Auto-transcribed via Deepgram nova-3. Speaker labels are approximate; light cleanup applied.
[00:00]
Jesse Buckingham: These tools have gone so good off the shelf where you could literally prototype in a day or two and get a functional demo that works on 1% of the use cases. Yeah. And when you try to go to production on the next 4%, that is when you realize, like, wow, this is a process. Every time I hear the build versus buy, especially in the AI landscape today, I think that you haven't really probably gone to production any sort of scale. I'm excited to share the next episode of the Freight Show with you.
Today, we've got a conversation with Jonathan Drouin, who is the product strategy lead at Worldwide Express Group. I really enjoyed this conversation with Jonathan. The main theme that I think is really interesting here is that I think Jonathan and Worldwide Express Group have done a really awesome job of being very excited and optimistic and encouraged by what is possible with AI in their operations in logistics, but also sort of pairing this with a really high level of discipline in how they assess opportunities, how they decide what to continue with, how they decide when to cut bait on initiatives. For anyone in the industry thinking about their AI strategy and how to make sure they're both keeping up with the developments but also making sure it's translating into business results — I think this is a masterclass in how to think about and navigate that. I hope you enjoyed this conversation as much as I did.
Welcome everyone to another episode of the Freight Show. Jonathan, great to have you on.
Jonathan Drouin: Thanks for having me.
[02:04]
Jesse Buckingham: I'm excited to chat because we've obviously known each other for a little while but I haven't ever really had the opportunity to fully hear your story. And I also really enjoy our conversations about AI in brokerage. Maybe to kick it off, you could tell me — you've had a very interesting career in logistics, as an entrepreneur in the space, on the operator side, on the product side. How did you sort of get into this world? Take us through the big brush strokes of your journey in logistics.
Jonathan Drouin: Yeah. Definitely. Thanks for having me. Look forward to the conversation. I had what I would say a nonlinear career path to getting to where I am today. I joke around for a while — I am French Canadian. English is my second language. When I mess up a couple words, that's when you know.
Jesse Buckingham: You'd really not know. I could not have guessed that English was your second language.
Jonathan Drouin: A lot of work. First couple of years, moved here right at the end of high school as a junior, and that was a rough season coming from Quebec, Canada, French speaking to good old Texas. A little bit of a difference. But origin story — I didn't go to college right away after high school. Why was that? I did not know what I wanted to do. I had a lot of ideas back then. Used to be, when I moved to The US, I quit playing hockey, picked up golf, got really good within a couple years. And I guess when you're 18 kinda daydreaming, you're like, hey, maybe I'm gonna make it pro. So I didn't go to college right away. I tried to just play a lot of golf, learned really quickly that was not gonna work out.
[04:07]
And after a couple bad tournaments, my dad showed up and he said, you know what? You're gonna work. My dad had a software company. So I was, you know, 18 going on 19. He said, here's a bunch of IT books. You're gonna learn the books and you're gonna show up at one of my customers on Monday and you're gonna be a software developer. So by 19 as a software developer, I did that for a few years. And really by the time my friends were graduating college, I had tried to do some startups at that time. And it's kind of funny — in 2011, 2012, I wanted to do something different. I've been doing clinical software development for four or five years. And I went to basically a job fair. And ironically, I walked up to a company called Bear Transportation. Somebody you know was there — Michael Kaney at that time was at that Bear Transportation booth. So we started talking. Lo and behold, a few months later, I was working at Bear Transportation. So I went from software development to freight brokerage in 2012. So I got my start there at Bear. Michael Kaney kind of hired me, trained me up in the initial years. I started on the freight brokerage side there. I did a very short stint at CH Robinson. Technically, they were acquiring Phoenix International at that time. So I happened to be in that transition, about a year. And I went over to PepsiCo — they were starting their own freight brokerage at that time. So I went to the very early years of PepsiCo and that freight brokerage. And at that point when I had been in the industry four or five years, I had a tech background, I knew the freight brokerage business. And I remember moving my first shipment in 2012 thinking, my gosh, we're gonna digitize this business within four or five years. There's no way brokers should exist. Back then, that was my initial thinking. So 2016, went down on my own, kinda started my own business, raised some venture capital. Started my own TMS company in 2016. Actually started my own freight brokerage in 2017. Small exits a little bit later on — great experience, great learning opportunities.
