We are all trying to figure out how AI will change industries. From the SaaS-pocalypse, to the CHRW stock rout, to the Citrini doom forecast, to the “nothing ever changes” fundamentalists, the takes have come hot and daily. I am incredibly bullish on the future of AI and automation in freight, but temper that with two decades of hard-won experience building technology and bringing revolutionary business models into the industry. Freight moves slowly, and there are real learnings from previous major technological shifts that can inform how AI will drive change. First, an anecdote.
In 2008, while I was working for Coyote, a couple of us went down to visit a customer, a plastics company southwest of Chicago. The facilities probably had not been updated since the ’70s, and as we were walking around the office, I noticed that every desk had four symmetrical holes. It was odd enough that I asked our host, “What are those holes for?” He responded, “Oh, that’s where the typewriters were. We just had them taken out last week.” I know some of you Zoomers probably think we still used typewriters in 2008, but this was a year after the first iPhone debuted, and at some plastics company in Chicago, management had ignored decades of progress to decide that typewriters were still good enough, 37 years after the first PC was released.
There is a natural adoption curve for any new technology. Don’t ever underestimate how long that tail can be.
The Slow March of Freight Technology
Freight technology has had many major transformations over the previous decades: containerization, xMS (TMS, WMS) proliferation; GPS and mobile proliferation; digital freight matching; cloud migration; RFID; an explosion of data tooling (BI, Big Data); massive advancements in analytical tooling; optimization model proliferation; and now the emergence of LLM’s, generative AI, and AI-native platforms.
New fundamental technologies, like cheap GPS chips, smartphones, or generative AI, consistently beget entirely new models, and many of these changes are still percolating through the market. When we launched CoyoteGO at Coyote, one of our key questions was whether enough drivers even have smartphones. In 2011, smartphone ownership nationally was just 35%, and even lower among truck drivers. No one asks about that anymore. Ask the same question about AI tools today and you get a similar answer: fewer than half of all American adults regularly use LLMs. It’s a reminder of where we are on the curve.
Another recurring theme in the freight technology evolution is that if you stick around long enough, you see the same problems solved and then solved again. Someone is graduating undergrad this year, thinking, “I can find a better way to optimize inventory,” decades of accumulated knowledge on inventory optimization notwithstanding. That’s the real opportunity for AI: not replacing human judgment, but tapping into a hundred years of supply chain expertise and making it accessible at scale. But I digress.
Four Waves, Three Decades
In the brokerage technology world, I’ve been lucky to be right in the middle of four distinct, overlapping waves, each of which unfolded over years and decades, not months.
Wave 1: The Four Walls Era (1990s to 2010s). The first wave was about making brokers faster inside the building. This was my focus at Coyote. How do we make our reps as efficient as possible? How do I empower them with up to date information on truck postings, bids, and available loads that are perfectly matched to their accounts. During this period, DAT went digital, TMS platforms simplified load tendering and carrier tracking, and CRM tools replaced Rolodexes. The productivity gains were real but narrow. People were still on the phone at both ends of the transaction. Technology in this era was about augmenting the broker, not changing the basic roles on the brokerage floor. Changes were incremental and the adoption curve was long. Many brokerages did not fully deploy a modern TMS until well into the 2010s, and it would not surprise me at all to find brokerages still running on AS/400s or spreadsheets today.
Wave 2: The Smartphone Revolution (2010s to 2020s). The iPhone put a computer in every driver’s pocket, and the implications for freight were significant and are still materializing. I had many shippers tell me when we launched Uber Freight that instantly after using Uber for the first time, they thought, ‘why don’t we have this for freight?’ The ELD mandates, mobile tracking apps, digital POD capture, and carrier-facing portals all emerged from this wave. The driver became a connected endpoint: reachable, trackable, empowered. Yet, adoption of driver-facing technology is still uneven. The cohort of app-centric carriers continues to grow, but a decade-plus after the smartphone revolution, the freight industry is still digesting it.
Wave 3: Digital Freight Platforms (2015 to 2022). Uber Freight, Convoy, Transfix, and others introduced something genuinely new: instant booking via upfront pricing, automated bidding and response, and the normalization of self-service booking and load execution. They (we) attracted billions in venture capital and reshaped expectations around pricing transparency. From the inside, it was an amazing period of innovation and talent injection into the freight tech space. Of course, the dream and the reality didn’t line up perfectly, and building freight technology and building a viable freight business proved to be very different things. Convoy shuttered in 2023. I then led the acquisition of the tech, rebirth and sale to DAT of the ‘New Convoy’, foundational to the automation vision that we are now bringing to market at DAT. Uber Freight shifted its ambitions with the acquisition of Transplace, becoming more traditional in many ways through that process. Transfix also pivoted to pure tech. I could write many longer articles about my thoughts on the why and how of how this era of digital brokerages ultimately fizzled out, but the underlying technology worked. Automation is real, and through our efforts at DAT, the market will continue to reap the benefits.
Wave 4: AI-Native Solutions (2022 to Present). AI is eating everything. We’ve already seen a surge and then levelling off in the scaling of AI tech companies in logistics, Happy Robot, Augment, Pallet, Fleetworks, and on and on came into the space with a head of steam, demonstrated real value, and have evolved and pivoted as the limits of impact get tested. We are early in this wave, but it’s already qualitatively different from the prior three: for the first time, the technology is performing tasks that previously required human judgment, not just human time.
