FutureLink Barcelona 2026 Keynote: TraceLink Powers the Agentic Supply Chain Operating Model Transformation

Narrator Voice: Welcome to the future of the supply chain, where events unfold across a global network, where information flows in real time, where agents observe, reason, and act, where humans and agents work together to move business processes forward, where agents and humans work in harmony.

Welcome to the agentic supply chain operating system.

Shabbir Dahod: First of all, thank you, everyone, for coming here. I hate to count. At my age, I hate to count the years, but this is my 24th year in the life sciences supply chain.

I have Wayne here, who has been on that journey for almost all of it as well, from Deloitte. He was at AMR. I think many of you have been on this journey with us for decades. When I look around the audience here, these are all familiar faces.

We've all worked together on so many different things, and for so long, we've become friends. We've become partners in this journey. I want to give a deep thank you for that, because the most important part of this journey is the relationships that we build, how we learn from each other, and how we try to improve each other.

We do that both as trying to do the greater good for your business, for patients, for consumers, and hopefully from what we can also provide to you as well. Most of you have probably met me at some point. I tend to travel a lot, focused with people in their offices and in their plants.

That's where all the learning really happens for us. You'll see most of us in the field, mostly because that's when we get to really understand and learn what you do, what your environment looks like, and what the challenges are that you have.

Then we take all that in our hands, and we distill that and try to create products, tools, and capabilities that can help you. That's just been our journey throughout the last 20 to 24 years. It's been about this very deep partnership and this deep level of trust that we've built over the years.

I want to thank you for that. I really want to thank you because it's been the most rewarding part of my life. I know for many of the people that have been on this journey with me, it's been the most rewarding part of their careers and their life as well.

Now we're not entering into a new chapter. We started that chapter last year when we started talking about what's possible within the agentic world and where we can go with it.

Last year, we tried to paint a picture on what is the art of the possible with having agents as users and the ability to create agents that can do work, that can be reliable. That's a lot harder than most people think.

I think most of you have learned that when you're looking at agents and the agentic opportunities, it requires a lot of capabilities to come together in order for it to really work. It has to actually work in a true enterprise environment where patients' lives are at stake, revenue is at stake, and the relationships you have with your partners are at stake.

You don't want to let anyone down. Therefore, you need to really create a platform and a capability that can address the full solution and not have to use a bag of tools and these piecemeal capabilities.

When we look at what we've built together over the years, we are in a perfect position to provide the agentic capabilities within the industry and within the organization. At the end of the day, the supply chain starts with a network.

We are all integrated. We're all interlinked in order for us to be able to execute on our job. You're not only integrated and interlinked within your functions within the organization, whether it's external manufacturing, commerce, your 3PLs, or transportation.

Those functions within your organization are also interlinked with everybody else that needs to participate with you and needs to collaborate with you in order for you to be successful. For your success is their success as well. Therefore, at the heart of it is about both [?] and methylmalts.

That's why the heart of the foundation we built is the network. As you know, it's a very large network, a very vibrant network. The pharmaceutical supply chain relies on this network in order for products to get to patients.

That network is built on a significant amount of trust. That trust is in how we manage all the information, all the processes, and the security levels. Many customers come to us, especially one came to us recently and said, "We need you to even do better on security."

We embraced it. We said, "Great. Let's do it," because we want to get better. If anybody wants us to get better on scale, performance, and security, we're always going to invest into it. We're always going to make it better, because we know it is the operating infrastructure for your business.

That operating infrastructure has to be trusted. That whole platform that we built on top of the network with that trust is a significant investment we've made in order for us to be able to digitalize all processes.

The OPUS Platform was created in a very unique manner to address the very specific needs of supply chain. It's not just any platform for local development. It's for multi-enterprise applications.

Why is that important? Because all of you are interlinked with each other. Therefore, your applications have to be aware and knowledgeable about each other so that when you exchange information, it's in a secure manner.

When you're integrating processes, it's done in the manner in which you can manage it and the information can flow easily. The actual capabilities that you have out of the box are already inherently network aware.

The OPUS Platform was constructed with that as a design point, which makes us very unique and very specific to the needs of the supply chain.

Then as we move forward, we digitalize all the processes. Whether they were the B2B transactions that you have to do with each other, which are in the hundreds, the variations of that are in the thousands, and all the different types of choreography that you do are in the tens of thousands across that.

We moved forward with that platform to support all the B2B transactions and now the collaborative processes that you have with POET. That created the ability for us to now take a network on a solid foundation that's trusted, with the ability to have these applications that can be built, that can be multi-enterprise inherently, to now digitalize all the processes so that information you may unlock.

With the information you unlock, that's where the real diagrams. Now, with the ability for agents layered on top of that, you can actually do work. That work is where the value really accelerates. You'll see the number here, the $1.5 trillion.

The way we come up with that number is to say basically, for all the work that people have to do that are supply chain knowledge workers, how much of that work can we have agents do?

I want to dispel one fallacy in that number, which is that this is not about replacing power. This is about productivity, because an actual value is not that productivity, actually. As you'll see later on, the real value is the business benefit.

When we look at that number of $1.5 trillion, we look at that and say that's the level of productivity we can drive. That productivity has an impact that's 10x or 20x more on your business performance.

Therefore, your goal is not to replace people. Your goal is to empower people, to make them even better, to make them even more effective, to make them even more intelligent and more knowledgeable.

This really runs counter to the words that you hear out there about people losing jobs. It's quite the opposite. What we observe, and what we'll unpack here, is that it's really about having greater output for the greater good.

I don't want anyone to ever walk away from agentic capabilities as anything else than really making the businesses better and making society better. That's really the opportunity we have. We're going to walk you through all of that.

How does that start? It really starts with the fundamental mindset of what we're trying to accomplish. When we design any solution, any application, we think about your process. We think about what you do on a day-to-day basis.

We think about all the knowledge we gained in being in those sites, working through what you have to do when you have to do production, what you have to do when you do shipping, and what you have to do when you're planning. We recognize that the opportunity here is that you can now add agents to that operating model.

The ability to say we can have an agent that can be a teammate, that can actually do a lot of the work that can make me better at my job and can make the business better as a whole. The ability to have the humans and agents work together in a symbiotic manner is the way that we will be able to create this agentic supply chain operating model.

We have to look at it process by process. We look at each process and say, "Let's unpack this process. Let's now reimagine this process, where we can have agents pick up a lot of the work that would be helpful for me for the agent to do, and then for me to be on top of that, to review that, to make critical decisions off of that."

I'm not burdened with the fact of doing things that I really don't need to be doing anymore. I can have an agent do that for me. Keep that in mind, because this is a very different point of view than somebody coming in and saying, "You don't need planners anymore. You don't need anyone to deal with your external manufacturing relationship."

"We'll just have a super agent, super intelligence take care of that." That's BS, just to make that very clear. It's really about a world in which we work together with agents and that the agents improve our lives, make us better, and make the business better.

The opportunity is that you have the ability to now have this teammate that can take on a lot of work and actually scale up and say, "I need you to work immediately right now on 1,000 things, and then don't go do anything." Now we have a team that can do that.

We can do that at any moment. As soon as a particular event or something occurs, pick it up and start doing this. This teammate will go do that. That teammate may have to specialize. "I just want you to focus on this, and only focus on this, because I need you to be a very specialist."

"Anytime it overcomes, that a meeting with the amount of inventory. Make these decisions and be a specialist on it. I will review everything, and I will then make the final decisions. We'll collaborate together amongst the whole organization that is full of humans and agents."

Our goal is to be more productive, so that I'm 2x more productive. With the agent, I'm 2x more productive, I'm 2x more effective, and I can drive revenue. We as a team can improve our service levels. We as a team can drive down the cost overall for the overall operations. We can improve our cash, and we can lower the risk.

That's the agentic supply chain operating model. It is not one in which you hand over the keys to some agent and you then expect it to run your business. It is a collaborative environment in which humans and agents work together. In order to have such a collaborative environment, you need an operating system.

Many of you have heard the term operating system. You use operating systems on a daily basis. Let me make it very simple. When you are using your computer and you move a mouse from left to right and you click, what's happening?

That's an operating system that's running in the background. What's happening is that as soon as you're moving that mouse, you're moving from one place to another, events are generated. As soon as you click, an event is generated. You open a window, an event is generated.