[06:08]
Then I went back into freight broker sales. And in 2019, I had the opportunity — I had been reached out by what I work for now, Wex Group — and they had said, hey, we like the technology you'd built a few years ago, we'd love for you to come on board, build something similar for us. I started on the truckload brokerage side, which was probably $200M of the $2B at that time. Fast forward to today, we're a $5B company. We did the merger with GlobalTranz in 2021. So definitely been a unique career path from starting in 2019 to 2026 now.
Jesse Buckingham: Going back to PepsiCo, why did they start a freight brokerage? Because it's quite a big brokerage.
Jonathan Drouin: And, again, I wasn't there for the time they had made the decision. But I think the folklore or the stories we would hear is, hey. We had executives at the PepsiCo New York headquarters discussing transformative ideas and, hey. We're spending billions on freight. We're also using co-packers for bottling some of our products. And they were looking at what are some of the things that we could take in house to reduce costs or maybe turn into a profit center. In that time, transportation was one of those. Think between PepsiCo and Frito Lay, you've got some of the largest private fleets in North America.
[08:09]
They've got plenty of backhauls, plenty of opportunities. And I think as they could see how much they were spending with CH Robinson and Coyote at the time, they thought, what if we could bring a percentage of that freight back in house and monetize that? So that was the early foundation. When I joined, it was very small, maybe $7 to $10M in revenue. By the time I left, maybe closer to $80M internally. And they've blown significantly past that today.
Jesse Buckingham: Super interesting. Are there other shippers that have gone this route?
Jonathan Drouin: Yeah. And I feel like it — a lot of people think because PepsiCo is one of the biggest that have done it, and it seemed more novel. But there's been other companies. Even Coca-Cola has their setup. There have been numerous shippers over the years that had done this internally. And I think it's always interesting to see the phases because there seems to be the same kind of track record: hey, your first $0 to $10M brings in a lot of executive excitement. And they'll typically start with their own freight. A lot of it will be breakeven, but you can see, hey, now that we're building these capabilities in house, should we sell to other shippers? So it's interesting to see all of them start with internal freight. They'll go get some external market freight, and they'll grow pretty well. And a lot of the times what we've seen is these companies are running the model at a very low cost to serve. Sometimes they're taking people from in-house with some brokers from the external, the real estate footprint is near zero. Some of the cost structure has advantages in the beginning.
[10:29]
You could see a lot of the ramp up, but it seems as you get to a certain plateau, especially in the last two to three years, you've seen a lot of them get into that trouble. All the market does fall off. They're at the end of the day the same broker. They have the same disadvantages as the broker. And you've gotta think for a shipper that's in the business of making a product, there's very large cash flow impact when they're selling freight and moving freight and dealing with the transportation AP side of the house. And I think over time, they seem to stall out a good amount and kind of just plateau.
Jesse Buckingham: Yeah. That makes sense. There's probably a lot of excitement about it and you start these things in market conditions where you're like, hey, why are we paying all of this? We can do it. And then you realize quickly that that becomes not that exciting in a world where there's a lot of cash being put up.
Jonathan Drouin: We'll see it eventually.
Jesse Buckingham: Is there any broader trend that you're seeing in the shipper market about trying to bring more stuff in house or outsourced?
Jonathan Drouin: I think it really depends which shippers you're talking about. I would say that if we're talking about the larger enterprise mega shippers, I think the trend in the last few years is they've become very smart from a data perspective. A few years ago, they were really good at having a really good RFP process and stuck with their standards in RFP. But then fast forward to now, if you're at the top echelon in the mega enterprise space,
[12:30]
you've got plenty of tools, plenty of data points. I think definitely at that echelon, you're having a very different conversation than four or five years ago. There's data. There's tender load to truck ratio. There's just a lot more information. I would say that has been definitely a trend on the bigger enterprise sector. But technology overall continues to work its way down to the mid market and SMB market a lot more than before.
Jesse Buckingham: Yeah. And so they just have more data about the market so whereas before there was opacity and they would discover price through an RFP, now there's better information about where the market's at. When you were building your own business, you did TMS and brokerage. Were those in that order?
Jonathan Drouin: Yeah. You know, it's interesting. It was twofold. I think one, the really honest direct answer is when you build a TMS company, if you think about a TMS for a freight brokerage — you are the heart, you're an organ. And if you're a new company trying to get brokerage customers, a brokerage that's making a decision to buy a TMS, yeah, they want the better tech, more efficiency, and all that, but there is a stability component. They've got to know that you're gonna be there for the next decade. You might kind of date your visibility vendor, but you're marrying your TMS company.
