Arun Rajan
Analyst · Ken Hoexter from Bank of America
Thanks, Michael, and good afternoon, everyone. As Dave and Michael described, we continue to execute our disciplined strategy, delivering for our customers and carriers while scaling several innovations that better serve our customers and widen our competitive moat. At the center of these efforts is our lean AI strategy, which combines our lean operating model with deep industry expertise and our proprietary custom-built AI agents embedded directly into the workflows within our quote-to-cash lifecycle. This strategy enables us to automate, scale and execute in a sustainable and repeatable way without letting external narratives blur the difference between perception and reality. We take a highly focused and disciplined approach to AI deployment, and there is no hobby AI at Robinson. We deploy AI where it delivers real-world results and measurable outcomes that show up in our P&L. We prioritize our efforts based on ROI, leveraging extensive instrumentation to identify the most manual and high-friction work and then scale our AI capabilities within our existing technology spend. Access to AI itself is not a differentiator and anyone can say they're using AI. But what matters is how AI is engineered, operationalized and scaled. And AI is only as effective as the data and context that powers it. Part of our competitive advantage comes from the scale, scope, depth and proprietary nature of our data and context, which I'll explain shortly. We combine that with a disciplined operating model that allows our tech to be continuously operationalized and improved. Before going deeper, it is worth grounding how AI works at a high level. In the AI ecosystem, there are broadly 3 layers. At the foundation is infrastructure, which provides compute and storage. Above that are AI models, which are increasingly accessible to everyone. Neither of these layers provides a durable competitive moat. The real differentiation and advantage exists at the third layer, which is the application layer. At C.H. Robinson, we own our application layer. This is where the benefits of AI come to life when deployed correctly and how we deploy AI agents is another source of our competitive advantage. C.H. Robinson's builder culture produced our proprietary transportation management system and an extensive application stack, including advanced AI and machine learning capabilities that sit on top of that. That same culture now enables us to design, build and deploy fit-for-purpose AI agents that drive value for the customer, carrier and Robinson. With more than 450 in-house engineers and data scientists who have domain expertise and deeply understand our business, we're able to deploy agents faster and with greater control than a buy and integrate model that relies on stitching together third-party solutions that are generic and lack the data set in context that represent the scale, complexity and nuance of our business and the industry. Our unmatched scale, proprietary systems and deep logistics expertise provide the data, context and human-in-the-loop oversight that makes our AI agents more effective, more reliable and more difficult to replicate. Our data and context advantage spans multiple modes such as dry van, flatbed, temp control, ocean and air. They also span multiple services such as short-haul, drop trailer, cross-border, expedited and customs as well as multiple geographies, customers and lanes. This level of granular disaggregated data cannot be purchased. And this depth of data, such as data on individual warehouses enables us to understand price and cost dynamics better than anyone in the industry. Scale, scope and depth of the context that we provide to our custom-built AI agents is also part of our moat and competitive advantage. Through our human-in-the-loop process and extensive instrumentation, we collect institutional knowledge from workflows and tribal knowledge from our freight experts into a context layer that enables our AI agents to execute and continuously improve alongside our expert logisticians. In effect, our people teach our AI agents in the same way they would train a new operations employee. Routine work can then be executed autonomously, allowing our teams to handle nonroutine surges in volume and higher value, more strategic activities for our customers. For example, think about appointment automation and the breadth of customers, freight dimensions and dock management systems we deal with. Every one of these customers, dimensions and locations has policies and nuances that are known to the appointment agent by way of an engineered context layer. Economically, this model scales efficiently. After the initial build and implementation, our marginal costs are very low. The ongoing costs are primarily tied to AI token usage rather than having to pay by transaction to a Software-as-a-Service provider. So owning the technology and engineering it in such a way that we have a scalable model is a critical component to widening our competitive moat. Our build model is also important from a speed of implementation perspective. If a company is using multiple third-party providers to create and implement AI agents, they are beholden to that external provider who doesn't know the business as well. With our builder culture, we're leveraging the vast domain expertise of our in-house team. Since we're building our own AI agents, we have more control over the implementation process and the speed of integrating those AI agents. That faster speed to ideate, build, operationalize and scale our AI agents is a differentiator, and it's showing up in our outperformance. Our fleet of AI agents is growing quickly as we continue to pioneer new ways to automate manual tasks and supercharge our industry-leading freight experts to solve for complexity and deliver high-quality service and outcomes to our customers and carriers. We continue to leverage and scale the use of Lean AI to power new capabilities that are backed by our unmatched data, scale and context, and we are continuing to disrupt from within. Agentic AI operates with a degree of autonomy and unpredictability, making its progress nonlinear and requiring ongoing human-in-the-loop oversight as it advances through cycles of progress and retrenchment. Our Lean AI process of discovering, learning and building where missteps and resulting learnings are milestones is not only necessary, but is the best path to uncover what truly works. Continued improvements of our service through cost-efficient AI task agents that listen, learn and act all day, every day enables us to deliver fast, accurate and personalized service at scale and in any market. We have a clear view of both what has been built and what remains ahead, and we are still in the early innings of our transformation. There is significant runway across our business to continue scaling AI agents, and we've automated only a fraction of the hundreds of processes and subprocesses that exist across the quote-to-cash lifecycle of an order. As Dave said, our strategy is focused on building the best model for demonstrable outgrowth while continuing to have industry-leading operating margins. Our technology is unmatched, and we will continue to disrupt ourselves to stay at the forefront of the AI revolution and to further widen our competitive moats. With that, I'll turn the call over to Damon for a review of our first quarter results.