infinite resources. Think about it. If you had an infinite number of people, if there was no cost to adding staff or having people answering calls or looking at data, how much more would we do? So now think about agentic AI and its ability to lower the cost of doing certain tasks.

So on the far left for me, I guess the far left for you as well, we have predictive, where we reserve it for only the highest payback problems because the cost is so high. And then as you shift over to the right, we can use it for the most common of tasks. So if you think about the demo from the keynote this morning, what are we trying to save there? A couple hours of resolution time. Now that adds up, seriously. But that would not have been enough to launch a whole software development project for a year involving advanced AI techniques from the left. But it's a perfect application for the right.

And so one thing I want you to understand about my belief about AI and how it's going to impact us is nobody knows yet. And the reason I want to make that point—I was talking to some of my favorite customers. Story from B2B: EDI was rolled out to get rid of paper. That was the purpose of X12 and automation and VANs and all of that. So how many people in the audience are familiar with the advance ship notice? Okay. Everybody uses them, right? What piece of paper did the ASN get rid of? None, really, right? Right, Don? Because we couldn't do that before. There's no way you would have mailed someone a sheet of paper to tell them that you were going to send them a truck. Or what you're saying is it came with the truck.

But think about what the ASN enabled. It enabled a whole degree of efficiency because you could communicate ahead of time. But when they were starting to work on X12 and EDIFACT and TRADACOMS and all those, they didn't know that yet. And we're right there when it comes to applying AI to supply chain. The biggest impacts we're all going to feel when we look back in 10 years, the biggest impacts will not be stuff anyone is predicting right now. That's been true for the bar code, it's been true for the container, and it's been true for EDI. And I think this is as big a shift as those three. And I do not say that lightly. I think this is going to be a game changer because it's removing these barriers.

From traditional apps to adaptive intelligence: The AI evolution

John Radko: So let's talk about how this could take place in our industry. There's traditional applications. And again, high cost, high barrier to entry. We build them, right? These are fixed workflows. They have rigid user interfaces. You pay developers to do it. You get into upgrade cycles and training and all that kind of things.

Now we're entering into this phase of augmented applications, where you see the little AI chatbot. So it makes it easier. And this is the first place we're going to feel this experience really strongly, is it will enable us to use effectively capabilities that we have, but they're hard to use. A lot of the demos you'll see are chatbots actually in the background calling APIs and doing things for you.

But the third one, the adaptive intelligence, where the application starts to actually decompose and operate as a set of agents. And so when you're asking questions, what we call the orchestrator agent, the one that's actually getting the question is choosing what tools to use through a variety of protocols. That is going to completely change the economics of software development, of services development, because the whole proposition of our industry is that it is very expensive to write custom code and do things custom in the beginning.

So we buy packages of applications that embody best practices. AI is creating a new middle ground where it can essentially design a flow for you in response to your question. And right now we're using that to make the existing applications easier to use, to make data more available. But it is not going to stop. And people are going to—many of the people in this room are going to find new and creative ways to use this technology.

How the business network data platform powers AI insights

John Radko: So how are we approaching this? Because that's the theory. Now you know what I'm thinking about and what we're thinking about it. This picture represents—you know a build's coming. You can see already the spots there. But this picture represents what we do today. We do business integration. I think this is a strong capability of ours. We provide collaboration portals. We have connectivity on all levels from devices, from IoT sensors and trucks, all the way up to talking to SAP systems. And we do this for a number of supply chain communities.

What we're trying to do to augment, though, is to put in a data platform. When we talk about the OpenText AI data platform, there is also a business network data platform. The goal of this data platform is to extract business context from the traffic flowing over our network so that we can use it in order to deliver additional services and provide additional benefits. And that will come in the upper layer, which is our AI-powered experience.

So this is where the agents Mark had in his pitch or in his deck a little bit earlier come from. So you take the data out, and the AI agents can work on it. We're doing this right now based on visibility data. If anyone's familiar with TG Insights, also known as Lens, we can use metadata. And that's how we do things like anomaly detection. Like, oh, usually you get a lot of orders from this partner. Now you're only getting a few. But the real power comes from looking inside the transactions, understanding the data in context, and that's our goal for the data platform.

And once we get this rolled out, and that is going to be a two to six quarter odyssey over the next 6 to 18 months, we will be able to engage in new use cases on a regular basis, and you will be able to create agents, just like Savinay Berry, our CPO, was saying, that we haven't even envisioned yet. And that, if you remember from the previous slide, that third category of augmented applications and adaptive, that's when the reality comes in place. So we create capabilities that you can address through agents.

OpenText Trading Grid Command Center: Your AI-powered control tower

John Radko: So I'm going to talk a little bit specifically about Opentext Trading Grid Command Center, which is the first place where we're applying this. And the idea is to pull data that we see from what's happening in your various B2B communities and put it initially into a view. Our original concept for Command Center was more of a dashboard kind of view, and then make the underlying data addressable via AI. That's the beginning. And then allow you to post agents against it so that you don't even have to be looking at it.

So how's this work? Well, essentially, at the base you have the business process data. And then it gets aggregated. And then it goes into that data platform category. And today, what we've been doing is putting monitors out. Integration monitors launch. That's where the anomaly detection is. There's performance monitor and some custom monitors we built. But the big change is the business intelligence UI.

So our intention is to make this data available through a UI and then through MCP interfaces against APIs that our agents first and then your agents second will be able to go after. Because we believe in the beginning, customers will like the dashboards and the ability to do a chat interface to ask questions. But we know all of you. We've all been together for a long time. You'll rapidly tire of that. And I've had a number of customer meetings. And there's a question that keeps coming up in every customer meeting: When are you going to automate this for me?