[14:38]
So in the very early days when we were with prospects, there was a lot of excitement seeing the technology, and it's like, wow. This is way ahead of your time. But there's always a feedback of, I just wanna see what's your staying power. I'm gonna give you time. I don't wanna be the first one. I don't wanna be early on this. This is too drastic of a decision. And that was the first aspect. And you know, as we were building this technology, we had some early adopters and we could see how efficient it was. And we'd walk into these brokerages and I could see the systems they were working on. And of course, in my mind, I thought all they had to do is move to ours and you'll see a significant uplift right away. And it was so hard to get that conversation. I remember talking to my private equity at the time. I was like, why don't we just start our own brokerage? At that point, right before we had launched, Echo had acquired Command. I was explaining to them, look, if you just have better technology, get efficiencies in a freight brokerage, you'll be able to scale. So we essentially started our own freight brokerage. We did use our own software to really prove, hey, this works, this is scalable. We joke around that our ad back then was we'd be driving down the tollway, the main highway in Dallas, and would post a load to the load board and our system would just — the workflow was so efficient. You'd be in a car looking to load a laptop, which ten years ago was far ahead of its time. So it was the workflow designs and just being able to prove that to the market.
[16:41]
Jesse Buckingham: And I'm curious because when you look at big brokerages and even across, there is this sort of build vs. buy decision around TMS. And I'm curious how you're thinking about that today. And like, what do you see as the scenarios where it makes sense to build versus buy, specifically with TMS? Because I'm also curious to ask the question as it relates to AI.
Jonathan Drouin: Yeah. And it's interesting. I feel like I've evolved over the last almost ten years from starting my own TMS company to now when I'm making a TMS decision, I'm considering a few billion dollars of freight. So the decisions, the impact, the risk is significantly different. I think if you're on the smaller side, you get to $100, $150M, you've got enough revenue, enough EBITDA to consider having that. But again, I would say there's still a very big aspect of our industry and growth that is purely around the hustle, the sales culture, just the energy. So when I talk to a lot of people that reach out like, hey, we're at this stage, $200, $300M, we're considering building a TMS — I want to make sure they've got the people side of the house really well working because I know that if they go build their own thing, it's not gonna take you from $250 to $500M overnight. It might gain some efficiencies, but today when I look at the build vs. buy, the stability is more important. Making sure the people and process side is perfectly aligned before you go right into the technology.
Jesse Buckingham: Yeah. That's interesting. And then the Worldwide Express Group is a really interesting business. I'm curious about the strategy because the breadth of the service offerings is very wide. I'm curious about the thinking behind the GlobalTranz acquisition and how the business has thought about those over the years.
[18:42]
Jonathan Drouin: Yeah. So it's grown quite a bit and it's changed over the years, but first and foremost, think the company is really focused on — we're a UPS reseller, so we do a lot of parcels. So today, we're about a $5B company. Our parent company is Wex Group. We have three go-to-market brands in the industry: Worldwide Express, GlobalTranz, and Unishippers. GlobalTranz is our agent network. Unishippers is our franchisee network. And then our Worldwide Express or W2 direct go-to-market brand. As far as mode, we're about 40% LTL, and then the rest is split between parcel and truckload. We are unique in the sense that we are a reseller of UPS, and that is a core competency and core focus of what we do. A few years back, if we go to 2016, 2017, there were really two large resellers of UPS: UniShippers and then Worldwide Express. And then we merged the two companies in that timeframe to become the largest UPS reseller. So typically our relationship with UPS is their team will focus on the enterprise, the larger accounts, and as you go through the mid market and really the SMB where we specialize, we handle the UPS relationship from that angle.
Jesse Buckingham: Does that sort of inform the strategy where you guys have a unique advantage in serving some of these SMBs who have a lot of parcel, and do you then target companies that also have LTL and full truckload needs?
[20:43]
Jonathan Drouin: I would say that overall, it's somewhat of a mix. The LTL and parcel, typically a rep will go into an account, and as they're doing discovery, we'll typically bring in a truckload counterpart as it makes sense or vice versa. So we're typically split off LTL/parcel sales, truckload sales, and then they'll bring in the opposing parties as it makes sense. But we'll typically operate both because there is some overlap.
Jesse Buckingham: When I've spoken to folks that sell multiple modes, if you could choose, you would sort of say, hey, it's a single relationship and you just work with me and I've got you covered. Practically speaking, why is it hard to achieve that? Why is it so different to sort of sell one mode versus the other?
