What does it take to bring AI into one of the most complex supply chains in the world - where long production cycles, global regulations, and thousands of product combinations are the norm?

At Syngenta, the challenge falls to Brandon Leander, Head of IT & Digital, who believes it starts with education, experimentation, and setting realistic expectations. Leander’s approach offers manufacturing and operations leaders wrestling with demand volatility, labor constraints, and global complexity a grounded look at how AI can deliver real value without overpromising.

As Syngenta explores agentic AI in its seed business, Brandon is focused on using intelligent agents as companions first, augmenting decision-making rather than fully automating it.

Doug Power, Event Director: What was the initial catalyst for Syngenta’s push towards agentic AI in the supply chain?

Brandon Leander: For me, the spark really came at the Connected Worker Conference last fall. I was listening to Michael Carroll and Ron Norris share Georgia-Pacific’s journey with causal AI, and it immediately clicked with some of the challenges we’d been talking about in our own supply chain. Over the following months, through more conversations with them at different Connected Worker events, the idea kept building. That’s what set me on the path of exploring how we could apply agentic AI within Syngenta.

Doug Power, Event Director: Can you walk us through one of the most impactful use cases? I know it’s still early, but what business value are you already seeing, or do you expect to see?

Brandon Leander: One of our biggest focus areas right now is the Demand Confirmation Tool. Today, it pulls information from SAP and IBP, things like demand forecasts and inventory, and then our planners manually work through it. We’re talking about over 2,000 seed varieties, which adds up to around 35,000 product and plant combinations. It’s a heavy lift. Right now, it takes about 20 days to get through that process. By the time the planning team confirms demand with the sales team, the data is already stale. That’s the problem we’re trying to solve. With agentic AI, we want to move beyond just confirming supply availability. If there isn’t enough supply, the agent can help trigger the process to determine when the product will be ready. If demand outstrips what we have, it can generate allocation recommendations based on forecasts and our business rules.

The real value here is protecting revenue and avoiding lost sales. If sales doesn’t know a product isn’t going to be available, that creates risk. There will be some labor savings too, but our main driver is giving the sales team timely visibility so they can keep selling at the pace we need. Looking ahead, we see another major opportunity in production planning. Seed production has long cycles and we can produce in different regions around the world. We want agents that can run scenarios, weigh factors like cost, lead time, quality, and regulations, and recommend the best plan to both optimize the supply chain and improve delivery to customers.

Doug Power, Event Director: How are you mapping AI tools to specific roles like business planners, crop strategists, and supply planners?

Brandon Leander: That’s probably the fun…and the challenge, right now. Everybody is coming out with their own AI tool. Almost every one of our technology partners has something with AI built in. What we’re doing is building a foundation for a Syngenta AI platform we can expand on. We’ve got strategic partnerships with AWS and Databricks, and we’re also a Salesforce shop, so we’re looking at what those platforms can offer. At the same time, we’re exploring custom builds and even some low-code agent solutions.

The point is, we don’t want to lock ourselves into one partner. Yes, we have strategic relationships, but there are a lot of other companies bringing good ideas, and we want to learn from all of them.

Doug Power, Event Director: So, how do you identify the right partners for such a complex operating environment?

Brandon Leander: Fortunately, I’ve got strong technical partners here at Syngenta who’ve been working in AI for a while. We’ve also been exploring agentic AI ourselves for the past 6 to 12 months. So I lean heavily on our technical team. And the truth is, we’re all learning together. Even with AWS, we’re all still figuring it out. What matters most is which partners are willing to learn, adapt, and collaborate. We see a future where agents will need to interact with other agents - Salesforce talking to a demand planning agent, which talks to an IBP agent, and so on. So we need partners who embrace open standards and can work well with others. That’s a big part of our selection process.

Doug Power, Event Director: What does the interaction between users, like business planners or crop strategists and their intelligent agent companions look like in practice?

Brandon Leander: We’re approaching them as companions and that's where we're using either a recommendation engine or a planning companion. AI is so new, and we want to be cautious. We’ve been talking about the idea of augmentation versus automation. Maybe in the future we get to automation, but right now we’re focused on augmentation. So the question is: how can this tool reduce some of the manual work they’re doing? How can it take on the menial tasks and then provide recommendations? There’s still a lot of business intelligence that goes into these decisions. As they use it, the companion agent will get better and then we will get to full automation. Our leadership team is comfortable automating the bottom 20% of our portfolio, taking an augmented approach with the middle 60%, and keeping the top 20%, because it’s high value, pretty hands-on. That may shift as they see the quality and speed of the agents. It could be that the top 80% is augmented and the bottom 20% is automated. We’ll explore that as we go further down the journey.