So the design already is aimed at enabling that automation. So the goal here, if you think to that data platform, it's to extract the business data from any platform we use, put it into a format where it has context and we understand it, and be able to allow queries and analytics against it initially, and then agentic agents to go against it, or agentic AI agents to go against it.

Using knowledge graphs to unlock B2B data intelligence

John Radko: So in order to do this, the basic mechanics, as you'd expect, is we have to collect the data and standardize it and then make it accessible through UIs. So the focus here is really on enabling access to data for people, yes, but also for agents. The data will have a context. There's a new phrase, at least for BN. It's not new for our industry that we're talking about, which is the knowledge graph. The knowledge graph—every B2B community has a knowledge graph. You have partners. You have transaction sets. There's business processes that fit in. We're going to consciously construct this knowledge graph.

The reason is because as we're looking at the data, today you can see a lot of your data through visibility tools, active documents, Lens—excuse me, Trading Grid Insights—but it tends to think about it as a document. We've done some features to allow you to see it in context, but I want you to think about seeing it first in context. So going into integration monitor, looking at your community, drilling into a particular partner that you're working with, seeing the business processes that they're working with, and then going in and being able to track the transaction flows and the performance across those flows. So starting first from a business mindset and then going to a data mindset, as opposed to the opposite.

But in order to do that, we need to understand the context in which every single transaction operates. And that's the purpose of the knowledge graph. So the knowledge graph is the network of relationships among all the pieces of information within your system. And this will all be based on the OpenText AI data platform that you've heard about yesterday from Savinay and Eric, and today a little bit more this morning.

Extracting powerful insights from EDI data with AI

John Radko: Now, we think AI can help obtain powerful insights from EDI data. What we're looking for and what we're hoping to partner with the people in this room is on our sample use cases, areas where you want to get more data than we can currently provide you. Now, traditionally, the most powerful usage is not predicted. As I said earlier, no one predicted the ASN. I think when people put the barcode on the first pack of Wrigley's Gum, they were thinking they'd speed checkout. I didn't think they could predict in advance how much effect it would have on inventory optimization.

So I believe we don't know where the best use cases are coming from. And so all of us need to adopt what's called a beginner's mindset. There's a quote I really like. It says, "In the beginner's mind, there are many possibilities. In the expert's mind, there are few." We're going to adopt a beginner's mindset and go on a listening tour. And what we want to understand is, what are the pain points? What are the areas you want us to go after?

I'm not sure—American Honda in the room somewhere from the demo this morning? Okay, awesome. Thank you. Thanks for doing that, by the way. That's a case in point. If you had asked me what would matter to Honda in this case, I would not have gotten it right, and that's not the end. I know that's not the end of your wish list, but so we want to do that with as many customers as possible.

So as we build out the data platform, we want to help deliver on use cases in two fronts. What do you want us to give you? What you want us to do for you as an agent? And what capabilities do you want completely independent from us? So these are, what tools and capabilities do you expect from us? And what empowerment do you want in terms of self-service and capabilities? So that's where we hope to partner with all of our customers as we're rolling out data platform and the agentic framework for business network.

Real-world AI use case: Automating deductions management

John Radko: So I'm going to give you one example of one we're working on and haven't built yet. This is a traditional business flow, and this is how we see it today, a series of documents. In the data platform, we get all of the detail. So picture all of this stored in a structured way so we can see it in a data platform. And now I need you to leap forward a few quarters with me in terms of AI agents. And I want you to envision an agent that is capable of consuming compliance rules from a retailer in this case, and understanding how to turn those compliance rules that have been shared into actions and analysis.

And now what you'll see is the deductions agent is actually looking across these and determining if a deduction is going to apply or if we're going to get our invoice paid in full. I want to return where I started, to the beginning of my talk. We can do this today. We do this today in Trading Grid Intelligence. But it is hard to scale. The only folks who will take advantage of in this room are people that have serious deductions challenges because they operate at high scale.

Imagine, though, if we can lower the effort down, if we can automate the configuration and automate the resolution, how much wider that scale will go. And that is the trend, when we talk about the barcode or EDI or the container. The trend is you have a technological change intended to address a problem. And then as a result of its effect, it widens the effect out to where there's a huge scale. And we think this is just one example of what's going to happen.

AI go-to-market strategy and roadmap

John Radko: So this is our horizon view currently of our AI go-to-market strategy. And for those of you that have paid careful attention to me, you'll not be surprised that I suspect this is going to change because we don't know what we're going to find. But it's a journey we're on together. And I did want you to know that though I think our direction in the journey will change, we do have a map of where we're trying to go together with you, our customers. So everything we do is aimed at the people in this room and your colleagues around the world. If this technology is not adopted and used at scale, we will not have been successful.

So this is where you're going. You feel it. We've got a lot of conversational AI capabilities rolling out in our latest releases. And then Horizon 2 is where we're starting to apply AI into our migration tooling. I announced at the CAB, we have a six-year program that we're embarking on to enable the migration of MS Classic customers to the Trading Grid. I believe this is the first event, the first OpenText World, where we have actually had an answer for the question that we've been asked so many times: when are we going to do this? And AI is going to play a big role. I would bet that we will see an acceleration there, but we have a current calendar plan to back that up. But I'm expecting we're going to see an acceleration as we dive in. I think we're going to make faster progress than we expect.

