At Citi. He's going to be asking the smart tech questions. We're thrilled to welcome C. H. Robinson to our conference. We have Dave Bozeman, CEO, Damon Lee, CFO, and Arun Rajan, Chief Strategy and Innovation Officer, informally CTO, I suppose is fair. It's interesting, just before we started, Dave and I were talking about the idea of a transportation logistics company at a tech conference might be a bit strange to some folks, yet those who are familiar with the C. H. Robinson story certainly would not be surprised by it. Just looking back at the last couple of years, if we think about your North American Surface Transportation segment, we've seen gross profit per employee up more than 60%. Headcount is down roughly 30%. Shipments per employee have increased at a double-digit rate.
There's really been some remarkable things that you guys have been able to do with technology and leveraging AI, and specifically you guys refer to Lean AI. Maybe you could discuss that just as an introduction. What are the changes that you've introduced that have been able to drive those gains? Specifically, it would help, I think, if you gave some specific examples around what have been the challenges that you've had to overcome to get to that.
No, yeah. Thanks, [Ari], and happy to be here at your conference. Thanks for having us. For us, as we talked before, it's not strange to us being here because we know the story and the transformation we've done at Robinson over the last three years. I'll just frame it and then have Damon and Arun get into it because it is a story. You talked about Lean AI. What we've done is introduced a transformation of a 120-year company to essentially make it a disruptor within this industry. We've done it in two vectors. One, a very different lean operating model based on lean principles. This is about continuous improvement, problem-solving, pace, speed, all things that have been around, but we've introduced into this company and I think also have been pivoting this industry.
That has really opened up and allowed our technology, which we are builders, not buyers, and always have been. It allowed our technology and our data set, which is pretty proprietary and the largest in the industry, to really come alive, to make that data intelligent. Between our Lean AI, the best technology, we think, in the industry, and our people, of course, which are some of the best logisticians in the world. Those three have been symbiotic in the system we run at Robinson and has really allowed some of those gains that you just talked about. That is what has driven it. Let us get into the technology in particular, and also our operating culture, which drives bottom-line results. We can do that. Damon, Arun?
Yeah. Arun, why don't you talk about the tech, and I will wrap it all up.
Yep. From a technology perspective, like you noted, we have improved our productivity by 60% over the past four years. It is sort of this journey of we worked backwards from what would a tech company do to disrupt this industry, right? When you work backwards from that, you say, well, a typical tech company would say, well, how do we decouple headcount growth from volume growth, right? Which is to make this a very scalable business, right? Some of us came from Amazon. That is the model you would do. The first thing you do then is okay, well, how do you engineer the system to ensure that it is scalable, effectively decoupling headcount growth from volume growth? We have been on that journey for multiple years now. And you see the 60% number.
One thing that has changed materially over the last couple of years is we used traditional software engineering approaches for the, I have been at the company for five years, but the first two or three years were more traditional software engineering approaches.
Over the last couple of years, in the past, a lot of projects would not make the ROI bar because there is a lot of software engineering effort. This industry is plagued by nuances that are specific to customers, that are specific to carriers because we sit in the marketplace, right? We sit in the middle between capacity and customers. Unlike a peer-to-peer marketplace, this is B2B marketplace, which has lots of nuances. And we sit in the middle of that and a lot of these projects did not make the bar two or three years ago. But with AI, now all of a sudden, engineering is massively better.
We have 500 engineers, but those engineers punch like they are 2,000 or 3,000 engineers as they start to use coding tools. That is on one side. The other side of it is we have engineered a platform, an agentic AI platform, that basically allows us to capture the collective knowledge of the company, the context of the company into our agentic AI platforms. All of a sudden, you are not building software the old way. We are not engineering every single rule or every single SOP into software. It is in a context layer. In a way, the company is being programmed in English. It is probably the simplest way to think about it with the advent of LLMs. This agentic harness we built on top of the LLMs has created this massive acceleration in our productivity in the last couple of years, combined with the lean operating model.
That is one side of it. That is the productivity side of it. The other side of it is, as a marketplace, what do you do? There is pricing on one side, how you price to customers, and then on the costing and the capacity side is how you procure capacity efficiently. In the marketplace, then it is like you make the spread. This notion of personalized pricing that drives our pricing algorithm. We have a machine learning-based algorithmic pricing, which we have had for 10 years. But as the data compounds and grows, this is traditional AI. I will call it classical AI machine learning that effectively drives our gross margins by driving the right pricing for the right customer for the right amounts of value we deliver.