An operating system thrives on an eventing mechanism. What we have built is an ability for the whole end-to-end supply chain within the organization, across the network, to now have events that are occurring on a continuous basis.

Those events are driven at the bottom-most layer by supply chain interactions. Whenever a shipment occurs, whenever a message occurs, as those supply chain events occur and information is exchanged, that's an event.

What we do is we integrate once into everyone so that we can then drive up the amount of flow of information and the flow of the events that are generated from that information.

You now have this digital twin that you can see about the events occurring, all the information flowing, and now you can have multi-enterprise applications that can understand that information, turn it into intelligence, and then we can have agents that are observing these events.

They can react to these events and execute based on the specific events that they care about. They can look at the information related to that event, look at the information that's across it, and look at that information that's related to the invoice.

When that invoice comes in, that's an event. An agent can pick it up, can look at the purchase order, can look at the ASN, can look at contract pricing, and then help make decisions. That is agentic orchestration. That is an agentic supply chain operating system.

It's one in which it's an active environment that's constantly live and driving activity throughout your organization, where humans and agents are working together to react immediately, to make smart decisions, and then move the process forward in the most effective manner.

The way they do that is through agentic business processes. Those business processes now encapsulate all the related information in order for all of us to work together.

When you look at it from that perspective, an agentic business process is really about the fact that you've got humans as teams. You've got agents as team members, where the agents report to the humans, but they're all working together. They all work together on one or more important processes. That's where they collaborate.

In POET, you have to lay out all the processes you care about. How should they update the workflow? How should they then proceed to the next step? When you proceed to the next step, another agent might kick in, or two agents might kick in and say, "We've now moved this purchase order from draft to approved."

From that perspective, now two of us hear about that, or this exception has occurred. The planner agent cares as well as the materials manager, etc. You can have that type of movement.

It means the work that the humanization is put together. MINT enables the flow of integrated information. It enables the flow of integrated information on an end-to-end basis, because we all live on the fact that these exchange information in our processes.

Finally, the OPUS actions are the things that get triggered. When an event occurs, an action triggers so that it can actually do work or execute some logic that may invoke one or more agents, may make some decisions, may update a public process. It's an active environment.

At the lowest layer, you have the integrate once network for the specific transactions you care about from a MINT perspective. You have Chorus now, which lets you have real-time access to information to any systems of record inside your enterprise or outside your enterprise into the Internet.

You can pull in information in order to make these decisions when these events occur. Then you have the lakehouse that enables you to maintain and retain information for long term and provide analytics on top of that.

That creates this agentic business process. If you have the ability to have very rich reports and dashboards that the agents can use to show you information, you can use to look at information.

The beauty of the OPUS Platform, because everything's metadata and everything's relatable, is that you can do reports and dashboards, employee processes, and processes integrating Chorus objects. This is the environment in which you construct an agentic business process.

What we're going to do throughout the rest of the presentation is do a little less talking and a lot more show. A lot of this industry talks. They want to show you how you can get to an agentic supply chain to operate more.

We're going to start with the very basics. The first basic is, let's construct an agentic business process. Let's construct an actual reliable, performance agent that's cost effective, because without that, we can't really have much. We'll just start with the first demo video. I may pause every once in a while so I can explain things. Please go.

Demo Narrator Voice: OPUS provides ABP packages to bring speed and accuracy to your supply chain. Every ABP starts with real business transactions. Inside the ABP package builder, we bind this MINT process network directly so the package understands the exact transactional context it will operate against.

POET manages agents' decision logs and enables human collaboration on the agent-generated data. We model the decision log object in POET, then link the POET process network to our ABP package, giving our agents a structured target to record outcomes for human review at any time.

Actions serve as the event glue. When an incoming invoice event fires in MINT, this Melody script automatically creates an appropriate decision log in POET and invokes the target agent objective with full context.

Agent profiles establish clear boundaries around what an agent is authorized to do. Inside each profile, we explicitly configure its core intent, objectives, granular tasks, decision boundaries, and compliance rules, giving the agent a precise definition of the business outcomes it needs to pursue.

Because every operation...

Shabbir: I'll just pause here for a second on purpose. What you saw so far is the ability to say, "Need to do this agentic business process. Here were these MINT transactions. I'm going to create this business process in POET that's going to track what's going to happen."

Then we're creating this agent profile. As you see here, the agent profile is highly specific. As you've seen, we've observed in working with customers that this could end up being 20, 30, 40, or 50 pages long.

Why? Because your work instructions are that. Whenever you sit down and do work instructions for an invoice certification or a plan approval or any of those aspects, you have to go through and look at many different rules and decisions you have to make, and you can look at many sources of information to do that.

When you have an agent profile, it can reach out to anything that it has permission for, and it can start to make those decisions. The first step is that agents themselves, in order to be effective, have to be very specific in what it is that you want them to do. I want to make sure we've made that point.

Demo: is unique. Customers have full control to define, customize, and govern their own agent profiles tailored specifically to their business needs.

Now, the ABP package definition is completed by adding the agent profile to the ABP package. Once the package is finalized, the agentic business process is operationalized by instantiating an active ABP, ready to operate and monitor the team of agents.

Agents can be seamlessly created directly from a predefined profile, automatically inheriting its complete IO TDR right out of the box. This ensures that every newly deployed agent aligns instantly with your organizational standards.

Shabbir: This is another important point. You create an agent profile. The agent profile says, "These are the work instructions for a particular type of agent." When you actually create an agent, you may want to create multiple agents that are highly specialized.

You may have an example where you have an agent that is doing the same voice processing, but you may have multiple different agents that are using that profile for maybe working with different people, different regions, or different types of materials that you're responding to.

There's an opportunity for you to create a fleet of agents that follow the same SOPs, but they have the ability for them to be in the same agentic business process. You can look across them all and look across the whole team and see how well they're all working together. Standards.

Demo: These agents operate under strict enterprise governance. Every agent is assigned a dedicated human manager and bound by explicit role-based access controls across all connected process networks.

Once the agent is initialized, the OPUS brain ingests...

Shabbir: Sorry. In this particular case, remember that an agent is a user. You give it permissions just like you're giving it to a user. If I'm only responsible for these five suppliers, then those are the only links I have membership to, and those are the only invoices I will see, in this case, from an invoice agent.

I don't see anything else, so I can highly constrain it. I can also make a sense that the agent doesn't have permission to approve anything. It just has permission to gather information and report to me the analysis that it's done.

I know security is a very important aspect in an enterprise. The way we define agents is that it has this level of security built in on top of the trusted pipeline.

Demo: ingests the complete IO TDR text and constructs a comprehensive execution plan designed to run on top of the OPUS semantic layer. Customers maintain total operational oversight at every single stage, with full visibility to review, evaluate, and approve…

Shabbir: I'll pause here a second as well. One of the things that we've learned is that AI has a level of hallucination that's built into the inherent aspect of the technology. When you look at any sort of LLM, that LLM is actually built to be nondeterministic.

What we do is we actually take that agent profile, and when we create the agent, we compile it into this inference plan. The way we do that is we constantly run testing against that agent in that context to now optimize it.

Until that agent passes 100% of the tests that have been generated, we don't actually finalize the plan. We let you see everything that we're doing from an inference perspective, so you have full transparency into what the agent is going to do in the real world with real data.

Then we optimize that execution path to not just be 100% accurate, but also optimize it to be 100% faster.

Demo: …the generated plan prior to live execution. As we saw in our real-world supply chain agent example, the underlying agent definition is exceptionally detailed and meticulous.

Every detail embedded in this text must be parsed and interpreted by the OPUS brain into exact, highly accurate downstream actions. Every single execution is engineered to be 100% reliable, completely transparent, and fully repeatable.

Shabbir: When you look at most AI, when you see all these benchmarks and hear, "They're really doing super on this benchmark, and they're at 75%. 80% if they're good," that means that when you're looking at AI, there's this level of uncertainty built in.

In order for us to feel confident that we could actually say, "Yes, go ahead and do this," I don't think you want to have a worker that's 85% correct in approving invoices or reviewing plans.

Therefore, we made sure that we got to a point where it's 100% correct for this data set. Again, when we looked at it and said, "We want a very specific set of tasks and objectives that the agent should do. We want to make sure it's tested and verified for that. We provide full transparency into it."