Then you'll see Horizon 3 safely two years out. That's a pretty safe window for us, where we're talking about AI and software platform for autonomous operations. It is very scary. It's interesting, in the CAB meeting we held this week, a customer asked us if we can stop traffic. And this was a legitimate reason to stop traffic, because it was such a high volume of traffic, it could actually injure their systems.

In Trading Grid Intelligence, we have a capability called quarantine, where we can intercept a document moving across the network until the customer chooses to release it. It is very rarely used. So the challenge for the two-year mark is, can we get good enough? Can the technology get good enough, and the trust get high enough that you would actually turn the decisions as to whether the traffic comes to you or not to autopilot?

And there's an easy metaphor, I think, for most of us in our lives. Most of our cars or many of you are driving cars that have some kind of assistive steering. You have adaptive cruise control where it adjusts the speed, or maybe the steering where it'll try and keep you in the lane. But I'm curious if anybody in the room has a car that drives itself, and you're not in control of it yet. Okay. But how many people think we will get there in the next 10 years, where you would have a car—okay. So if we can do that with driving, I've got to believe we can do it with supply chain. So that's the goal to where it's an exception-only rule.

So that gives you an idea of what we're thinking, where we're going, what's driving our technological strategy. What I would like to do now is kind of come back to Earth and invite Shailesh Jha, a good customer of ours from Mars. He's the VP of AI data analytics integration at Mars, and hear about it from the customer side. Please help me welcome Shailesh.

Mars case study: Building AI maturity in a 115-year-old company

John Radko: So Shailesh, if you don't mind, before we start in on the regular, tell us a little bit about your role at Mars and just your background a bit.

Shailesh Jha (VP of AI Data Analytics and Integration, Mars): Sure. Shailesh Jha. I've been with Mars for about 18 months now. I'm the vice president for data analytics and integrations. It kind of sounds that I do every work there, but I do not. Yeah, and prior to Mars, I worked with Amazon for about eight years, both in advertising and their retail function. And at that time too, we used Contivo, one of your products.

John Radko: Absolutely.

Shailesh Jha: And before that, I was working at Tibco, probably a competitor in some ways, but yeah. So been a builder most of my life, and now I'm on the other side buying technology.

John Radko: Okay. Now, one of the themes we keep coming back to is that data is really the foundation for so many of these products. So can you start by telling us about how you've built a foundation of trusted, high quality data to power your AI analytics strategy?

Shailesh Jha: Right, and it's important to understand the models in the context for that. So Mars, not a lot of people know about Mars because it's a privately family-owned company. It's not listed. So Mars is a 115-year-old company. And within Mars, we have three major segments—snacking, food, and pet care. Not a lot of people know, but we are the largest veterinary health provider in North America. So it's a diverse set of business grown over the last 115 years. So scale and complexity come naturally to us, that part of the system.

And the way we have started addressing that is through a layered approach, building that high quality data. And the foundation of trust is clearly built around the five principles. So at Mars, we have our five principles, which are quality—starts with quality that addresses the data part—and responsibility, where we have been trying to achieve a higher level of data stewardship, responsibility of system, assigning that. The third thing is mutuality. That's the third key principle. And in that what we want to make sure is that the data is shared with every associate partner in the right way, in the right context. Make that available to them.

And yeah, and fourth is efficiency. Cost is always on our minds, being a CPG company. So we want to do it as efficiently as possible. We want to minimize redundancy in our storage, in our compute, in our networks. And this is how we are building it. And so far, what we have done is we've strengthened our data foundations, we've standardized our governance models, we have standardized our data sharing through API gateways. We've started defining canonical data models to share that across multiple systems.

John Radko: That's excellent. Thank you. Mars is a very sophisticated operation, and I'm curious to know, where would you rate yourself on the journey in terms of maturity and evolution? And maybe describe your organization's current level of AI maturity, and what key milestones or inflection points have shaped your journey so far.

Shailesh Jha: Right, so like I said, in Mars, you can find a team at every level of journey, because we are so diverse in the parts of our veterinary health which are far advanced, where we are trying to bring AI for improving pet health. There are parts of business which are very traditional, and there have been in the system for over 100 years. So if you're looking for me for a score, that okay, give me a number, we probably are not looking at AI maturity in that way. We are looking at it as in terms of where does it fit best?

So rather than chasing the hype, we are evolving with intent. And by that I mean, whatever tools are necessary for that line of business for that marketplace, we are ensuring that we provide that to the business.

John Radko: Okay. I really am going to go off script slightly for a second here. I really like that notion of intent, because when we have a new technology like AI and it's very exciting and dramatic, it can be tempting to just chase the fads. And this focus on outcomes and achieving goals and that and the recognition that it's different within each line of business, I think, is very powerful.

So one question, because I talked a lot in my previous section about scale, because in B2B, scale always becomes ultimately the biggest challenge. How are you seeing—what do you see as the barriers to scaling AI company-wide?

Shailesh Jha: Right, right, that's an interesting one. So I'm going to give you a different answer, probably a little provocative for this room. I think the biggest challenge or the biggest barrier for AI adoption, I would say, is the pace of AI. And what I mean by that is, just look at it, in the last two years, we have gone from experimenting with generative AI tools to now talking about agentic AI. We're talking about human outside or human in the loop. Now we are all comfortable with automation taking over completely.

So I think AI is moving at such a rapid pace that it's very difficult from a capability standpoint to baseline it. So as enterprise, when we see an AI partner come up, okay, this is my solution, take it, we are looking at, okay, in three months, you're going to come to me with a very different solution, a different capability. Like you, we don't know where it's going to land. And from enterprises, that long-term ROI is always the key. When we are looking at purchasing a solution, the long-term commitment to it.