Likewise, on the capacity side, it is about cost discovery and how we procure the capacity at the lowest price, the lowest cost to us, to put on a particular load. Again, if I was to step back and sort of summarize what I said, which is, well, if we were trying to disrupt C. H. Robinson as a tech company, what would you do? You would create this scalable model to drive down to unit economics, and then you would drive this intelligent pricing and costing discovery to drive the best sort of gross margin number. And that then compounds, ultimately, what do you do with the lower cost to serve, like we have accomplished? You parlay that into growth by driving prices lower.
That is the sort of formula for driving our business. Lower cost to serve, higher gross margins, but you parlay some of it back into growth by lowering prices to customers.
Just to put a bow on what Dave and Arun said, since the end of 2022, we've not only benefited with 60% productivity since the end of that year, but we've also seen demonstrable revenue growth, we've seen demonstrable gross margin expansion, and certainly the operating margin expansion we've realized has been on the back of the productivity. But for us, AI, Lean AI specifically, it's more than just productivity. We see revenue growth, we see revenue management capabilities, so think better price, better procurement of freight as well as productivity. We see the benefits of our Lean AI strategy up and down the entire P&L. We know what we're doing is very defensible. We have 450 engineers that build our own tech. This is custom tech for our company's specific problems.
Very difficult to be able to go off the shelf and try to replicate what we've done. We've estimated that you would have to partner with maybe 15- 20 individual AI platform companies to replicate what we've done. Even once you did that, you're getting a generic solution set for a company's specific set of problems and opportunities versus our customized solutions. I think maybe the most important thing outside of that is the cost.
Our marginal cost of ownership once we build an agent is close to zero. Whereas if you're using somebody else's third-party tech, you're going to pay by the drink every single time you use their technology. We've generated hundreds of millions of dollars of operating income value since the end of 2022. On an annualized basis, we spend less than $1.2 million on tokens. Just soak that in. In an ecosystem on an AI where you cannot find a company that's generated a positive ROI from their investment in AI, we've generated hundreds of millions of dollars of value since the end of 2022 with a very immaterial amount of token investment on an annual basis.
I think for this audience, from an investment perspective, this has been a structural change. Everything that Arun and Damon just said, we are very, very purposeful about where we put the technology, and that is embedding it into workflows. For us, it was the order-to-cash workflow that lent itself very much coming from machine learning into generative AI, and now we are actually doing agentic within our other business that will come all around. That has allowed us to automate, essentially, our back end or operational type of roles that it does not matter if the market takes off or if the market stays lower for longer, this system is now going to be structural in that change. We will not add in a number of humans in this kind of order-to-cash type of process. We always talk about transactional quotes.
We have a mature agent that is doing our transactional quotes that come in with email. Used to only get to 60%, now we do 100% quoting. We do it in 31 seconds. It used to take 17- 20 minutes. We do it back in a conversational manner. The point on all of that has allowed us to win more freight, see more freight. It also says that if we are doing 600,000 of those quotes, you can add a zero, we will do 6 million, and we will not be adding humans to that because that is a mature agent that is placed within that realm and that workflow. I know that was a long answer, but it covered a lot to show why this is symbiotic and why this is structural from an investment perspective, and we feel good about that story.
Yeah. No, that is a great overview and really impressive productivity growth over a short period of time. Maybe diving in a little bit more on the technology stack. It sounds like a lot of this is homegrown, purpose-built for Robinson. How do you kind of see the-- as investors here are looking at technology companies, who is gaining wallet share? Is it mostly the infrastructure providers? Who is losing wallet share if there are technology vendors from that perspective? Then, one of the interesting things you said was just $1.2 million of annualized token costs. I am sure some other companies in the Valley would like to be that low in terms of token costs, what we hear. What are you using? Is it the frontier? Do you go more the open source route? Would love for you to hit on that as well.
Yeah. No, great question. I would say strategic partnerships, or we are a builder culture. Which means we build our software, which means we will evolve to be an AI-native logistics company, right? The question is, well, what are the underlying infrastructure providers we use? We have long had a partnership with Microsoft. They are cloud platform, and they are also the access to multiple models, right? Through Azure, we can access multiple LLMs. We have a partnership with Snowflake. Again, I think of these as infrastructure players.
The real sort of value that is coming from AI is from our custom-built harness that sits on top of the LLMs, right? Think of it as, we have the ability to route a given workload to any LLM, right? We could use a frontier model if it is a complex reasoning problem. But if it is a simple transactional thing, we can use open source, or we can use an older version of the models. We have a router that will route the work based on the sort of complexity of the work to the appropriate LLM. In some cases, we just host our own open source LLMs. The bottom line is, in the end, because we are a custom-built shop and we have our own harness, and we built these agentic workflows, the way we architect our platform kind of accrues most of the value to us, right?