Then when we actually create the agent and deploy that agent, we make sure it's optimal and that's fixed. Why? Because we know. We work with you. You go through validation and verification, and you don't want to have any issues if it's going to be looking at invoices, purchase orders, quality documents, or anything else.

You want to know what it's doing, because you have to dial it. It gives you its quality. It gives its compliance. This is all built for that level of transparency all the way through.

Demo: From there, you can deploy a dedicated team of agents based on configured profiles, tailoring specialized agents to handle specific trading partners, regions, or product segments. These agents are added to the ABP, from where they can be monitored.

Shabbir: In order for us to feel confident about using agents in a true business enterprise environment, we knew that the foundation of having agents that can be validated, verified, be able to define a profile that was specific to your business, to business processes, and make sure that we train the agent to be 100% and fast and cost effective to validate the work.

Then what we did was we measured, and we were able to get 100% reliability. We were able to get 100x improvement on speed, more than that. For us, we first started doing invoicing for 28 minutes. Then we applied this approach, and we were getting about 60% accuracy.

We applied this approach. We're getting 100% accuracy and 100 times faster. Good for you, it's also cheaper as well, because our optimization helps you have it be cheaper. This is now an agent that runs in the fabric, where we can have thousands or hundreds of thousands.

Nova Scotia can support millions and billions of these agents because of our core scalability of core. Again, we're making a decision about what's the right platform, what's the right operating system. The foundation has to be reliability and accuracy.

I want to hit first. Now I'm going to try to be one of you, if that's OK, for a little while. Again, for us, it's important that we make it very real. Very, very real. There's so much hype, and the number one thing that I would say I hear from customers, or I hear in the undertones, is there's a lot of skepticism by the AI.

Many of you have done projects that haven't gone well, or it's been OK, very costly. I was talking to a CIO just last week who said that they put an agent to work, and it was actually costing more than the human doing the work.

There have been all these unfortunate situations. What I'm going to do is walk through from the perspective of I am working at Linkiva. I am responsible for external manufacturing, and I am very excited about what we can do with agentic business processes and what we can do with OPUS.

There are many different areas that I can see that I can impact. There's roughly about $2.25 billion worth of external product sales. I've got about 60 CMOs I work with. I go to 50 markets. It's not just a regional player. I've got 10 3PLs that I have to work with.

I observe that in order for me to be successful, there are these ongoing areas that I need to improve on. The first thing I need to improve on is product availability. I need to make sure I have the right product at the right time in the right place. How do I achieve that?

There are areas I've looked at to focus on, making sure that my POs are the right price, making sure I have full transparency into how the product is flowing and I'm monitoring that. Related to that is, at the same time, I have quality processes going continuously.

I've got batch records I've approved, and I know that all that's related to the production process as well. I have to make sure that I get right first time, I get optimal cycle, and I'm going to reduce my cycle times.

I've also got to make sure that I'm working well with my suppliers, because I want to improve the actual relationships and effectiveness of not just myself, but the suppliers that I work with. I want to make sure that we, as partners, are improving ourselves and taking out cost as well on both sides.

Sometimes that cost is shared by me because I don't understand the forecast on time, or I may not have the right artwork I've sent to you. Things really work. I have to also make sure that whenever I release a product, I have market access.

I can go ahead and make sure that it gets registered to the EU hub or any other hub that it needs to be. I have to make sure that I have good supply continuity, because it's not just about me producing it and getting it to the CMO. I have to make sure it shows up on the customer side on time and full.

My on time and full is really, from my perspective, getting it to the customer. I have some challenges there. I'm going to improve on that as well. I have to constantly monitor inventory, because there are sales forecasts that I have to compare, but I've got to look at inventory.

My inventory is everywhere. My inventory is in my warehouse, my inventory is at the CMO warehouse, my inventory is at the 3PLs, and I have high and low watermarks. Not everybody looks at all the alerts. Not everybody looks at all the inventory levels.

I have to make sure that I stay on top of where the inventory is at on a daily basis, or maybe even twice-a-day basis. Finally, I want to make sure that when I pay my invoice, I'm paying it accurately. It's correct, so that I'm checking everything and double-checking everything.

I believe by being able to improve all my end-to-end processes and making them more authentic, there's a lot of opportunity, because I see that I can improve my productivity and the revenue by having a more predictable supply chain.

I can increase the revenue, have the right product at the right time, in the right quality, and reduce my risk with the right first time. I've got better supplier relationships because they're communicating. We have clear data that we're sharing with each other on it.

I'm making sure that I don't have issues with market access, any sort of issues where their EMVS hub maybe refuses my shipments for some reason. I make sure that I can get my service levels up, that I improve my service levels to my customer, but that means I have to monitor through transportation.

I'm constantly keeping my hand on the pulse of inventory. I know it. Every day, I've got a good report to tell me about it. I'm making sure that I'm paying what I have to pay and I don't have any sort of verification.

I want to also make sure that I pay on time because their partner fees on time. I don't want to be one where I don't know that something's been paid yet. I believe that by doing all that, there's tremendous opportunity.

Now, let's start on the right. My workforce costs me $45 million a year. My issue is not reducing them. These are great people, but my real value is how do I make those people more productive, drive up their productivity?

The real advantage that I have is that I've got revenue at risk of about $225 million, but I know that's revenue that if I can continue to chop away at, then I can increase my revenue. If I get better on my market availability, I've got about $394 million of spend that I think we can do better with.

I think that if I have that collaboration with my suppliers, better visibility, and better transparency, then I'll be able to bring that spend down. I've got about $155 million of inventory that I think I can make a dent in, too. I can have leaner inventory because I'll have better visibility and better transparency.

Now we're going to go on a journey about making me successful. I'm going to be successful as well. The first step is really, how do I get this kind of visibility into what's going on from the moment I issue a PO, making sure that then I can make sure it's the right price?

Do I get the PO acknowledgment? When was the batch started? When was the actual production process going on? Where is it in QA and the QC process? Do I have the yield information? Is it released? Is the ASN? Has it shut down?

I have to know that so that I have visibility, because I've got 60 or 70 CMOs. I've got production going on all the time. In order for me to be effective, I don't have enough people to chase all this down.

Even if I got digital integration with everybody, just monitoring and managing all that is hard, even with the people I have. With the people I have, I need to have better visibility.

Let's move to the first part of this, which is how do I integrate everybody to get the information, and then how do I track the actual process through it?

Demo: Linkiva creates a purchase order in its ERP for its CMO. The order flows into MINT in draft for pricing validation before submission.

The standard options available for customers and partners to link directly to their enterprise systems using HTTPS API protocols include NetSuite, MS Dynamics, JD Edwards, Creation, and SAP S4HANA. This company, Linkiva, happens to use SAP S4HANA.

Now let's look at some of the other integration options available beyond HTTPS APIs, which are available to any customer and any of their partners. We will start by opening the Extensible TraceLink Transfer application, where we see options for SFTP, AS2, and SMTP email.

The last protocol we'll explore in XTT is SMTP email, where you or your partners can use simple email attachments to exchange files with TraceLink. Now we'll switch to XTT exchanges, where you see support for many common industry exchanges, such as SAP Ariba, GHX, and IBM Sterling.

If you or your partner needs to exchange data...

Shabbir: This is all about, I need to get data from my system in and out of my system. In this particular case, I've got S4HANA. TraceLink, out of the box, supports S4HANA APIs. I can integrate in and out of that.

It also supports my partners using any systems they want, any way they want to integrate with me, whether it is via email even. We have customers sending us PDFs and saying, "Here's the purchase order PDF." Then we take that PDF and we convert it into canonical that becomes a purchase order transaction.

The first step is really making sure that there's the foundation to integrate once, but integrate however people want to integrate on any transaction.

Demo: from a system that can't be integrated with, like a file system, then the MINT user interface is always available. Regardless of how your trading partners exchange data, your company receives that data using its preferred technology.

Linkiva needs to verify...

Shabbir: One of the issues I have is that when the purchase orders go out, the price is... Stories I hear about. When the purchase order goes out, the price that's on there needs to be verified.

My contract management system is not really tied to my purchase order issuance system. I also have multiple sites that are oftentimes ordering at the same time, and they're not aware of the actual price that they can get.

I need to have the ability to, before the purchase order goes out, verify all that. In order for me to do that, I need to access another system that's not my SAP system to actually verify that.