And that's where it's one of my biggest learnings. And as I moved from tech to CPG, they are very different. Whereas in tech, fail fast is a culture, in CPG, it's buy and commit. And I think that's where the pace—and the second part of it, equally important is the cost angle. With these AI systems, what we are seeing is there's increasingly a debt. And because they are so discounted right now, because you've seen each one of them is funding the other, so that there is a debt, which is floating around and somebody has to pay it, and it's going to be the customer at the end.

So we are very careful of not—and I think your CTO called it out very well with the agent sprawling concept, and I think we need to be careful of that. So I think the biggest barriers is the more our companies can mature, the capabilities standardize it, because enterprises are traditionally risk averse.

John Radko: I think you'reabsolutely right. And I think there's a bit of fear of missing out in the AI industry that may be leading us to jump from type to type. So let's talk a little bit about systems and data integration, though. Which enterprise systems and data sources do you think are most critical right now for feeding your AI ecosystem? And how are you ensuring interoperability across them?

Shailesh Jha: So being a CPG company, we've got the standard ERPs, we've got supply chain systems, the manufacturing. We also have a lot of point of sale integrations with our partners, retail partners. So obviously, these are very important systems that are—plus we've got trading partner networks like yours, OpenText, a very important system for us to integrate.

So to me, these are the important systems, and how are we ensuring interoperability. So a few things that our team is building on is one thing we started on a journey in the last one year we've picked up on that is defining standard data domains. And then because our business is so diverse, we're trying to find the common denominator that cuts across all of them. And with that, we are also defining a canonical data models that support, irrespective whether it's data or integrations, we're using those same set of schemas to make sure.

The third thing that we are doing is also building on our API gateway so that the data distribution is standardized. One thing that we have today is multiple gateways. So we want to standardize into a single place where people can consume data from or integrate into the other system. And the last thing is cataloging, which also goes hand in hand with API gateways, ensuring that all our data and API products and ML and AI products going forward are cataloged in a single place. So we have our partners of choice there that we are working with to build out that infrastructure.

John Radko: Well, that's very interesting. Yeah, we're trying to build a catalog, at least for our own AI capabilities, because similar to what you said earlier about the pace of change, even within an organization, there can be AI-based tools that exist that employees aren't even aware of just because news travels more traditionally.

Now, next generation AI, how specifically are you leveraging generative or agentic AI within your organization, and what capabilities are efficiencies are emerging as a result?

Shailesh Jha: So I think for them, for generative AI—so we started our journey. So one thing that we started building was we started realizing that there's too many of these LLMs and models like you have from Anthropic, Google, many other vendors playing it. So we wanted to make sure that, first off, before jumping the gun, we had to define a responsible AI framework for it to operate in. And then what we did was, just last year, we trained about 18,000 of our associates on responsible AI. So we had an internal certification. So people, when they start interacting with them, we know what they are.

And we have our own internal system, which uses a complex algorithm to select which is the right LLM for the use. So that's how we're using generative AI. We're using it across our marketing function. We're using it across finance. We're using a lot in conversational reporting, as we're calling it. I think you showed that in your slides too.

Agentic is something that we are looking at very closely from replacing the traditional RPA, Robotic Process Automation. And we are also seeing some good results there as we move towards agentic in that space.

John Radko: That's exciting. I look forward to continuing the conversation and hearing more. One last thing I'd like to ask is, what guidance would you share for your peers? The truth is I know you came from tech, but in your current role, you have more in common with most of the people in this room than I do. And I think they'd be interested in hearing your suggestions as to what they should be doing.

Shailesh Jha: Right, right. So if you're beginning your journey with AI, I think one of my advice and one of my learnings is don't start with AI. It's the shiny object, but like to your point too, which you mentioned, that don't chase it. It should be seen as a last mile. It's the foundations that are really important to it. The road to the AI and underneath is your data, your governance, your stewardship, how are you going to distribute it, responsible AI, because this will sprawl out very quickly. You'll see our assets that can get into the public domain pretty fast, your IPs.

So as enterprises, I would say if you're starting right now, start with the foundations and then use AI for your last mile where the true value of AI is seen today in automation tasks, processes, productivity.

John Radko: Okay. Shailesh, thank you so much. I appreciate it.

Shailesh Jha: Thank you.

John Radko: Please thank Shailesh.

What I'd like to do now is invite Mark back up to share a little bit more about the plans we've got going forward around our roadmaps and innovations. Mark.

The complete OpenText Business Network platform

Mark Morley: Thanks, John. So I'm going to be the warm-up act for my colleague Sushil in just one moment. But what I wanted to do was provide a reminder of our portfolio today, where it stands. So we have a number of key capabilities within business network over and above our core integration capabilities that you see in the top right-hand side there.

Supply Chain Orchestration, which we'll expand on in just one moment. I've got a separate slide to cover that. Supply chain insights. We've heard about integration monitor. You'll see a live demonstration just one moment. But this segment is really around our Command Center. How can we get different insights to support a company and the changing dynamics of the market that they're working in?

Supply chain traceability, some of you may not realize we actually have an IoT platform. We have a QR code-based solution as well for asset traceability. So we've expanded our portfolio through the various acquisitions that we've made over the last 10 years.

Secure collaboration, there is a dedicated session straight after this one that looks at secure supply chain collaboration and what it takes to actually ensure that everyone in the supply chain has a digital identity, they have the right permissions to be able to get access to the right information at the right time. And my colleague Vesna is going to be leading that session straight after this one.