Jonathan Drouin: The one sales rep that can sell all modes — I had listened to even your interview with Doug a few months back, and he was talking about how they're selling one sales rep tries to sell all. But I think what we've seen historically is if you train one sales rep to go after multiple modes, they will naturally go to what they prefer. So I think by just breaking it down to, hey, your sales and your comp plan is focused on these one or two modes, you're focused on the other, I think it drives more of the right sales behavior. In a perfect world, yes, you'd have probably both. But I think the ramp up from a sales perspective — they'll stick to a couple of items. So I think our model just doubles down on that approach.
Jesse Buckingham: So take us through your sort of current world and orbit and what you're focused on. And then I'm curious to start diving deeper into how you guys are thinking about what you're prioritizing today.
[22:51]
Jonathan Drouin: Yeah. No. I'll give you where I started and where I'm at now because it's changed quite a bit. So I started in 2019 originally to come build out the truckload portion of our platform. And 2020 happened, kinda slowed things down. So we really focused on parcel and TL as the core of the company. Then in '21, we did the deal with GlobalTranz. So it went from, hey. You're here to build. Now you're here to integrate multiple companies. At that time, GlobalTranz had gone on a buying spree and had bought multiple different entities. So when we got together, the mandate was how do you consolidate all the truckload business units into one platform? So it was a few years to go through that process. In 2024, shifted from truckload migrations to a heavy focus on revenue operations. And then I switched over to the technology team later in 2024 working for our CTO. And then I went from truckload product to product strategy, more of an overall oversight on the product strategy. And by the end of 2024, going to 2025, very, very focused on the AI side of the house. The models were changing. The vendors were coming on the scene. So by the end of 2024 going into 2025, we were starting a lot of our initial projects within the company. And then as we shifted to 2026, it kind of leaned towards product strategy, heavy focus on our agent channel at the moment.
Jesse Buckingham: How have you thought about AI strategy? I'm curious — what are the use cases you go after? How do you think about build vs. buy? If you're buying, how do you think about vendor selection?
[24:56]
Jonathan Drouin: Yeah. Love that question. So because I was an operator, I would say probably the quick answer was I did a lot of it by gut feeling. I knew the business very well. So I could already tell pretty quickly, here are the pain points. Here's where there's a lot of work. There's a lot of exceptions or very repetitive tasks in these buckets. But in the beginning, we spent a lot of time remapping all the workflows. We went through team by team. We remapped the whole organization. And at that point, we had a good understanding of the technology at that time. When you can look at the workflow, how the business was set up, and some of the tasks each team was doing — we looked at okay, what's the available technology that is working today that makes sense? So we did the full order to cash, or really from quoting to cash. We also partnered with Vooma on the quoting side, shipment creation. We went role by role, team by team, basically workflow by workflow, and looked at what was possible. In the first year, we spun up close to a dozen AI initiatives. Now some worked really well, others were great learning experiences. But we went pretty wide pretty quickly to really stress test — okay. What's the boundary? Because I think when working with emerging technology, there's just a lot to learn and nobody's got it figured out. And I think it's better to just — action beats inaction. This is one of those cases where we went pretty wide, pretty fast, and got a lot of learning under our belt in 2025. So as we're taking out projects this year, it's a lot more strategic, very precise on that approach going forward.
Jesse Buckingham: That initial piece of work that you did around use case mapping — how did you sort of assess and quantify opportunity?
[27:30]
Jonathan Drouin: So that was actually challenging. Yeah, we worked with the business stakeholders that knew these key areas really well. But I think what was challenging sometimes is, what's the art of the possible? These team members know the process better than anyone. They've been doing it for five, ten years. And it's a lot of times it's, okay. Here's the current workflow, but we're gonna try to reimagine it with the possible tools that exist.
[29:01]
And when it came down to how do we quantify — some areas were very straightforward. I've got 5,000 clicks a week in this area on these one or two tasks. Some of it is a very big time efficiency and how do you measure that? It's like, yeah, sure. I think it's gonna save fifteen minutes. We're doing it a thousand times a week times x amount of folks. So it varied quite a bit. But I know we've mentioned this before — we targeted really three things. We were very binary in that sense. What will help a customer, what will improve customer retention? That was a key factor we're looking at. What is gonna grow margin? And then what's gonna reduce our cost to serve? While we had a lot of assumptions initially, when we did a full quarter with our business leaders, we had about 60 or 70 ideas that came out. Then we picked 10 or 12 to get started on. But the ones we did pick had to hit one of those three initiatives. And if they were too vague, we just kicked them to the side. Because at the end of the day, when you're PE-backed, you've gotta show results. We had to make sure we delivered value very quickly.