In this particular case, what we're going to do is show you how you can use Chorus to access any system and then be able to bring it as a model object so that now you can see it in its actual model data, but the agents can also use it. Of course, internationals can also use it.

Demo: purchase order pricing against the agreed-upon pricing maintained in its contract management system before sending the order to its CMO. Chorus makes those pricing records available through the familiar search screen.

Opening the Lumetrexa record, we can see the unit price of $10,000, matching line 10 of the purchase order. With the link now established, let's return to Linkiva's purchase order and the pricing check triggered in MINT.

Here is the purchase order created in draft state in MINT by the inbound XTT Link Actions transaction. Saving it as part of that inbound processing automatically triggered an OPUS action that invoked the price validation analyst OPUS agent.

The price validation analyst evaluates...

Shabbir: This is the agent that I've created in order for it to check every outbound person. I can check on everyone, go through all those profiles of decisions, go check the catalog information, do any other calculations, and make sure this is the right price.

Demo: purchase order lines against eligible contract pricing accessed through Chorus. It records the price evidence and a recommendation for review, preserving the original purchase order.

This price validation workspace brings together pricing items that need attention, as well as recently completed assessments. Here, an unavailable contract price needs investigation, while another recommendation is ready for review.

The agent retrieves the corresponding records through Chorus and Bridge. The prices for the other five lines match as well. Chorus makes external business data available to the agent through a standard OPUS object, eliminating the need for someone to look up and compare these prices manually.

The agent records its line-level validation results in POET decision logs. The summary and reason confirm that all six prices match the external pricing records. This gives Linkiva a traceable record of the pricing checks and their outcome.

The successful case shows how pricing can be checked before submission. Now consider a separate exception example. For this purchase order, Chorus returned eligible pricing for 11 lines, but no eligible contract price for Lumetrexa on line 40.

The agent records that line for investigation and recommends hold full for review without changing the PO. Expanding the progress log shows how the agent checked the purchase order, retrieved the available pricing evidence, and recorded its findings.

The reviewer can investigate the missing pricing in the authorized source and decide the next steps. Returning to the successfully validated order, Linkiva submits the PO to its CMO.

Shabbir: In that particular case, we're having to modify the purchase order in [?] through Chorus, reissue it, change it at that point. Whatever the business process requires, the agent could do at that point, or the human could initiate, but then the script could do as well.

Now we're going to move forward. If the purchase order's gone out to my CMO, I need to now coordinate and get acknowledgment on that purchase order. A common issue right now.

Demo: The process also monitors whether purchase orders sent by the MAH have received the expected acknowledgment from the CMO. If their response is missing, the scheduled action identifies the affected orders and surfaces them to the MAH for follow-up.

The MAH expects its CMO to acknowledge purchase orders regularly. They create a scheduled action with a daily recurrence that evaluates purchase orders sent by the MAH and compares them with purchase order acknowledgments received from the CMO.

It identifies the purchase orders that have not been acknowledged within the configured response.

Shabbir: I hear this often, that I need something to keep tabs at all the purchase orders that are outstanding, that I don't have acknowledgments for, and that I need to be able to follow up on that because I need an answer as to whether this is going to be in that nowhere that I can actually rely on.

In this particular case, we have a scheduled action that will just every day, or whatever periods you want, look at all the open orders, the ones that haven't acknowledged, and flag them to be followed up on.

Demo: The result is captured in POET as a reviewable set of overdue purchase orders, so the MAH can initiate the appropriate follow-up.

Production has started, and order status transactions are flowing back to Linkiva. These received status messages all need to be checked, because if any product is identified as running late, action needs to be taken. Production schedules need to be assessed and potentially modified to ensure lines don't sit idle.

The order fulfillment monitor brings these related transactions together to track each order and identify where attention is needed. The planner can see missing acknowledgments, delayed commitments, and quantity or shipment gaps in one workspace.

Let's open an illustrative delayed commitment example to see the recommendation and supporting evidence. For this order, the delivery assessment indicates a seven-day risk against the required date. The agent records a follow-up recommendation so the planner can review the evidence and coordinate next steps with the CMO.

The expanded progress log shows which records were checked, how the delivery risk was identified, and why follow-up is recommended. This helps the team manage orders by exception rather than manually compare every transaction.

Linkiva gains a traceable view of commitments, order progress, and supply risks with clear recommendations for human follow-up.

Shabbir: The first leg is making sure that I get my orders there, I get the order status, I follow up, and all this work that humans have to do to follow up. A lot of times, it doesn't get done. When it doesn't get done, what happens is that it impacts our ability to have fun.

The first step is to really tighten that up. The other aspect of this is the ability that, as I have now tightened this aspect up, I've got a parallel process going. For sure, I've got batch. I've got batching reviews.

I've got multiple sets of different types of transactions, the collaborative transactions, the actual collaborative processes that I have for things like artwork, etc., that needs to go on. That's where POET comes in.

POET is a tool, a no-code tool for you to construct and compose any business process with any amount of information. What we'll do is a little deeper dive on all the different POET processes that we've already built that are available in the catalog, that you can take right out of the catalog, modify it to your desire, and create new ones very easily as well.

This is a means by which you can digitalize these processes that are people to people as well. Let's into this thing.

Demo: Let's look at the breadth of collaborative processes POET can support across the supply chain. Starting with external manufacturing, here we have batch record reviews, where manufacturers and their external manufacturing partners can coordinate review activities, clarifications, responsibilities, documentation, and decisions around a batch.

The information shown is specific to that process. Within Commerce and Direct Supply, teams can manage compliance exceptions, capturing the nature and priority of an issue, establishing responsibility, and collaboratively managing it through investigation and resolution.

Within transportation, POET supports processes such as indirect supplier incidents, giving teams a structured way to collaborate around issues involving supply chain partners. These are examples of the processes POET can provide out of the box.

Instead of creating another spreadsheet, mailbox process, or point solution every time a new collaboration need emerges, organizations can bring these processes together on one common digital collaboration platform.

Shabbir: Now you have this tool that can enable you to basically visualize any process. We've obviously created this at a very short time period because it's all done through no code.

We research, but within your organization, you can design it however you like. Like I said, it's all done no code, drag and drop. We'd be able to do that.

Now let's switch to the fact that I've got certain specific processes, which are around quality management and supplier management, that I need to improve on in my organization, Linkiva.

I want to make sure that I've got how quickly we're doing the batching review, how effective we're doing that. Then I want to make sure that I am aware of and collaborating very well with my supplier about the ongoing challenges that you're having bidirectionally as well. Here's what can inject to that.

Demo: POET transforms fragmented people-to-people interactions into structured, shared digital supply chain processes. Like MINT, POET is a multi-enterprise application.

Whether you're interested in exchanging data with internal company users or users at supply chain partners, all a customer needs to do is create a link with their partners, assign them roles, and start collaborating.

Let's start with one of these marketplace processes for batch record review. Just like with the base work item, its priority, due date, and the companies involved are captured. The review documentation, along with attachments and review notes, which are an integral part of the process, are included.

The outcome of the review is captured explicitly as approved, rejected, requires clarification, or canceled. The complete history, including conversations that led to decisions and final outcome, are all in one convenient place.

Because this information is structured, it can now be searched, measured, automated, and analyzed. Since Linkiva also uses MINT, we can leverage the network lookup by value field type and extend the batch record review object to create a relationship directly to a MINT transaction.

Next, locate the view edit page for batch record review and modify it by dragging the new PO number field to the general section of the view edit screen. That's it. The next time a user...

Shabbir: In this case, I need to add the PO number to the batch record review, so I just went into the open solution environment, took that particular batch record process that was in the catalog, added that reference to the PO, and now whenever my team is entering in information about a particular batch record, they can enter in the PO number.

That PO number will link directly to the POs that I have in MINT.

Demo: interacts with a batch record review object, the PO number field will be displayed and available for editing and traversing. We also associate the purchase order for business transaction context related to this batch.

Imagine for a moment you are one of the people involved with this production run. You have a question about some of the activities during production and want to understand what decisions were made and what led to those decisions.

You now have all the technical information, collaboration, and even business transaction context in one place. In POET, an email notification can be sent to the assigned partner when a batch record review is created or updated.

The partner, Tarivia in this case, can simply review the content of the email and reply. This reply is captured in POET and added to the originating batch record review as a...