Certainly, there is some token costs that OpenAI probably gets a giant share, that $1.2 million token spend that Damon described. But there is a bunch of volume going to open source. Over time, we will continue to route the simpler workloads to open source models, which is why the costs are so low. The question is like, well, how do you do it? Just like I think of this as you kind of roll back 15- 20 years to the cloud, and when the cloud first came out, I think you had this tendency of people to sort of say, "Oh, the cloud is here. I do not have to provision hardware in my data center, so I can just spin up this instance in the cloud." Spending went crazy, right?
I think the same thing is happening with LLMs, where it is like, well, I can easily access this intelligence, so let me point my application at it and just, hey, look, it is great, but it costs a lot. Because of the way we built our harness, and the way we have engineered our agents, each agent has a very specific purpose, and it has a very specific context, which means it has a limited context so it does not hallucinate. But equally, its token consumption is very limited, and also we can use an older version of the model or an open source model to do the work. I think this all goes back to sort of, to me, this era looks like you have to get your platform engineering and your infrastructure engineering right to be able to get the true value of LLMs.
If you do not create this harness, I think you end up in a lot of labs, [Rich], all right?
Yeah.
I think, again, for this room, Damon used the word earlier, and why is this defensible? It is defensible because one thing running on the back of all of that, we have said this a lot already in the past, is our dataset.
And we have the largest dataset in the industry. It is 100 trillion data points that have been accumulated over time, and that dataset is proprietary to Robinson. And when you build a bespoke platform like we are doing with that dataset, that makes that very powerful and very hard to replicate, even if you are an AI native company starting off to do that because you do not have access to that data. You have access to some data, which can be averages of averages, but not the level of data that we have. And what Arun just laid out is super important on why this is a deeper, wider moat.
There is no hobby spend on AI at C. H. Robinson, right? So every dollar we spend on tokens, every dollar we spend on engineering capacity has a ready-made high probability ROI assigned to it, right? So we do not just give 9,000 employees or 10,000 employees Copilot license and say, "Go try to do something creative," right? The only dollars we spend are based on a very high probability outcome from an ROI perspective, which is why we have had the success we have had.
Yeah. And I guess as you think about, you talked about the productivity gains, which is one side of it, but then those incremental revenue opportunities, right? Like building a more sophisticated pricing engine. How much would you sort of attribute these AI investments to driving productivity versus driving revenue? And then maybe just give a sense for the audience, what are your biggest areas ahead as we look forward in the next few years that AI will tackle?
I will start and then hand it to Arun. I would say because we let the highest ROI project dictate investment, difficult to say what percentages of our benefits come from revenue management versus productivity versus growth because they are all in the same funnel competing for the same investment dollars. I would tell you, though, the revenue management unlock has been substantial, right? I would say as little as four years ago in this industry, just the way in which the industry priced was very unsophisticated. Very low frequency, meaning you'd set a pricing strategy at the beginning of a month and maybe determine at the end of the month did you win or not from that pricing strategy. Whereas today, with our technology, with our discipline approach from Lean, with that dataset that Dave referenced, we're setting pricing strategies on a seconds and minutes basis.
An example we use a lot is we come in Monday morning, set a pricing strategy at 8:00 A.M. If that's not yielding the volume margin expectations that we anticipated, we could change that strategy tens of times an hour, hundreds of times a day, thousands of times over a quarter. Whereas as little as four years ago, you may have only changed that pricing strategy once or twice in a 30 or 90-day period of time.
Just the frequency in which we're interrogating the market from a revenue management perspective, and just the ability now to access that data set that Dave mentioned is over 100 trillion data points. Before the advent of AI, the ability to analyze that data was extremely limited. Now with our advanced machine learning, with our predictive analytics, with generative and agentic AI, now we can use a significant portion of that data to drive arbitrage opportunities in our marketplace. And we believe, as I mentioned before, what we're doing on revenue management in the logistics industry, we believe is unmatched.
Maybe I'll add. I think the way I would say it is that an agentic AI platform or a harness, you kind of combine that with first principles, and I think you get the same approach we've taken to productivity or the approach we've taken to revenue management and gross margins. The same applies to pretty much everything. So if you convert that into, say, go to market and our sales and account management motions, right? So think about a typical sales situation. There's a bunch of prospects that we have to call out to be top of mind, right? A lot of that's handled by AI, right? Because those prospecting or reactivating customers, especially small and medium customers, we reach out to them purely via AI, right? Because now the humans can focus on something else, right? They can focus on actually serving the customer.