And then finally on the industry applications, a large part of our business here within business network relates to financial services. So we have, for example, integration—we are a SWIFT bureau. We have a number of corporate bank solutions as well to support bank connectivity. But also on the corporate side, being able to support treasury management and the financial transactions that have to be exchanged between a corporate and a bank and various clients. So we have some of the world's largest banks connected to our network. We're not just supply chain focused, which is a key point.

OpenText named leader in IDC MarketScape for fourth consecutive year

Mark Morley: I wanted to share this slide. This is literally hot off the press. As of about an hour ago, IDC informed us that they're going to be publishing their new MarketScape on Friday this week. I managed to get a sneak peek of the actual layout here, which you can see. So thanks very much to Simon Ellis, who's sitting in the audience here, for providing this graphic for me to be able to share in this presentation.

So we are a fourth time leader for the fourth consecutive MarketScape with IDC. And you can see the quote there in terms of OpenText Business Network, the ability to deliver a secure, compliant and future-ready solutions backed by unmatched professional services and innovation positions OpenText as a trusted partner for organizations around the world.

So as I said, that MarketScape is going to be published on Friday, and we will make sure as a follow-up to everyone that's in this room, that you'll get a copy of the MarketScape so you can read the assessment and how we compare to other vendors.

31 billion transactions: The scale of OpenText Business Network

Mark Morley: I talked right at the very beginning about the importance of data and the volume of transactions moving across our network. So we're sitting on a very rich seam of data across our global business network. 31 billion transactions being exchanged on an annualized basis. If you imagine each one of those transactions having 80 data fields, that equates to 2.5 trillion pieces of data moving across our network to support our customers.

And if you take that one step further, if you look at the average value of the transaction moving across our network, that represents around 11 trillion in network commerce. That equates to 10% of the world's GDP. So I think this slide, if nothing else, actually illustrates the importance and the reliance that our customers have on our network and being able to make those supply chains operate seamlessly and reliably.

What is supply chain orchestration and why does it matter?

Mark Morley: I mentioned about supply chain orchestration just a moment ago. And I've been doing my own research in terms of where we think the market is going to be going. I know that analyst firms are looking at this space as well. And when you look at the IDC MarketScape that I just showed you, if you look at previous magic quadrants, MarketScapes and everything else that the analyst firms have been producing over the last few years, there's been a strong focus on B2B integration, iPaaS platforms, for example.

But I think with the introduction of AI, there's an opportunity to look at things slightly differently. And when you combine and look at the capabilities on those bullets, whether it's end-to-end visibility, multi-enterprise integration, AI-driven decision intelligence, this is exactly the direction that we're going with the roadmap that you're going to see in just one moment, and the capabilities that we're going to be providing to our customers in the future.

So supply chain orchestration, the definition given here, is really the real-time alignment of people, processes, and systems and partners to ensure that supply, demand, logistics activities operate in harmony, a bit like running an orchestra, dynamically adjusting to disruptions, changing customer needs, new business opportunities, and mentioned the disruptions, the supply chain disruptions that are highlighted right at the very beginning of this presentation.

So with that, I'm going to hand across to my colleague Sushil Pancholi, VP of Product Management at OpenText, who's going to talk a bit more detail about where we're going with the roadmap and talk a bit more detail around the integration monitor that we've introduced as well.

What's next: OpenText Business Network product roadmap

Sushil Pancholi (OpenText Business Network): So hi, everyone. And I also want to say thank you very much for the opportunity to share our roadmap plans with you. We'd love to get feedback from you at our booth later on in the sessions. And let me walk you through some of the highlights from this roadmap.

So during the week, you've been hearing a lot about our plans on AI and data analytics, and for sure we're very excited about that. But I also wanted to reassure you that we haven't taken our eye off our core B2B integration capabilities, which is why most of you are customers and in the room today. So let me cover some of those things that we want to continue pressing on our advantage on these new capabilities.

So I'll go towards the middle of this slide. You can see we're planning on commercializing Trading Grid MFT as a service. So this is something that we're planning on commercializing in 26.01, and it's the secure efficient movement of massive files. In the past, we've seen customers and prospects have to go to alternate providers for this because they typically don't see us in our platform as a solution for this. This now is something that we're going to be offering on Trading Grid, and the advantage is obvious. You get a single platform for both your full service managed services, B2B integration, as well as your MFT traffic, and you also get to take advantage of all the innovations that we're talking about on Trading Grid, Command Center, AI-enabled analytics, self-service capabilities. All of those things will be available on the MFT solution as well.

Next thing I want to point out is e-invoicing compliance. So many of our customers are being faced with that challenge. International companies like you have to comply to mandates, not just one country. This is expanding almost all of Europe, Latin America, APAC as well. So as Mark had mentioned earlier, we support mandates from 50 plus countries, and we're going to continue adding that. Unlike regional providers who typically support one to, say, two or three countries, we can support basically the whole geo areas that you need to get covered.

So one of the big ones that's going to go into effect is in France in September 2026. Not only are we in certification testing for France already, but we're going to continue adding other countries as well to our capabilities.

The other thing I want to point out is our tight integration with API-enabled integrations to ERPs. So this is something that we have enabled for off-the-shelf support for order to cash and purchase to pay processes, business processes for major ERPs like SAP Public Cloud, SAP S/4HANA Public Cloud, NetSuite Dynamics. So customers, if they're using any of those ERPs, they can use our off-the-shelf adapters and be ready to integrate and support those business processes without too much hand-holding and EDI expertise.