Shabbir: Again, my partner can be in email, respond to back, and it ends up in the comments. I can also, within the company, follow certain batches. I'm a planner, and I follow a particular batch. I can say, "Follow this batch and send it. I want to know when it's going to be completed." I can do that. At least you can do that as well.

Demo: as a comment, keeping all the process collaboration in one place and fully accessible by those with the appropriate permissions.

Here is the corresponding batch quality record in the external QMS accessed through Chorus. BRR-126 is currently pending. Opening it, we can see the same Lumetrexa product, Tarivia manufacturing partner, and purchase order associated with the review in POET.

When the review is approved in POET, that approval triggers the configured workflow to update this QMS record through Chorus and Bridge.

Shabbir: I have my own QMS system, Vivo, whatever it is. Through the scripts and through Chorus, I have access to it. When this object is done, it's just simply closed out the batch, and its disposition needs to be updated. It can automatically go ahead and update the QMS system as well.

Demo: Its status now shows approved. The decision made in POET is reflected in the external system without manual reentry, keeping the quality records aligned and reducing handoffs.

Producing a batch is only part of the process. The team also needs visibility into whether the required evidence and quality reviews are complete.

The batch readiness and risk monitor brings together MINT batch and material evidence with POET batch record review, document reviews, incidents, CAPAs, and change requests. It checks material traceability, outstanding reviews, and unresolved issues that could affect readiness.

The agent records a ready-for-quality-review, follow-up, or investigate recommendation with supporting evidence and next steps. This gives quality reviewers a clear view of what remains outstanding.

Here, the workspace brings together the items requiring attention. The expanded progress log shows the evidence gathered and how the traceability difference was identified. In this illustrative example, investigate remains provisional, and the remaining quality assessment is pending.

The discrepancy requires an authorized resolution before readiness can be confirmed. By identifying risks early and providing supporting evidence, agents help quality teams focus their attention, make informed decisions, and maintain human oversight of batch readiness.

Shabbir: As I've improved the batch process, I've also got the agent that's observing it and identifying things that look like they need more attention. Then provide me the reason why it needs more attention so I can focus on them.

To make sure that I get those approvals done as fast and effective as possible, to make sure that product gets to the market based upon the commitments that I've made to my commercial partners.

In order to do that, I also want to continue to improve my supply performance. Because this is all on a single platform, complete metadata, a process is all digitalized, and all information is digitalized, I can have real-time reports.

I can have real-time reports on exactly how long these are taking, where the issues are, what the incidents are, and how long it's taking them to close. I can provide all that information to myself and with my partners so we can look at that and say, "How do we improve?"

We can take this dashboard of, "We need to improve our OTIF. We need to improve how quickly we can respond to each other." We can come up with the metrics that we care about, and then we can create the dashboards that we track with each other.

By doing that, then we will improve ourselves, bring down the cost, and drive up effectiveness for both our organizations. We'll show you how you can do that as well.

Demo: Supplier management requires a view of the relationship over time beyond individual orders or incidents. The supplier performance monitor evaluates linked POET supplier assessments, issues, CAPAs, change requests, and audits.

It checks applicable onboarding, qualification, and technology transfer obligations, then identifies recurring issues and overdue resolution work. The agent records a supplier performance assessment for each CMO with evaluation results, supporting evidence, and recommended follow-up.

OPUS Reports & Dashboards can use these recorded results to build supplier scorecards. Here, Linkiva can see the Tarivia items requiring attention, recurring deviations, overdue corrective work, missing audit evidence, and a technology transfer deliverable.

Let's open the recurring deviation assessment. In this illustrative assessment, Tarivia has four recurring deviations against a configured threshold of three, together with three overdue CAPA or resolution actions.

The findings are ready for review, with options to inspect the linked incidents and assign follow-up. The expanded progress log shows the evidence gathered and the evaluation behind those findings.

This example is awaiting review before the assessment and follow-up assignments are finalized. The supplier relationship manager and supporting teams can use those findings to identify where attention is needed and coordinate corrective work.

The value is greater accountability, clearer priorities, and less time spent manually assembling supplier performance information. Once these processes are running in POET, the structured data they generate can also be brought together into dashboards.

This gives teams a broader view of different collaborative processes and business areas, rather than looking at each record in isolation. We can use these metrics to understand patterns, priorities, workload, and activity across the network.

These dashboards can also be shared with participating suppliers, with each supplier seeing the information relevant to its own activities and relationship.

Shabbir: This is an important part. I know that many of our customers have to create special reports, share them in email, etc. Going back to the beginning, because OPUS was built as a multi-enterprise platform from day one, everything we build, the reports and dashboards, end up being multi-enterprise.

Meaning that the same important dashboard you can share with all your partners, and they only see the information relevant to them. They don't see their scorecard. You only need to create 50 scorecards. I don't think I'm to create 70 scorecards for my 70 CMOs.

I create one scorecard or a set of scorecards, and then I just make that available. Whenever each partner comes in, they see their scorecard. I see across all of them. I can slice and dice across all of them. They just see the ones that's relevant for them.

This is a way for us to be much more productive together and be much more productive in how we actually share information as well.

Demo: Because the data is linked to the companies participating in those processes, we can also look at the same information through a specific partner lens.

When we filter the dashboard for Tarivia Biomanufacturing, the view immediately narrows to the processes and metrics related to that partner. This gives teams a much more focused view of their collaboration with Tarivia across the different business processes.

Shabbir: We've gone through the process of being able to have the production all the way through, be able to see the actual batch record review process, and manage my suppliers. These are the areas I focused on.

Obviously, everyone has different areas they focus on. The beauty of the agentic operating system is that you can pick and choose your erectile journey as well. We'll get into the next few slides later on.

My journey isn't done yet, because I have produced product, but I also have issues of making sure it's available on the market. I need to have the ability to track the CMO reconciliation with the serial numbers. Almost everybody needs to be serial numbers.

Make it a three-way reconciliation in the CMO, we can report the ERP, and let's make sure they agree. I want to create agents for that as well. First of all, I need to deal with EU exceptions because something like that occurs and I need to jump right on it because now that product is unavailable.

We'll show you how we use agents to help with EU exceptions as well. Next.

Demo: Following CMO production and product release, Linkiva must confirm that the required serialization data has been successfully reported to the intended national system.

In this scenario, Linkiva has received serial number data from its CMO, triggered product release, and generated the EMVS product pack data report. Generating the report is one step. Confirming its acceptance and successful processing through to the destination system completes the reporting process.

The EMVS Alert Investigation Analyst supports the investigation when an alert occurs. It evaluates serialization records, product master data, reporting results, and EMVS evidence to identify the likely cause and recommend corrective action.

Authorized personnel review the findings and perform any required correction. Let's examine an alert involving a reporting distribution failure to the German national system. The objective brings together alert investigations that need human attention.

We are at the final reporting stage of Linkiva's external manufacturing process. The serial data has arrived, the product has been released, and the product pack data report has been generated.

Let's review the German alert. The agent confirmed that the serialized product data matches Linkiva's SOM records, but identified a failure in distributing the product pack data report to the German National Medicines Verification System.

This highlights a critical reporting requirement following CMO production and product release, confirming that the required data has been successfully processed by the destination system.

Let's examine the findings and recommended corrective action. The agent has completed all seven assessment tasks. Product master data and SOM checks passed, but the reporting check identified a distribution failure to the German national system. EMVS verification returned "Batch not found."

Based on this evidence, the agent classified the issue as NMVS distribution failure and recommended resubmitting the EMVS report. An authorized reporting user reviews the evidence, performs any required correction and resubmission, and verifies successful processing to the German NMVS before resolving the alert.

The agent provides the recommendation and supporting evidence. Corrective action remains with authorized personnel. Let's review the evidence behind this recommendation.

First, the agent verifies the product master data and confirms Germany as the intended market. It then compares the product, serial number, lot, and expiry information with SOM records. All four relationship checks match.

The reporting evidence identifies the issue. Product master data was accepted, but the product pack data report failed distribution to the German NMVS. No product pack update was submitted, which is not a separate failure. The empty disclosure history is consistent with the unsuccessful reporting to the national system.

The agent records these findings as part of the assessment. For Linkiva, the investigation pinpoints a reporting gap following CMO production and product release.

It gives authorized personnel the evidence and recommended next steps to correct the issue and verify successful processing before resolving the alert. This reduces manual investigation and helps teams address reporting issues that could affect product availability.