Jesse Buckingham: I sort of love it. It's at some level super obvious, but I don't think everyone just articulates the structure quite that simply and clearly. Does it retain customers? Does it improve net margin? And does it reduce cost to serve? And did you for things like measuring time — how do you sort of even get to those metrics?
[31:36]
Jonathan Drouin: I would say overall, a lot of them are 80/20. I think when we've gone deep into the data, what we've seen historically, for example, if we take data around our CRM for sales and our operation roles or revenue operations roles — you could see a lot in the data. You could see they touched the case at this point, they did this action on this case. So 80% of the times, it's three minutes from this stage to this gate. But then there's the exception layer. So we relied a lot on what I've said is directional time study. And I think a lot of people get lost at some point when they go too deep in those areas. I would say we stuck to things that looked pretty obvious, that we wouldn't have to have three analysts go looking at too much data to determine if there's gonna be an ROI.
Jesse Buckingham: Yeah. It just stood out. And then how did you go from — 60 initiatives, working with business leaders to sort of uncover those — what did that exercise look like?
Jonathan Drouin: Yeah. I would say it was, you know, product 101. We always have kind of a list of the backlogs, the issues, what do we work on to improve the business and the tech and the automation. But we did a very large reset at the end of 2024 because quite frankly, the technology had changed so much. We just wanted to refresh all the assumptions. So we went with our core groups — the sales side, operation side, finance side, accounting, all departments — and just like product discovery 101 of the art of the possible, what are the things you wanna solve? And we were very intentional about starting over from scratch, just retesting every assumption. Because the technology we've got now, there are things that are doing today you couldn't do four years ago. We wanted to take the fresh lens. And it was very productive. We actually communicated very closely with the business leader of each department of, like, here are the specific things to look for and bring up. So we did AI education, a lot of AI automation education before going to those sessions so that once we were in there, people weren't just throwing a couple ideas that looked very basic or simple. They were opened up through a lot of ideas.
Jesse Buckingham: Yeah. It's interesting that people's ability to generate ideas really is like a function of how well they understand
[33:41]
what is possible. Because I've even seen big step changes where our teams experiment and play around with certain tools, then their ideas become much better about the solutions that they come up with.
Jonathan Drouin: Yeah. No. I think that's critical. I think in this world, especially the rate of change, I think the rate of change in the product world, AI world today requires people to be quote unquote tinkering on the latest tool to see, okay, how do you connect the dots to what you're doing? Because it's just moving too fast.
Jesse Buckingham: Then okay. So you guys have sort of mapped out the use cases and there's been conversations about build vs. buy. What have you learned about your framework for thinking about when to partner, when to try to tackle things yourselves?
Jonathan Drouin: Yeah. It's evolved quite a bit. So when we started and kind of finished '24 going to 2025, we did multiple partnerships, and we did take on a lot of internal work. We've got a pretty big dev shop. We took on some of our own work. We kind of took some of the top developers we had that were very interested in the topic, and they're playing with the large language models nights and weekends. So we got the missionaries — they were very focused on it, kind of started a quick AI pod within the company. So we started, again, Vooma on the quoting, the quoting shipment creation, different ones for the voice AI and different stages of the process. We had pretty much every touch point from quoting to cash with some sort of AI initiative kicked off. As far as the builds, what was really interesting is if you asked me in January 2025, I really thought the internal build would be a little bit easier.
[35:43]
There's definitely a learning curve when you're building anything AI, but we'd taken on some projects that were small at the same time still pretty impactful. And I think what we learned pretty quickly is when you're working with new technology — and we built some of our own multi-agentic workflows and orchestration across a dozen or so agents with their specific tasks — that became challenging very quickly. And when we look at the landscape, the reality is a lot of these companies that are entering the scene are backed by great VCs, hyper funded, they've got the best talent around, and when they're working on one of your initiatives with you, they're putting five, six, seven engineers from MIT or Carnegie Mellon. And we just saw a drastic shift of us building versus them building. So we really increased and kind of leaned into partnerships the rest of the year. I think right now it's a great time for companies to actually partner versus building. The technology is changing too fast. VCs have really flooded the market, and it's good for brokers. It's keeping their prices with vendors in a really healthy spot, so it allows us to do a lot more experimentation if we partner versus if we try to build in house.