This notion of how do you take customers from the top of the funnel and drive them down lower into the funnel. An example might be, we're not just connecting supply and demand, we're solutioning for customers. Customers are asking us, well, because, the logistics sector isn't like a monolithic single dimensional market, right? There's flatbed, and there's temperature controlled, and there's bulk movement. There's all kinds of different types of freight, I'll call it modes and services. So when customers ask our account manager for that expertise, often they have to call an expert, a subject matter expert, to come in and join the call, right? So now we have AI agents that are trained to be that subject model expert or are trained to be a supply chain engineer because we only have so many subject matter experts and so many supply chain engineers.
Now you apply the same principle of scalability. Now you can take these roles, kind of encode them into an AI agent that participates in the call to help us close deals. Again, it's the same principle that we applied to productivity because this is a different type of productivity, but it's in the sales motion.
I think just to put a bow on all of that to your question, I'm super excited in the next chapter of Robinson. We always say last two years have been awesome. The next two years are going to be very much more exciting than the last two or three. I'm super excited about what the teams are building on our agentic platforms when it comes to, say, our global forwarding business that will ultimately go to our NAST business as well. Very, very complicated business that if you do a quote, it can take upwards of 10- 12 days, in a sense, to put together a really complicated quote to move things from, say, China to North Carolina. Now we're building a platform that could potentially have agents do that quoting in a matter of hours versus days.
That's pretty significant in doing that. Super excited about that technology. That also comes back into our NAST business as well. Where that takt time is immediate, now we're able to take a technology of agentic from generative, and reapply that, bringing some things that we couldn't do below the line, above the line. That's why it's going to be super exciting for Robinson going forward.
Dave, so I think you guys have done a great job of describing what C.H. Robinson is doing that's different from competitors and difficult to replicate.
Yes.
I know we don't have a ton of time, so I want to make sure we hit on kind of broader transport type of questions. One of the concerns that a lot of people have had recently is potential slowdown in the macro. Obviously rising interest rates put some pressure potentially on the consumer, on industrial activity. Speak to what you're seeing out there from kind of a supply-demand standpoint, because a lot of people think that we've kind of experienced this freight cycle inflection and that there's room to run. Do you agree with that? Do you still see that as the case? What does the earnings growth look like? Let's assume freight demand remains somewhat tepid. How important are these tech initiatives in kind of still being able to drive earnings growth regardless of what the macro environment looks like?
Oh, good question, and we like that because it gets down to the receipts in our business, as you know, [Ari]. Damon and I, we always talk about the receipts. First of all, on the macros, you guys see it out there. You are correct that this has been somewhat of a supply side correction, meaning supply has tightened up, prices have gone up. We see that, of course, and we're participating and doing really well when it comes to the spot side of it. But we're also doing very well when it comes to the contract side of the business of what we're doing. From a demand perspective, you called out industrials. We see some industrial technology out there, be like data centers and things like that. But we're cautiously optimistic on that because you really have to continue to watch retail, housing, and manufacturing.
Those are the things that are going to really drive freight, for the most part. I think some of those are a bit muted right now, and some here and there, green shoots, but we are monitoring all that. But the thing with Robinson is higher highs, higher lows, as you know. We are certainly winning at a very four and a half year type of freight recession. We think we will not only linearly but exponentially have a curve that when the market inflects with the thing we've built, it's going to generate even more. But we'll get into some of the actual numbers and why we feel that as well.
Yeah. So just on the earnings potential, just a couple of double clicks there. I would say, in an almost four-year freight recession, right? Certainly in 2024 and 2025, we had over 20% earnings growth in both of those years. Certainly, the market didn't help drive any of those earnings performance. Certainly if consensus holds this year, it will be another 20% earnings growth year, in 2026 with again another flat to down market.
Q2 was, I think, a real important quarter for us. You had the market down again 4.5%. That was with spot cost up over 30%, and yet our AGP per load, which is a key KPI for us, was flat. If you would have asked somebody two years ago, could a broker have flat AGP per load when spot rates were up over 30% in a market that was down 4.5%, they would have told you it is physically impossible. We demonstrated that in Q2. I think the other exciting point in Q2 that just shows the potential for our earnings growth is our operating leverage. In Q2, we had substantial operating leverage. AGP dollar flow to operating income flow over 90%.