Shabbir: Now the product's in the market. It's on its way to the customer, but my job isn't done yet, because my job is to make sure that it reaches the customer on time and full.

I need to have the ability to track it through transportation and make sure that it gets to the port, the dock, or the actual warehouse itself, wherever it needs to land. Now we need to have the ability to track it through transportation and manage any sort of errors, etc., that could occur.

We'll see how we can do that through agents and the old platform as well.

Demo: Once products leave the CMO, tracking their progress and coordinating arrival can require significant manual effort. Delays can lead to costly disruptions, particularly when shipments sit at ports awaiting pickup.

Let's see how Linkiva monitors shipments, coordinates arrival, and assesses delivery performance. Production and quality checking at the CMO has finished, and Linkiva has received the ASN. The ASN identifies the products, quantities notified as shipped, and expected delivery date.

This gives the team the shipment context needed to prepare for arrival and track delivery against the order. After the CMO fulfills the order, the MAH receives carrier shipment status updates. As transportation updates arrive in MINT, Linkiva can track the shipment's expected arrival.

The configured hook evaluates the carrier shipment status. When the hook identifies that the expected delivery falls...

Shabbir: I'll just pause you for a second. What you can do in the agentic supply chain operating system is that you can create what returns all those hooks. What it does is it actually lets you look at any event and create a rule around that.

It says, "I've got this carrier status. Maybe I'm going to a carrier status. I want to evaluate it against that specific status with the actual expected delivery date. From that basis, then I'll know if it's going to be late."

If it's going to be late, then I want to raise an exception event that says this is going to be late. I'm actually taking information and putting a hook to process it so that I can have this richer understanding of, "I'm a late shipment coming."

Now I can do something about that. That's what this is doing. It's setting up that hook. You can set up a hook at any event with any sort of rules as you're creating your agentic supply chain operating model.

Demo: within the configured 24-hour arrival window, it generates an event. The action creates the arrival coordination work in POET. Linkiva and its partner can see what needs to happen, who owns the next step, and when it is due.

The work remains visible as the teams coordinate pickup or receiving and resolve any arrival exceptions. Coordinating arrival is just one part of achieving service levels. Linkiva also needs to know whether the shipment and receipt outcomes meet the agreed order requirements.

The bound action invokes the shipment and OTIF monitor agent. The shipment and OTIF monitor agent uses purchase orders, advance ship notices, and receiving advice or acceptance. It compares shipment information with the order and evaluates actual receipt timing, quantities, and condition when receiving evidence is available.

The workspace brings together delivery risks and exceptions requiring attention, so logistics and distribution teams can focus their review. We

Demo: now move forward to the delivery assessment.

The agent compares actual receipt timing, quantities, and condition with the order requirements. The assessment shows the on-time and in-full findings, any delivery exceptions, and the recommended follow-up.

Expanding the progress log shows the linked evidence and the checks behind each finding. The ASN explains what was notified as shipped. Receiving evidence supports what actually arrived and was accepted.

The exceptions are created in POET so the responsible teams can investigate service failures and coordinate corrective work with the partner. Together, transportation visibility, timely arrival coordination, and evidence-based OTIF assessment help Linkiva respond to delivery risks and improve service accountability.

Shabbir: Now I have the ability to track the shipment all the way through so I can improve my service levels. Next step is inventory monitoring. I've got inventory multiple, I've got 1,000 SKUs I'm working on. I've got inventory all over the place.

I've got different thresholds that I can set, and I can have alerts that kick off. Who's monitoring those alerts? Who's actually creating and looking at the inventory on a consistent basis?

The opportunity for me is that I really have more real-time reactiveness to changes in inventory, so that I can make better decisions on maybe certain orders I will place or some others I won't place, things I could put on delay, things I could communicate with my CMO on, and things I can communicate with my 3PL on.

That enables me to have more effective use of the inventory that I have currently available in the network.

Demo: Inventory monitoring helps Linkiva maintain sufficient stock while avoiding unnecessary excess. MINT provides visibility into inventory balances reported by CMO and 3PL partners with product, location, quantity, and reporting context.

Inventory balance transactions provide a snapshot of inventory by product and location, including the reported quantity and unit of measure. Inventory update transactions communicate inventory changes as materials and products move through the network.

MINT provides inventory reported by Linkiva's CMO and 3PL partners. Chorus extends that visibility to Linkiva's own warehouse locations by integrating to their WMS systems. We can search and view that external inventory through familiar OPUS screens.

Here, Linkiva's warehouse location has 1,500 units on hand, with 1,200 available. The inventory level monitor agent can retrieve this same data through Chorus and evaluate it alongside partner inventory from MINT, using the appropriate stock quantities, units, and reporting times to identify where planners need to act.

The inventory level monitor is scheduled to run every six hours. It reviews...

Shabbir: You can also have agents that are scheduled to run. You can say, "This agent's going to run every day." In this place, every six hours, I want to check inventory.

Demo: CMO and 3PL inventory evidence from MINT and fetches live warehouse inventory from Linkiva's WMS through Chorus. The agent compares each product and location against its defined minimum and maximum levels and logs confirmed shortage or excess exceptions without duplicates.

It recommends replenishment or rebalancing follow-up for planners to review. Humans approve the operational changes. The workspace brings together inventory items that require attention. Planners can review confirmed shortages and excess stock while missing, stale, or incompatible evidence remains not evaluated.

Let's review an inventory assessment to see the findings and recommended follow-up. For MR22, East 3PL is 270 cases above its maximum, while Central CMO is 140 cases below its minimum. The other locations remain within their defined ranges.

The assessment gives the planner a clear view of where follow-up is needed.

Shabbir: The key in that chart is actually pulling data from different systems. We have the MINT data coming in, but it's also looking in your own general warehouses as well to create that chart. One chart, I can't cross that here.

Demo: The expanded progress log shows which balances were read, how product, location, stock basis, and units were matched, and how each location was evaluated against its limits. This makes the findings traceable to their evidence.

The agent logs confirmed shortage or excess exceptions. Inventory planners receive the notification for the affected products and locations, supporting evidence, and recommended replenishment or rebalancing follow-up. Humans approve the operational changes.

The result is earlier visibility into stock imbalances, clearer follow-up, and better control of stock availability and working capital.

Shabbir: Almost there. I've got inventory that I'm tracking. Now I'm going to also make sure that I pay my suppliers on time. I've got to make sure I pay them the right amount.

I need to make sure that I do it efficiently, that I break down my cost from constantly reviewing every single line item, cross-checking everything. I understand. I heard you guys say that it could sometimes take tens of minutes to hours just to do all that final supporting information.

I want to deploy an agent that can be highly specific about everything it checks. I can be very diligent about following up on any of those particular items. I'm going to create an agent. It's my invoice processing agent.

Demo: At 5:00 PM, the accounts payable manager at the MAH reviews the invoice analyst agent before signing off for the day. A quick look at the agent shows 37 invoices fully processed today with every item resolved.

The weekly report confirms 200 invoices completed this past week and zero pending, leaving a clean workspace overnight. While the analyst is away, the agent operates continuously overnight, eliminating manual reconciliation backlogs.

Starting from yesterday's 37 completed items, a flurry of 17 incoming invoices hits the network. The agent ingests them instantly, pushing the active count to 49. Within minutes, it processes every transaction.

10 clean invoices are automatically verified and recommended for payment, while seven with line item exceptions are flagged with detailed evidence for human review. Overnight, the agent effortlessly evaluated all 17 invoices, delivering 10 to 100 times faster cycle times without manual effort.

At 9:00 AM, the accounts payable analyst logs in to start the day. Opening the 24-hour report, they immediately see a high-level breakdown of the 17 invoices processed by the agent, categorized by payment status.

The analyst is presented with a structured summary for each transaction. Rather than manually sorting through raw invoices, the agent guides the analyst directly to the high-priority items that require human attention.

The analyst opens an invoice recommended for payment. The agent begins by confirming foundational rule compliance, verifying that the PO is active and valid. Next, it performs a complete line item match across unit prices and quantities, validating both total amounts and term compliance.

Finally, it displays its step-by-step reasoning through visual charts before confirming its final pay disposition. Next, the analyst opens an invoice marked for hold. While three line items pass inspection, the agent isolates a price discrepancy on the fourth line item.

It highlights the variance in a visual comparison chart and logs the exact discrepancy details directly inside the decision log for human review. Next, we introduce the Procurement Investigation agentic profile.