Jesse Buckingham: Yeah. It's interesting. This conversation comes up quite a lot with folks about building versus buying. And I always sort of — I'm of the mind like, you should figure this out for yourself. Because it's obviously self-serving for me to say, hey, you should partner. But as somebody who spends my life building these systems, and when you're dealing with the thousand details to get these things to work reliably at production scale, it always — whenever I hear, hey, we'll just go build this myself, the thing that goes through my head is like, well, have you done this yet? Like, do you have proof points
[37:44]
for getting to production rollouts on this? Because I think you may be discounting the cost. And it always is interesting to me the sort of total cost of ownership because you're right, we are sort of amortizing our learnings and investments, knowing that sometimes we will front load things because we have the tools and capabilities over time. I think if I was in your shoes, I would also be looking to partner on a lot of these things. I think part of maybe where it doesn't apply quite as much is if you have a very organizationally specific workflow, where you might get somebody to go build it for you and it'll almost feel like a discounted dev shop.
Jonathan Drouin: Maybe that's kind of how I'd think of it. And I think I would double down on one of things you said. And I think this is probably the biggest issue right now that I'm seeing — I love vibe coding. I love the whole concept of building whatever you want. And I think right now because of last few years, these tools have gone so good off the shelf where you could literally prototype in a day or two and get a functional demo that works on 1% of the use cases. And when you try to go to production on the next 4%, that is when you realize, like, wow, this is a process. And so every time I hear the build versus buy, especially in the AI landscape today, I think that you haven't really probably gone to production at any sort of scale because it is — the prototype has gone almost zero time, but the zero to five is quick, but the five to get to 50% production AI, it is a journey.
Jesse Buckingham: Yeah. I know. I viscerally feel you. One of the things that I've appreciated about you and Worldwide Express Group has been that you are both very AI forward,
[39:50]
but also not super AI pilled in a beer goggles kind of way. There's still quite a lot of discipline about how you've thought about it, and it sounds like that's been a little bit of a journey — an exploratory phase and now sort of sharpened your pencils. I'm curious — what's the perspective of your private equity owners when you're sort of dealing with this very exciting technology shift?
Jonathan Drouin: Yeah. So we've been around a minute. Twenty-five, thirty years. So I think one, there's a longevity aspect. We gotta be — we're PE-backed, so there is a return we have to provide to the market. So there is a lot of built-in discipline. But I think just from a culture and philosophy perspective, I would say that if you looked at the company overall, we are a sales organization first and foremost. So a lot of the focus is there. So when we do take on tech and product, it had to just deliver the result and mentioned those three pillars of, you know, customer retention, growing margin, cutting costs. I think it is embedded in our DNA and we just don't deviate off of those. So maybe sometimes it makes us a little bit slower on the front end, but when I look at the long run discipline, I talked to a lot of our peers in our space and, you know, I see them using sometimes even similar technologies deploying the same. But I think our approach is — there's so much discipline that we're getting to the outcomes much better and much faster. And actually I think our change management is significantly easier here than a lot of the peers I communicate with in the space.
Jesse Buckingham: What yeah. Why is that? And when you say discipline, what do you mean?
Jonathan Drouin: We're not gonna be the ones to splash a press release on anything we do. We're not gonna go on the cutting, bleeding edge. It's like, you know, there's Myspace that's the cutting, bleeding edge, but then there's Facebook that got all the market share.
[41:53]
So we're really focusing on letting others go first, watching what they do, learning from them, and then once the technology is showing promise, it is working — then we'll go aggressively and make it happen.
Jesse Buckingham: And then so it's like making sure that there's proof points, that there's ROI there before you embark on it. And then what about the realization of that ROI? Like, how does this sort of discipline continue at that point?
Jonathan Drouin: So we're even to that point very tactical. Any project we take on, for example today, we have an agent steer co every two weeks. So every project that we take on — we've got an agent steer co, and it is reviewed by our top leadership, all the business stakeholders. Everything we take on, we take on a project as a tech product org. We set the KPIs up front, and of course, at times they will change, but they are known and public from day one. We execute, and then we've got the follow through, in this case an AI steer co where everybody sees the results. So if you're not hitting the results, very quickly, hey, we're gonna abandon. Like, this is not working. Or, hey, business leader, we've delivered this. You said we would get XYZ growth. We're not seeing it. Why? So I think there's a lot of visibility and accountability.
Jesse Buckingham: Yeah. So it's this sort of like setting — what is the business case? How are you gonna measure success? And then having an operational cadence to see that, and then knowing when you're gonna like cut or continue, and being intellectually honest about those.
Jonathan Drouin: Very, very — that intellectually honest being the key. You've got to be. And I think in this space too, there's just so many things happening so quickly. I'll talk to a lot of peers that say, hey. We deployed this AI and, yeah, we like it. It works well. And it's like, okay. What was the outcome you got? And it's like, oh, quality of life got better. I'm like, okay. Is that it?