New AI analytics features in Command Center

Sushil Pancholi: Okay, now back to Command Center AI analytics. So here we're—anomaly detection for Command Center is something that we're commercializing soft release this quarter. And we're going to have an official rollout for it in 26.01. So the way anomaly detection works is we're going to have a baseline developed for your traffic using three to six months of data. The more data we have, the better the predictive capabilities and detection capabilities become. But basically, we'll develop a baseline and then have a threshold that we will define. And anytime those thresholds are crossed, you will get notifications for it, and you can take proactive action on it.

Okay, next is forecasting of transaction volumes. So the concept is very similar to anomaly detection. We, again, are developing a baseline, and then giving you the ability for operational purposes for planning to ensure that you know what's coming from your partners in terms of how to plan for your ERP systems, your internal systems, so that you can be ready for whatever changes there are in EDI traffic.

Okay, Mark touched a little bit on the IoT side. We're also adding intelligent capabilities, Aviator capabilities on our IoT platform. Here you can actually get real-time notifications and real-time visibility in the physical world. Most of the capabilities that we provide in terms of status is coming from—oops, it's changed on its own.

Okay, let me talk about the IoT Aviator here. On Aviator here, you can ask questions such as, what is going on with my partners? Who are my typical partners exchanging 850s breakdown by TPs that are active versus inactive. You can also do comparison of transaction volumes of given TPs between specific time frame. Like, you can say how many TPs sent me purchase orders in October '24? And what was the change in October '25? So those are some of the capabilities that you can interactively ask via IoT Aviator and get insights that way.

Okay, now I'm going to turn it over to John, and he's going to bring on two other customers of ours.

Hershey and Lids: Using EDI data for supply chain visibility

John Radko: Thanks, Sushil. Please help me welcome Scott Stevens and Jafrina Jabin from Hershey's and Lids, respectively.

So thank you so much for joining me. We really appreciate your participation here. Can we have one more round of applause for them? Because they have been troopers about the preparation and getting ready. And I'm really excited to have them share their information with everybody. But before we do that, Jafrina and Scott, what I'd like to do is have you set the scene. Can you start by telling us a bit about your company and the role your EDI infrastructure plays in connecting to your supply chain? So, Jafrina, let's start with you.

Jafrina Jabin (EDI Analyst, Lids): Thank you. My name is Jafrina Jabin, as you already know. I'm from Lids. I'm working as an EDI analyst at Lids. So at Lids, we have supply chain, which is like a diverse network of domestic and international vendors, retail partners, and logistics providers. So through OpenText, which is our main EDI provider, we connect all of these parties together so that all of these entities can communicate with each other. My role actually is to monitor all of these transactions and work on the errors, so that the operation can run smoothly.

John Radko: Okay, thank you. Scott.

Scott Stevens (EDI Analyst - Hershey's): Yes, I don't think there's a lot that I need to say about Hershey. I think most people know them. But I do want to add, we are more than just candy. We are also branched off into a salty snacks division. I'm sure you've probably seen some of that Dot's Pretzels. I don't know whether you knew that that was ours, as well as things like SkinnyPop. And I'm actually—just found out before I came in here today, we just finalized an acquisition, and we're going to be providing a product called Lesser Evil, which is an organically made product. So that's all fresh off the press today.

So from Hershey perspective, we do all of the EDI transactions that everybody else do. I mean, we have what I call the core four, which are the purchase orders, purchase order acknowledgments, ASN, and invoices. But we also have other transactions. We have all of our warehouses that we have to do transactions with. We have co-mans and co-packs, co-manufacturers and co-packers that do work for us as well as banking. So we've got a very diversified slate of EDI transactions.

From an OpenText perspective with Hershey, we brought them on when Hershey transferred over or migrated from our ECC environment to our S4 environment. So we're fully on board with you now, and you guys are doing a great job, and I'm enjoying it.

John Radko: Thank you. We're enjoying having you. So, Jafrina I'll start with you. One of the things we wanted to explore is the insights we can get from data. So what insights are you currently able to extract from your EDI transaction flows, and how does that relate to the key supply chain KPIs you care the most about?

Jafrina Jabin: Great question. So from all of the EDI transactions that we are using, we extract insights that directly influence decision making and our supply chain performance. So some of the KPIs that we extract are purchase order accuracy, ASN timeliness and completeness, fill rates, back order trends. If we are getting back orders, we monitor the trends. Then we also monitor EDI compliance and recurring error patterns because if ASN is late, we get a lot of errors, especially from the location from different type of error debt. So we also monitor those. And obviously, through the vendor management platform from OpenText, we monitor invoice mismatching and the financial alignment. So tracking all these metrics, that really helps to run our operation smoothly.

John Radko: Okay, thank you. Sticking with the analytics but branching into AI a bit, Scott, how's your organization using—how's Hershey's using analytics today? And have you begun exploring the use of AI to enhance visibility or performance?

Scott Stevens: Well, that was a great question, and from an EDI perspective, and I've been in EDI for multiple decades, I won't say how long, but multiple decades. And we're not getting into the AI as much in EDI yet. However, I'm relying on some people in this room to assist us as we go through the Command Center. But if I look at the business itself, I know one of the biggest things that I found was that we're using SAP data sphere in the form of dashboards and predictive simulations to try to help the business come up with new ideas and thoughts.

John Radko: Okay. Maybe we could dive a little bit deeper into use cases, where you're applying that for your supply chain operations, for example, maybe forecasting supplier collaboration or exception management.