Previously, the invoice agent generated a decision log flagging noncompliant line items requiring review. An OPUS event fires immediately, routing the flagged invoice directly to the procurement investigation agent.

Applying its IO TDR rules, the agent isolates the specific invoice discrepancies. It creates a procurement incident, logs the evidence, and assigns it to the CMO. Every OPUS agent is driven by a defined...

Shabbir: What you saw there was the ability for it to evaluate the invoices, have the ability to identify ones, then another agent picks it up. Now I'm going to collaborate to get this resolved and work with potentially a different manager to accomplish that.

Now what we'll do is go into the transparency that we provide around agents, because one of the challenges that I've heard is that you want to have much more clear understanding of exactly how much it's going to cost you to have these agents, and the ability for you to be able to predict that.

Based on the type of agent that you build, this is what the cost is going to be, let's say by a line item for a PO or an invoice, etc. What we've done is created full transparency around all the work the agent does.

Everything an agent does has an end-to-end audit trail. What you've been seeing is the main interactions with the human nature's progress log. You see how everything is done, how it's reading through things.

The work log identifies all the specific work it's done. This is what our pricing model is built on. It's just the work that we do, what the agent does. It's not on tokens. It's not on anything else that's hard to measure.

Then when you actually say, "Built an agent. It's going to do this much work. I have these many line items. I do these many POs on average, or invoices, or batch records," this is what the agent is going to cost you. That's what it's going to cost you.

The evaluation log actually gives you insight into what the agent did while it was processing. You can have inspection of exactly what the agent did in processing every single item from an inference.

That can be leveraged on an ongoing basis to make sure that you're monitoring and doing the correct thing, and that you don't need to modify that agent anyway to find any issues with it. Again, this is the way that you can have full transparency.

It's not some black box. You just close out and it runs, and you hope it does well. We provide you full transparency, full control, full accountability, and the ability for you to have a very defined set of metrics that you can actually see, as you see in the next set of...

Demo: business outcome. The agent's functional work is recorded in the progress log for analyst verification, while its systemic impact is tracked through granular operational and decision metrics.

Finally, the evaluation log captures system-level execution and timing, together forming complete agentic metadata that allows different personas to verify and govern agent performance.

The Worklog dashboard translates completed agentic tasks directly into transparent, value-based pricing. As teams of agents execute across our profiles, such as invoice verification and procurement investigation, the platform automatically measures work output across operational steps, decision...

Shabbir: The work log is based on the operations that the agent does, the actual decisions it makes, and the communication. This is effectively the work it's doing, as you can see exactly where the expenditure is going for that agent or any set of agents. You can accumulate that to full transparency of the national pricing model and predictability of it.

Demo: By aggregating these work metrics across all active profiles, organizations gain complete commercial transparency into the exact return on investment produced by their agent workforce.

Shabbir: I am now feeling confident that with my workforce being much more productive, I can address some of the issues with revenue, make sure I have product available. I make sure that I have reduced the risk from a quality management perspective, improved the productivity, and better supplier management.

I make sure that my market access is always going to be on time so that I don't have any issues there to resolve them. I make sure I track my service levels all the way down to my customer, constantly monitoring the inventory, and making sure that my suppliers get paid on time and with the right amount.

What do you guys think? Thank you, Mr. Narrator. The team did an amazing job. I'm going to take this hat off because I don't want your job. It's too hard, honestly.

It's just amazing to me how you're able to be so successful in such a complex world. You look at the job that all of you do. You have so much uncertainty to work with, so many different variables, so many different SKUs in the hundreds to thousands, sometimes 10,000, partners in the tens or hundreds, scarcity of information, scarcity of people to actually do their job.

I think that this opportunity, in my perspective, is the biggest opportunity in agentic, because I think supply chains, through digitalization and through agentic, can really drive significant benefits to humanity. We will have better products and lower cost with more effectiveness, and I think we'll have a happier workforce.

Otherwise, as we all know, I work with all of you. It's challenging. I remember I only named the company as a person, but whenever I used to visit head of supply chain, it was like locked in a room on the phone all the time, dealing with all the emergencies.

"OK, this has to clear customs, or this needs to be expedited, or that batch is released late," and it's all through emails. Then trying to chase people down and what's happening here. We can make the world a lot better than that. We can make the world significantly better.

We can improve what the business does from a revenue perspective. I think that's really the opportunity that we all have together. We've been working incredibly hard. We are dedicated to this supply chain, and we've been working feverishly, first building the office platform as a unique multi-enterprise platform.

Creating an integrate once capability that is scalable, that can deal with the variety of transactions, hundreds of thousands, with all the different systems, delivering on MINT from a transaction perspective and supporting all end-to-end transactions, powered from a no-code perspective.

A scalable open support and dashboards, now the lakehouse capability for us to be able to do analytics on that. The ability to also now have governed agents that are truly reliable, trustable, trusted teammates. The ability to have the no-code agent profiles, which can be very extensive.

It's not simple. What people's jobs are in the works are not simple. Being able to combine that and being able to create multiple agents off of that. Now, of course, you can access any system. The agent can access any system. You have visibility into that.

Now you have an agentic business process that encapsulates all that. We had it all put together, but it's going to be a journey to get it all there. Obviously, at the heart of it is the integrate once and the ability for us to bring all the partners in from an integration perspective.

Make sure that no matter how they want to integrate or how they want to have the ability to work with you or work across your different systems as they evolve, as you go from ACC to HANA to who knows what else next.

You can integrate to any of your other systems, can access or anything else, and integrate with your partners. Like I mentioned, with MINT, you have the full range of transactions, because a supply chain is a network.

It's a network of different types of nodes and the ability for any party to participate. Many of you are pharma companies. Some of you are from 3PLs. Some of you are CMOs. We have many customers in the retail supply chain.

All different perspectives have to be incorporated and making sure we understand the transactions that flow in each direction and the needs of each of those parties. We addressed all of that through MINT with all three to three transactions.

The adoption is going very well. We've been working very hard. It's like laying roads and bridges and all that. We know we have to do it because without it, we can't move forward.

We've been successful in driving up the number of customers, the number of links, and from an activity perspective, that's also driving up. It is across the whole supply chain. We have customers in every segment and participants in every segment of the supply chain.

The amount of transactions is 3x, integration up 16x, MINT links up 13x. It's accelerating. I'm actually finding that I think these numbers will continue to get better as we go every three months, six months, nine months, because I believe that we've all started to see the actual benefits of doing this.

With Chorus, we have the ability to integrate any system. You have the ability to create these rich metadata-defined objects. You can have an object and read, "This is my contract object. It is the one way I look at contracts, but I'm inputting multiple different systems."

Chorus will say, "For this particular object, I'll go to six different systems, pull out information, present information, and consolidated views." You want query, you can do that. If you want to change any one of those items, it will go back to the right system and update it.

You can also pull information from any external Internet system as well. Sometimes you have to go monitor airline flights, but you have some API to go to system and pull that information down.

Chorus can do the full end-to-end set of integrations to any real-time information that needs to be pulled in, and the agents will work across any particular systems out there.

We know that 64% of enterprises run multiple ERPs. Some lucky few can run one ERP plus everything. The majority of the industry has multiple ERPs. 15% to 20% only rely on one ERP.

There are about 18 daily apps that you're interacting with. You have to literally go log in, pull, pull, pull to make a decision. Here with Chorus, you can have one application, one platform to reach out to any of those systems to pull information in, in the right context, to make the right decisions.

You can do it. Agents can do it. You can all work together on it. With POET, you have the ability to define any sort of business process you want. You can compose your processes from any sort of workflows you're going to need.

You can create a very rich set of structured information with rich definitions of your data. It's highly well defined and structured. You can have work items with due dates you can track, and they can monitor, and they can fire if anything's going to be late.

You can create a list of work items and items. Now we just introduced knowledge bases. You can take the rich content information you have on hard work and other aspects, other SOPs, body agreements, and preliminary knowledge base, and make it available for agents to also access as well.

You can leverage that knowledge and operate on that knowledge in its actual influencing as well. Now that enables you to have your ability to have both the collaborative composable processes, as well as with the information and data, all well defined.

Again, the ability for you to not only have age time, but also have analytics on those processes as well. You've got the ability to have, across the board, all these processes that are available in the point catalog.