In fact, [Rich] at Deutsche Bank reminded us, that was the best operating leverage performance of any company in logistics, including the assets. Right? Here you get a broker demonstrating operating leverage. That is a concept nobody thought was even possible. It is supposed to be a variable cost model. We have transformed Robinson now into a semi-fixed cost. What you get with Robinson is you get the best of both worlds. You get the operating leverage of an asset in an asset-light model. Right? We believe we have created something quite unique at C. H. Robinson, which is why we believe we will continue to outperform the market, both from an outgrowth perspective and both from an earnings perspective as we go into the future.
That is important, [Ari], and you see it, and that is in a really tough backdrop. As this inflects, this system only goes wider and deeper.
We are excited to see where it goes, certainly. I know we are close to time here, but last question because it is probably the question I get most often. Obviously, the stock has sold off a bit on concerns around broker liability and Supreme Court ruling that opened up brokers to liability in the case of accidents. Speak to that for the investors in the room or the investors listening in, who might say, "I cannot get comfortable with the C. H. Robinson story until I know how this settles out."
Potentially we are looking at years of brokers fighting in courts or fighting these claims in courts. How do you think about that? What would you say to investors to get more comfortable around that? If we could tie it in quickly to the tech point, talk about how you can leverage AI maybe in carrier screening or what is being done there as well.
Yeah. So three, four vectors you are really calling out, and we will try to do it very quickly for you here. From a Montgomery perspective, these are just facts. We are a data company. We talk in data and facts. What the Supreme Court ruled on that was just no more of a preemption for brokers. But the facts are, prior to Montgomery, that is just one defense now that is off, but we had to deal with well over 30 states that did not have that anyway.
For Robinson, we have always had a docket. We have been public for 28 years. We have had a docket during that time. Everyone in this industry has a docket that they are dealing with. We have had tens of cases in that docket. We ship 37 million shipments a year, and over that time, that is hundreds of millions of shipments with tens of cases.
That just tells you that, one, we are disciplined, we are measured, we know how to defend a docket, but more importantly, we run a very, very safe and disciplined company in doing that. The lower side of tens of cases, hundreds of millions of shipments. You break down that. So that is the facts of that. The second vector then for investors is, "Okay, Dave, what about the impact of inflationary insurance because of this? Will you get that?" Let us talk about the facts of what that is, and maybe we will finish off with the technology on what we are doing on vetting, which we think is the best in the industry.
Just to round out Lipe, which is the last case. We feel really good about the facts of that case. We chose not to settle that case. We feel like we will prevail on appeal, so that is the facts of Lipe. 98% of all of our cases either get dismissed or settled. We do not think that trend is going to be disrupted post Montgomery, post Lipe. So we still believe the vast majority of our cases will be settled. Average settlement amount has been somewhere between $1 million and $3 million. We think that trend probably holds into the future as well.
As it relates to insurance cost, I honestly believe you do not have to wait until Lipe gets through the appeal process to get comfort in C. H. Robinson. We are going through insurance renewal right now. I believe once the insurance companies essentially provide their verdict on C. H. Robinson for 2027, I think that will give you great insight to what they view as the risk profile of our docket and what they view as the risk profile of Robinson. Some of the more bearish sentiment on the street that insurance costs are going to go up hundreds of percent, we do not view that is going to be the case for C. H. Robinson.
We think inflation will be a very manageable number, that the majority of it will get passed through freight rates anyway. Ultimately, the consumer will bear the majority of that cost. Any legacy costs borne by Robinson, we get paid to offset that anyway. Just a baseline fact to show you how immaterial insurance has been to us historically. Insurance plus claims is less than 50 basis points of gross revenue for Robinson.
Automobile liability insurance on its own is less than 25 basis points of gross revenue. So even if we did see a material increase in inflation on insurance, it is not going to have a material increase on our earnings. Look, we feel really good about where we are at. We ultimately believe the current legal landscape will drive a pretty accelerated consolidation of our industry. Which will be, once we get through the fog of war on Montgomery and Lipe, we feel like this will be a very strong bull case for C. H. Robinson on the other end. So we have been active buyers of our stock. We continue to be active buyers of our stock. We are putting our capital where our words are.
Finally, we are driving a legislative and rules-making vector as well, as we are working with FMCSA to get a standard through the Department of Transportation, and have a lot of our transportation industry peers that are following with us on Responsible Freight. I will be in D.C. next week. Also working a legislative solution to this as well, with certain bills that are going through that we think that will apply the right accountability, responsibility. We are going to continue to lead the industry on that. Thanks for having us, and hopefully your investors understand our story. More exciting to come in the next few years.
Yeah. Thank you.
Absolutely. It's a great story, and you guys tell it well. Dave, Damon, Arun, thank you all.
Thank you.
Thank you. Appreciate it.