Scott Stevens: Absolutely. And I'm going to be honest with this, like I said, I've been in EDI and not in the business world, so I didn't know about it. So I actually used AI to find out what Hershey's doing with some of this AI. So that's pretty neat. But what we found is that we've got multiple things going on in their supply chain and manufacturing. They've actually spent—I think I remember, it was $250 million to enhance that. And it's to provide robotics and automation and improving the efficiency and the manufacturing optimizations.

We're also using it within our marketing and consumer engagements. We've got areas where we're targeting our customers using targeting campaigns. We've got where it's giving us personalized recommendations and also to assist us with customer service. And also in areas of product development, I know they're using it there, because they're analyzing the market trends and what the preferences are. So they're using AI for all that as well.

John Radko: Okay, that's very exciting, actually. I think in their introductions, they were both a bit modest because they're operating in one of the most challenging spaces. I think many of the organizations that you're exchanging data with are among the most demanding and exacting, and that's because it's such a changing and crazy world.

So last content question, Jafrina, before I ask the two of you for some insights. In terms of visibility and orchestration, do you operate a supply chain control tower or a command center that consolidates EDI and other data sources to provide visibility across your network?

Jafrina Jabin: So we don't use a specific command center because we use multiple platforms. So we have OpenText Trading Grid, then Active Intelligence, and our in-house BI reporting platform to get a unified view across vendors, DC operations, and all the stores. We are using OpenText Trading Grid for monitoring all the transactions, and we are using active intelligence for all the deductions and the back orders. So together, we are using all of these platforms for getting all of the views for our operations. And the goal is to monitor real-time data so that we can make our operation more efficient.

John Radko: Okay, thank you very much. Hey, I want to thank both of you for sharing the current world and your reality. I think it's very meaningful to certainly all of us, but all of your colleagues in the room. Now, I want you to—I'm going to give you the license to predict a little bit and look ahead. So let me ask you, I shared earlier my thoughts, but in your view, how could AI reshape your business in the future? I'm not even going to ask where—I lead the witness. You tell me how you think AI is going to change the world and the operations you're in. Maybe, Jafrina, we'll start with you.

Jafrina Jabin: So looking ahead, I believe AI has the potential to reshape our operations even more profoundly. I see opportunities to improve demand sensing where AI could be used to predict the increase and decrease for product supply and demand, which can be done through the history of our previous data. AI could also help for deeper understanding of consumer behavior and optimize fulfillment and insights. So if I want to say, in short, so AI would be the ultimate tool which would allow us to move from reacting to problems to anticipating them, which will actually be the future of resilient supply chain.

John Radko: Okay, thank you. Scott, thoughts on looking ahead?

Scott Stevens: Looking ahead, as I was saying with my years of EDI experience, I think one of the biggest changes I'm seeing happening and I'm actually seeing it through multiple of these sessions is the ability to be more proactive using EDI data to come up with some of this information instead of relying on your ERP, whether it be S4 or whatever. And I think that's very exciting. It's an area that I never thought would happen, so I think that's great.

I wanted to start with some small things. I wanted to—I've challenged a couple people here to—can we give our business people some tools that they can not have to come to us to ask? Like, did I get an acknowledgment for an invoice, or did this purchase order actually come in via EDI? I want them to be able to go into the OpenText tool and look that or ask those questions themselves without giving them a bunch of technical knowledge or skills to have to do that. So that's where I think AI is coming.

I'm seeing, and you guys have seen some of this today, I was talking to some people this week about it, like the chargeback help and being able to assist with advanced warning and notices and that things and helping with remediation, I see that. And then the whole error handling, I think we all know how that works and how complex that is. And I'm looking forward to some of that.

John Radko: All right, thanks. Final question. Advice for others. Share your advice to us and to everyone else in this room about for companies that are really just starting their analytics or AI journey, what practical advice would you offer, and where do you think they should begin to see quick wins? I started with Jafrina last time, so I'll start with Scott this time.

Scott Stevens: One of the things that I would advise, because I'm doing it right now too, is looking to the Command Center. What they're talking about looks like it's going to have some great opportunities for us and for growth there. I'm also—would say look for some quick wins. I just mentioned some things about maybe quick AI looks for the business to help look things up without the technical EDI team having to get involved with that.

The other thing that—it's big at Hershey, and we actually created teams that are set up to review all AI initiatives, they want to make sure that it's secure data and it's factual data. So there's teams that have to review that before we're allowed to go into some AI spaces.

And as I mentioned in the last question, I think the importance is try to use and leverage the EDI data, which is way ahead of the time, before it gets into ERP to get your data.

John Radko: Okay, thanks. Jafrina.

Jafrina Jabin: So my biggest advice would be to start with the data that you already have, because EDI is already consistent. It's suitable for analytics and AI. So it's a great beginning point to start. But data is the backbone of EDI and AI. So we have to begin what we have. And we have to standardize and clean the data that we already have. And after that, if we can incorporate EDI and AI together, it can remarkably change the supply chain game in future.

John Radko: Okay, thank you so much. And Scott and Jafrina, I just want to thank you for sharing with us and for being here with us at OpenText World.

Scott Stevens and Jafrina Jabin: Thank you for having us. Thank you.

John Radko: Thank you, everyone.

See Integration Monitor in action: Anomaly detection demo

John Radko: Ah, Sushil, I'd like to invite Sushil up to tell us about Trading Grid Integration Monitor and provide a demo. I think you're going to set the stage, and then let us go.

Sushil Pancholi: All right, thank you. So this is something that we're very excited about. We've been talking about anomaly detection virtually at most conversations that you've been having with BN over the last couple of days. You've heard about it. So we've actually got a demo here.