It's growing at a very rapid rate, but it's just a starting point. You can go ahead and take these processes and customize them, configure them, create new ones, be innovative if you want. You can share them, and we'll put them in the marketplace for others to use as well and attribute it to you, obviously.

It's a rich environment for you to continue to do that. You have agents now. The agents are able to be completely governed. You can trust them. They're verified to do the exact work that you need.

They're going to be fast. They're going to be reliable and available. They were predictable, the cost on the work it does. You have the ability to have that agents now that work inside of the full range of work that you wanted to do, defined however you want to create your operating model.

We provide you with a set of agents in the catalog. This is a growing set. We're building very quickly. These aren't simple ones. We're actually building them by talking to you, understanding, and then being able to put them in the catalog so you can have a starting point for your agent.

You can use it right out of the box. There's no additional cost to that. If you want to modify it to your work instructions, your SOPs, you can do that. Obviously, this is a no-code agent platform. You can create any agent you like.

You're not limited in any way. It's not like, "We've got 20 agents now. We've got five agents now." No. These are your starting points. You can create as many agents as you like. Again, it's up to you as to how you want to define your processes.

You have the ability to have these events that occur. These events, like I said, like with the operating system, these events that occur throughout. They can be object events. They can be hooks that you want. They're trapping on. They can be scheduled events.

It's a live, vibrant environment in which agents and yourself work together. As these events occur, you can all react in real time and address it. Multiple agents, just like most of the humans, can listen for these events and can act on these events simultaneously, because oftentimes the single event of a relationship that many people have to act on them.

Many agents can act on them simultaneously. Just MINT alone has so many different events that are going on. 2,900 you can trap on at different points in the process. You can pick the right process time if you want to trap on these events and then be able to act on them however you see fit.

The richness of those events is the richness of your reactivity and the points of time that you want to actually take action. You've also got the actual lakehouse.

The first set of data, obviously, is the track and trace data that we put into the track and trace lakehouse for you to be able to use very high-volume reporting, scheduled reports, and dashboards.

We're also putting all the other information for you as well. We're going to, in a Q1 time period, put the MINT and POET data and other data so that you can now leverage the lakehouse capabilities to create rich analytics that you can use in making decisions, and agents can use it to make decisions as well.

Now that really enables you to have this full fabric on an end-to-end basis with the EPCIS information, and then we'll layer on top of that all the other information as well. You can have deep analytics. You have to go build a separate lakehouse. It's built in with metadata, with richness.

You just go build that up yourself. You have the open reports and dashboards that support access to empty stores and to integrate and join the meaning of the data.

You saw the actual ability to go from POET to MINT to Chorus. You can create reports and join across all of those to pull together information you want from any source. You can do it all in a no-code manner, so you don't have to sit there and script it.

They also provide you SQL if you want. If you want the full bridges of SQL, you can also use SQL. It's enterprise scale, so this is not like one of those systems we had 2,000 rows. We support millions of rows. We've tested up to 20 million rows in a specific dashboard as well.

Then we could have AI agents also access this information from a reports dashboard and reports perspective. On top of that, what you saw was reports were agentic reports. That actual agent created those reports on the fly leveraging ORD, the OPUS reports and dashboards, in agentic mode.

We've actually created an ability for the agents to tap into the data set with full permissions only for whatever data they have access to, to generate those reports on the fly in objectic capability.

That's an area that you'll see us investing more in so that you can have a gentle interaction to create the type of analytics you want, all the full data set that you have.

Again, we provide a rich set of reports and dashboards for all the different orchestrations. We continue to grow these because we make them no code, so it's fairly quick for us to build. You can also create new ones, put them in your company, users can create them, and you can also share them across your company as well.

You have full analytics there. We also have agentic business processes at agents. We're creating agentic business processes and starting to put them in the catalog as well. You can pull it together and say, "The production tracking process, I want that."

It has these agent profiles, these MINT transactions, this process, and I want to deploy that. That gives you the starting point, and you can modify it as well. It gives you the full building blocks you need to create the agentic supply chain operating model that fits your company, fits your needs, and evolves with you.

Our real goal here is that what we're creating is a full anti energetic business network, because what you see in the confluence of all that is that no matter who you are, your supplier, CMO, your manufacturer, your logistics provider, transportation provider, we support you.

We have customers across all of these areas. We have customers that are some of the largest pharmacies, hospitals. They want to become champion. All of what you saw is available to them for themselves and for them to work with you on.

Same thing was for wholesale, transportation, and logistics. There is no limit to who the customer can be. The reason why that's important for you is that if they're able to advance their ability to get digital, their ability to get agentic, it makes all of our lives, and your lives specifically, better because they're more effective and more efficient in working with you.

They're able to get to this, enter an agentic business network. As we now conclude, how do you get started? How do you actually get started with this process?

What we've created is this objecting accelerator program. Because everything is no code, because everything is accessible and available, we see that there's lots of opportunities.

We can work with you to identify an opportunity. We can align on that opportunity, and within a 24-hour or further, it can be done over 30 days, can be done over two days. However quickly we want to do it, we can take a specific process and basically identify it.

You can see it working in a tabletop environment, so you can have your own sense of, "This is what I want to do." It's a very boxed-in capability you want, because I think it's the best way to start. As that works, then you do more and more and more.

What we see is the ability for you to start small. It's not like you have to boil the whole ocean, create a $10 million program to actually do this. You can start small and grow.

The opportunity is really for a whole ecosystem to be created around this, because we, TraceLink, are not the subject matter expert in your processes. We are working very closely with our partners, many of whom are here.

What we recognize is technology is only 1x the actual work that has been done. 3x of the work is at the process definition and the ongoing work that you have to do to find the right process. 5x is actually the change management that needs to occur, because you need to have your training, your SOPs, and work together.

We are working very closely with the partners in creating and educating them. Many are going through training right now on MINT, COVID, and so on agents in order for them to come up to speed.

What we've created is, Caitlin has called it, the Gentry Conservatory. It's actually a dedicated area within our offices in Massachusetts. We can come in, and we specifically created this area to build agents, to build agentic business processes, and to be able to learn, to then rehearse and perform and compose and graduate from it.

We're bringing partners in. We're bringing partners in with customers. We'll have stations where everything's set up. We have collaboration areas for us to work together with you to be able to achieve this.

We're fully invested in making this happen and working with you however you want to work with us in order to achieve success. We believe it takes a village. It takes partners. It takes us. It takes you for us to all work together to achieve this.

We're going to invest into it to the level necessary to make sure that everyone's successful. As we know, the companies that transform first become the leaders of tomorrow. Moving in an intelligent, deliberate, but very meaningful manner with some haste is going to be critical for your success or everybody else's success.

We see that there's opportunity to grow the business. We're not looking at cutting costs. We're actually looking at driving profits, lowering costs, improving their actual revenue and the top line. That's a real opportunity. It's not about cutting costs. We look together, and we actually look to work with you on this.

Now, we're doing something unique. It's a little out of character for us, to be honest with you, but I think the times need it. In order to support and encourage creativity, because again, the reason why we picked OPUS as a word, why we use all these creative terms like poem and so forth, is that we think the opportunity in this new world, of this agentic world, is to really open your mind.

Try to drive creativity across the industry, multiple industries, multiple different types of players, to come up with unique ways of creating agentic business processes. We are issuing a challenge to the industry to create the most valuable agentic business process.

In that challenge, what we're seeing is that we have created Link Labs, which is available to anyone. You can go in there in a safe zone. We'll give you a company. You can go create an objective business process. You can make it as meaningful as you like, and the more meaningful you make it, the higher you score.

The people that score the highest win. It's winner take all. That means that there's opportunity for you as a company, as an individual, as a team, working with your partners, working with even colleges, because we need to get creative about how we can identify the anti supply chain.

The prize is a winner-take-all prize. We will select through third-party judges, who will select the winner, and the prize is a billion dollars. Hopefully, this encourages people. I see people already taking notes in the back. I know him. He'll be all over it.

This is an opportunity. We're here to invest into this. We really believe in this opportunity. We feel that this type of encouragement, this type of creativity, creating a lab area that everyone can create and then be able to demonstrate.

Then we intend to announce the winner at our next FutureLink. The process will kick off in January. We'll then register your interest. When she will register your interest, as police information will be done in January, we'll keep you updated.

Thank you very much. I hope this was useful.