Doug Power, Event Director: You often hear horror stories about user adoption when it comes to connected worker platforms. Have you encountered any concerns from your workforce around AI adoption, and how are you addressing them?

Brandon Leander: We haven’t gotten specific feedback yet, because it’s not real for people. Right now it’s more like, “Hey, this could be coming.” I’m working with our communications team to explain what we’re doing with AI and why. In our case, we’re trying to grow the top line, but we’re expected to keep the same size workforce. That’s part of the story we’re telling: we want to grow, and AI helps us do that. At some point, though, employees are going to realize: if this automates part of my job, does the company still need as many of us? We haven’t hit that inflection point yet, but it will come. And when it does, it’ll come down to how we take care of people as an organization. Are there opportunities for them in other roles? So for now, it’s really about communicating the purpose. Yes, growth, but also efficiency. And with efficiency, the reality is it eventually touches headcount.

Doug Power, Event Director: As you move from pilots to delivery and scaling, what challenges do you anticipate?

Brandon Leander: We sell all over the world and we have multiple product lines, or species, as we call them. So the plan is to start small: a couple of species, a couple of geographies. Learn there, then share those learnings across the organization. We’re not trying to hit a home run on the first at-bat. Some of our other use cases are much bigger, but right now we want some wins. The thing I’m a little apprehensive about is the quality of the AI agents. In my experience, when a computer does something, the organization expects it to be 100% right. When, in reality, the human may have only been 75% right. Is the computer better than the human? That’s not always the standard we’re judged on. So it’s about being realistic. First-phase agents may be mediocre. This isn’t magic. Like I’ve said before, AI is about understanding inputs, learning how the best people would execute, and then letting the agent grow from there. What I’m telling our team is: don’t expect us to flip a switch and suddenly reduce headcount by 50% in a role. That’s just not realistic.

Doug Power, Event Director: What lessons have you learned that you can share with other manufacturing or supply chain leaders considering this path?

Brandon Leander: The first thing is education. Educating your business leaders on what agentic AI actually is and how you tell a story that resonates with them. The second is partnering with your data teams. Whether that’s Databricks, Snowflake, AWS, whoever it may be, you need to understand your platforms. Just yesterday I was working with someone on our team who’s building an AI agent, and he was showing me how much more contextualization we need in our data lake for the agents to really work. He walked me through an example where, the further into the dataset the agent got, the more accurate it became.

But when we tested it again from the beginning, it had actually forgotten what it learned earlier. So it wasn’t as accurate from start to finish as it looked in the middle. That’s the kind of thing you need to be realistic about. So my advice would be: look at a couple of different partners who really understand this space, keep leadership expectations in check, and don’t oversell what these tools can do right now. We are still very early in the game.

One of my favorite quotes sums it up: “It’s not what you don’t know that gets you in trouble—it’s what you think you know that just isn’t so.” You have to stay humble, be willing to learn, and work with good partners.

Doug Power, Event Director: You’ll be speaking at the Connected Worker event in October. What are you most looking forward to at the event?

Brandon Leandor: I’m really looking forward to hearing from other companies about what they’re doing and the lessons they’ve learned. Like I said at the beginning, hearing from Georgia-Pacific’s Ron Norris and Michael Carroll at past Connected Worker events really kicked off our own journey. That kind of networking and learning has been invaluable for me. So for those coming this year, I think it’s a great opportunity to see what other companies are building, what the technology providers are bringing forward, and to make those connections. For me personally, it’s been a great experience over the past 18 months, and I’m excited to continue that.

Want to hear more from digital manufacturing leaders like Brandon?

Join us at the Connected Worker Summit in Chicago, October 7–10, for real-world case studies, workshops, and practical guidance from pioneers in digital manufacturing. Brandon will be there too, drawing on his experience at Syngenta to share how agentic AI is being applied to one of the world’s most complex supply chains. His session will explore how AI companions can support planners, improve visibility for sales, and set the stage for scaling while keeping people at the center of decision-making.

Download the agenda to see the full speaker lineup and what’s in store.