[43:53]
Jesse Buckingham: Those three pillars, you've done quite a lot of projects and experimentation. In the sort of short term or immediate term — customer experience retention, net margin, and cost per load — where have you seen the projects gravitate towards, or the results be the most obvious?
Jonathan Drouin: Yeah. I mean, I would say anything that's repetitive has been the areas where we've seen the most amount of impacts. I think from a tech perspective, it's definitely around the medium. I think right now, we've got email AI, we got phone AI. I think things like what you do on the email AI — I think that has been one of the best types of product we've deployed. And I think part of that is if you look on the quoting side for what you do, the customer doesn't know there's AI involved in the process. So there's no customer moment of, this is all AI. It's, hey, they get a quote back faster. So there's an improvement in process without them knowing AI is involved. And in different areas — for example, on the carrier sales side, when our carrier is reaching out by shipment and they're interacting with the email AI on the carrier side, we've seen a lot of very positive engagement. So I think there are mediums like email that are doing really well. But again, it goes back to repetitive type tasks, linear type tasks work really well. And I think over time as the technology gets better, they'll probably go a bit more upstream.
Jesse Buckingham: So channel is one piece. Yeah. Of course. And the repetitiveness. And I know that one of your perspectives is that you're actually maybe more bullish than others about the scope that you think AI can get to over time. Because it sounds like today a lot of that is the repetitive work. But in the fullness of time, what do you think is gonna be possible in logistics with AI?
Jonathan Drouin: I believe the boundaries are gonna get pushed. I think today, hear about we're gonna do the repetitive tasks quite a bit. So I do think there's gonna be things like managing a large customer or managing a very large RFP or very complicated situations that will stay as is.
[46:06]
But I do think there's gonna be a lot of AI enhancing everybody's role, and it's taking a part of it naturally. I think it's gonna be — I don't know the percentages, but maybe it's only 10% of the repetitive work. But I think AI is gonna enhance a lot larger percentage. And I will say this — you tinker with the latest tools that are probably not corporate approved or enterprise approved, real frontier cutting edge right now. These tools are doing a lot more. They're able to interact with an email and a phone, and I think they just get smarter as they have more context. And I think — I give the example of — if you deploy AI or language models in production, a lot of the times they get limited. They'll do the tactical repetitive tasks really well. But when there's exception management, they kind of start breaking down. And we've got AI in our back office doing a lot of exception handling. And when you look at some of the problems, they're really basic sometimes, but they still need a person. It'll be a carrier invoice or we send an invoice and it didn't have the right PO starting number, somebody forgot to put it in the TMS or the CRM, and now it's a bottleneck on the finance side. Well, if you think about the future of AI and you've got AI in your emails, AI in your phones, and the large language models and their memory have that context — I think they're just gonna be able to chip away at more and more exceptions that look impossible today.
Jesse Buckingham: In a lot of ways, it feels to me like the intelligence is very good already and that the challenge actually is in the context engineering — how do you make sure that the agent has the right context now. Of course, too much context and they're still not amazing.
Jonathan Drouin: But if you can get the context right, then the intelligence is almost good enough for a lot of this stuff. And to your point, the context is really simple. If you just get some documentation right into one area, suddenly the intelligence is three times the results on the back end. And I think that's one of the skills of AI to master in the coming year or two.
[48:16]
Jesse Buckingham: When you sort of look across the industry today, what are some of the pitfalls that you see people make?
Jonathan Drouin: I would say it's definitely simple change management. I think change management, but not having a clear goal. I think a lot of the peers I talked to, they'll set a goal. Hey. We're gonna onboard xyz for AI in a workflow. And the mentality is too often it is: we'll see where it goes. And then you deploy the technology. You're now paying the SaaS cost or the token cost for the technology, and you've helped the business a little bit, but you haven't really delivered a true measurable outcome. And I think the biggest mistake is really not setting up the outcome on the front end. And again, the outcome can be incorrect as you go through the process. You'll learn something. You'll have to change the outcome. But I think it is starting without being clear. And once you get in, you're ninety days in, you're thinking I don't get adoption, it's not working. I think a lot of the times it is it is cultural, people side versus technology side as it stands. Because I can say that there's multiple projects we've deployed that some of our peers haven't. In theory, they should have the same results, but we see that happening, I think even more in the AI language model world than probably previous types of technology.