Okay, so the setup for this scene is we have Oliver, who's a new EDI analyst and has been charged with monitoring his company's B2B community. And Emma is a senior EDI analyst who has offered to bring Oliver up to speed using TG integration monitor.

[Demo Dialogue - Emma and Oliver]

Emma: Oliver's company uses Trading Grid. Oliver suddenly inherited responsibility for his company's frontline support for working with partners. Emma is mentoring Oliver in using integration monitor, an essential tool for Oliver to monitor their B2B community.

Don't worry, Oliver, I'll help you get up to speed. Here I'm showing you an Integration Monitor screen. IM gives a total picture of the history and activity for our Trading Grid integration solution. This one shows information about your different B2B documents. I showed you yesterday how to filter document level information by partner, document type, and all that.

Oliver: I looked it over, and it's impressive, but I did feel overwhelmed at one point. I'm responsible for B2B performance, but we have so many partners. I mean, I could spend all day flipping from screen to screen, just trying to find problems that might be happening.

Emma: No wonder you were asking me this morning about how to detect things. Can you see the anomaly screen? They are the red lines on the screen. Each anomaly shows when the document levels were out of range. IM lets us configure integration monitor to generate a forecast of expected future behavior based on history. Can you see the forecast on my screen?

Oliver: Yes. Oh, Latham. All this tariff stuff had us scrambling for months.

Emma: Yeah, for good reason. They're an important supplier. We were just looking at the overall B2B traffic, and Latham B2B documents are so few, they barely affect that graph. But if we have a disruption with them, wow, do we hear about it. That's why anomaly detection is so important.

Oliver: Like if there's a bunch of extra stuff happening?

Emma: Extra activity is a kind of anomaly. But there are other kinds. For instance, if something is missing, that's also an anomaly. You said yesterday that anomalies are strange things in the data, and that our goal is to identify the anomalies that show when something is impacting our business.

Oliver: So there can be important things that are missing and we need to know about them?

Emma: Yes, exactly. We had an incident four months ago when Latham wasn't sending ASNs for days. Trucks started arriving, and warehouses didn't know what to do. An assembly line even had to shut down. It was a big deal. Turns out it was a weird glitch in their system, and they fixed it immediately. But nobody noticed for days.

Oliver: I heard about that. Didn't we have an off-site that day?

Emma: Not a good combination. Well, the problem actually started days earlier. They had the same problem recently, as you can see on the screen. Latham doesn't send many ASNs. We weren't getting any of them, but we didn't have anomaly detection then, so nobody noticed the problem. The business impact was huge. Now we're watching for lack of ASNs from Latham and other critical suppliers like what you see on the screen.

This may be confusing. IM produces forecasts of future behavior based on what happened in the past. Anomaly detection then can keep comparing activity against the expected behavior, the forecast. For the Latham ASNs, IM compares the forecast and the current behavior to identify when traffic was expected, but just wasn't there. To the left of the screen is history, and on the right is future expected activity.

Oliver: But what's that fuzzy region showing?

Emma: It shows the range of normal activity. Depending on the hour of the day, day of the week, week of the month, and so on, the fuzzy region shows the likely minimum and maximum for the activity, in this case, the number of documents. Anything outside the fuzzy region is unexpected, a kind of smart adaptive threshold. IM uses that for anomaly detection, since activity outside that range is anomalous. For Latham, we set lack of ASNs anomaly detection for when there isn't any traffic, but some traffic was expected. It's pretty cool. For other partners, we check for different anomalies.

Oliver: Emma, so we can check different anomalies based on what we do with different partners? That is really cool. I see I have another meeting in a moment, but could you show me more tomorrow?

Emma: Sure. If you think about it, anomaly detection packages up years of experience. It can help you take care of our partners, even though you're new to this.

Oliver: And that's encouraging.

Emma: Tomorrow, I want to show you how to set up forecasts and anomaly detection, so you really understand what's critical in our business.

Oliver: Thank you. I'll see you then.

Sushil Pancholi: So just one last thing, I want to just remind everyone to please come visit our booth. We have Command Center integration monitor fully, as well as the anomaly detection use cases there to demo. We'd love to get some input from you. And we can talk about different other use cases as well. Okay, I'll bring Mark up here. Thanks.

Key takeaways: The future of AI in supply chains

Mark Morley: So thanks very much, Sushil. I think you'll agree that was an excellent demo. It shows the general direction of where we're going, combining AI with our transaction flows. I just want to say thanks for attending. I think we're actually spot on time, just maybe a minute over. I think you'll agree the customers did a fantastic job in the fireside chat today. I want to thank our guests and John and Sushil for providing the direction of where we're going in the future.

As a reminder, if you've got the mobile app, you may have noticed that we've actually got two other business network sessions directly after this one. I think it's about a 10-minute gap, in room 208A, which I think is one corridor down. So if you're interested in learning about secure collaboration and also the second session there, which is going to be running around supply chains brain from transaction to intelligence, and Vesna running the power up your supply chain collaboration, please feel free to attend those sessions directly after this one.

So thanks for your time and thanks for your attention.

Related content:

• How Agentic AI and IoT prevent production disruptions: The rise of self-correcting manufacturing
• What’s New: How B2B Integration Essentials simplifies global SMB supply chain connectivity
• From visibility to autonomy: AI’s role in optimizing supply chains
• From conversational intelligence to connected visibility—turning IoT data into action

The post AI-optimized supply chain community playbook:  How to win the supply chain game appeared first on OpenText Blogs.