Jesse Buckingham: Yeah. It's interesting. I do see this a lot, right, where this sort of initiative is like, hey, let's go learn and figure this out. And so then you sort of back into the objective and then it's assess whether or not you think it's almost good enough. But I like that sort of push to articulate it at the front end. On the change management side, this is actually still one of the biggest barriers to ROI realization. Right? It's getting people — a lot of this stuff is the AI sort of operating autonomously and maybe the change management is less, but usually there still is. I'm curious what you've learned about that. Like, when do you push as a business because you're realizing that the resistance is cultural, but the destination is clear versus, hey. This is actually not the right product?
[50:17]
Jonathan Drouin: I mean, I think our answer is is is somewhat simple as we're just committed to it. And I think for us, it's at this point good year and a half into the journey — everybody knows we're committed. It's more of like, hey. We're gonna do it. So I think at this point, we're past the stage. And I do think a lot of people kind of tiptoe on that, and we made a decision we were gonna go head first into this and never really looked back, never stopped. So I think it was known from the beginning. For yeah, of course, we want people to buy in, but if you're committed, we have to keep moving. The technology is moving too fast. So we made that decision up front. And again, I do think our culture is unique in that sense of we've gone through multiple transitions and organizations through M&A. We do a lot of M&A. So we've been at this point multiple times that we'll rally the teams and, hey. Here's the decision. Here's the path. It's not gonna be perfect, but we're gonna go down that path. Again, do think culturally, we're very unique in having better change management than probably others.
Jesse Buckingham: I'm curious how you guys have created this because I feel the same way in our business. There is so much that is affecting how we write code today, how we think about operations, how our sellers sell — where there is huge opportunity for AI. And I have taken that stance in leading up. Now, we don't meet a lot of resistance. We're an AI company. We hire people out of Silicon Valley. Everyone is so excited about the potential. But it's also like not optional is how I think about it. Because there is — and I don't even mean this in logistics, I just think for every company, we're at this inflection point
[52:27]
where it's so critical to make sure that your culture is aligned. And I do see examples of businesses where it feels like business leaders are getting almost held hostage by employee resistance in a way where I'm like, this is going to potentially be existential. Maybe not in the next six months, but in the next three years.
Jonathan Drouin: I mean, I frankly agree to a level. And I think if you have the team members that are resisting to the example you said, I think they're about to be the next Kodak moment. And I think you do not wanna be that company. You can't afford to be that company anymore.
Jesse Buckingham: Yeah. Yeah. I think it's a wild time where there's so much opportunity and I also think there's risk by not acting. And it's tough because we're in the fog of war so to speak — not everything is clear, you've got to kind of feel your way to truth, because there's a lot of hype out there as well. So how do you simultaneously be very optimistic for the future, but also build the muscle for cutting through the noise to signal?
Jonathan Drouin: Yeah. And I think it's interesting. We're recording this as Klarna just made the big press release of 40% staff layoff. I think there's been challenges of a lot of folks on the Silicon Valley or AI front releasing earnings discussing, hey. We did this. We cut staff by x because of AI. And I think sometimes it is AI, but sometimes it's underperformance, and it's messaging for better results in the public markets. But I think some of that fear escalates so quickly with this technology that I think it's kind of strange. It's moving faster, so you need to adopt it faster than anything before. But there are pockets of resistance because of the type of messaging that came early on around these tools and products.
[54:32]
Jesse Buckingham: Yeah. Yeah. AI is definitely part of the picture, but it's almost certainly not all of the picture. And there are some interesting dynamics at play with public businesses of, you know, easy way to get some multiple uplift by making sure that you're perceived in a particular way.
Jonathan Drouin: Very interesting.
Jesse Buckingham: Jonathan, I've really enjoyed the conversation. What kinda gets you excited in this space and for your role and what you guys are working on? Yeah. I mean, I would say that I'm definitely bullish on the AI front. I think our business, our industry is being transformed.
Jonathan Drouin: And I think it's hard to predict in two, three years what it will look like. But I think right now, we're in the midst probably of more opportunity in our business, in our industry than I've seen in the last ten, twelve years. So really excited to see how this technology will just transform where we're going. Again, as a company, as an industry, I think we're not even scratching the surface on what will be possible, so that definitely gets me going every day.
Jesse Buckingham: I love it. Jonathan, I always enjoy these conversations. Thanks so much for sharing your insights about how you've approached and tackled these.
Jonathan Drouin: Definitely key. Definitely key. Awesome. Thanks so much. Appreciate it. Thank you. Have a good one.