How AI Will Be Different, and Why It Will Still Take Longer Than You Think
Understanding how AI adoption will unfold requires recognizing that we are not experiencing a single technological shift but a cascade of three interwoven solutions.
Phase 1: Generative AI (2022 to 2026, ongoing). Large language models can draft emails, summarize load details, generate carrier outreach scripts, answer shipper questions via chat, and assist with RFP responses. The potential use cases are limitless, but the practical use cases have proven to be supplemental, focused on automation or elaboration. The productivity gains are real and measurable, but often incremental. The org chart does not change, the business model does not change, and the strategic moats remain. Generative AI is an enabler, not a new operating model.
Phase 2: Agentic AI (2025 to 2030, emerging). Agentic AI is where the architecture begins to shift, and expand the scope to full displacement of manual processes and even roles: monitoring load boards, identifying coverage risks, reaching out to carriers or brokers, updating information, and escalating exceptions, all without a human in the loop. We started using Happy Robot and others fairly early in this journey, and have seen direct measurable results. In brokerage the pitch can sometimes hit walls when the intangibles of a role are implicit to the core value prop of a broker. If carrier sales is your advantage, then what advantage do you have when a 3rd party provider is doing all of your carrier sales with AI call bots? These would have been unimaginable conundrums 10 years ago, but I’ve heard that exact objection more than once. I think the answer is that you build back in differentiation through expertise in how to manage the AI agents and solutions. To truly build value, you have to internalize the intuition and expertise with how to build and manage the agents. The anti-pattern I’ve seen is when teams throw their hands up because the agent can’t do EXACTLY what they do. It won’t. It’s different, but learn and it tune it to your needs.
Phase 3: Embodied AI and Robotics (2028 and beyond). The longest arc belongs to physical automation. Autonomous trucks, robotic dock operations, and AI-orchestrated inventory management, material handling, and warehouse systems represent the embodied phase of AI in logistics. The technical barriers are real, but LGV’S are also loading trucks at Niagara facilities today, and Aurora is delivering loads without drivers. If you haven’t used a RoboTaxi or Waymo yet, do. It will change your world view. The tech works. Full-cycle automation from shipper dock to consignee dock without human intervention will be a decades long journey. The regulatory environment may be uncertain, and the capital requirements are significant, but believe it. It’s coming. This will be the single most disruptive change on the horizon, and will impact all of our businesses in the space. Don’t ignore or dismiss it, but also know that it will take a long time to scale. Think through how you will adapt.
Lessons from the Trading Floor
There are several parallels in the financial and banking industry that I see as informative to thinking about the transition we are going through today. When ATMs arrived in the 1960s and 1970s, there was much ink spilled over the pending elimination of bank tellers, but contrary to belief, teller headcount actually grew through the 1980s and 1990s as banks opened more branches, enabled by the higher efficiency and lower cost to operate. Conversely, when electronic trading platforms arrived in the 1990s and 2000s, it impacted entire categories of floor traders and back-office clerks. The bustling trading floor is a relic of history today. ATM’s enable tellers to focus on higher value tasks and scaled teller count. Electronic trading automated execution and eliminated whole categories of roles.
The analogy carries forward, and I expect we will see a bit of each as AI continues to scale. Much as we still have tellers, financial advisors and fund managers, I never see the need for expertise and relationships waning. Shippers will still look to experts, partners, and operators for help. The broker as freight advisor, risk manager, and procurement partner persists. The execution gets progressively easier and lower touch.
The Block Signal: What Dorsey’s Layoffs Portend for Transportation
The freight industry would do well to pay attention to what happened at Block on February 26, 2026 and what is now happening at Meta and Amazon. Jack Dorsey’s company, parent of Square, Cash App, and Afterpay, announced it was cutting more than 4,000 employees, 40% of its global workforce, reducing headcount from over 10,000 to just under 6,000.
The business itself is strong: full-year 2025 gross profit grew 17% to $10.36 billion, and Cash App gross profit surged 33% in Q4. Internal AI tools, including a proprietary automation platform called Goose, had fundamentally changed how work gets done. Smaller, flatter teams using AI could replicate and exceed what larger teams had done before.
In his shareholder letter, Dorsey was blunt: he doesn’t think Block is early to this realization. He thinks most companies are late.
“Within the next year,” he wrote, “I believe the majority of companies will reach the same conclusion and make similar structural changes.” Wall Street’s reaction was instructive: the company’s stock rose 24% on the news, a signal that investors believe smaller AI-augmented teams can sustain or accelerate growth at meaningfully lower cost.
What does this mean for transportation? A few things.
First, the back-office functions where agentic AI is advancing fastest are precisely the ones that have always been the most labor-intensive: carrier sales, load booking, document processing, customer service, billing, and collections. A mid-size brokerage running 200 operations headcount today may find that 80 to 100 people supported by AI agents can handle the same throughput within the next three to five years.
The question isn’t whether this happens, but who navigates the transition proactively and who gets forced into it.
Second, the Block case illustrates a dynamic that will repeat across industries. Companies with strong financials and proactive leadership will restructure on their own terms. Weaker operators will face the same pressure without the balance sheet to manage the transition well. Freight is already consolidating toward larger players, and AI accelerates that trend. Well-capitalized brokerages and 3PLs will have the resources to invest in AI infrastructure and manage workforce transitions thoughtfully. Smaller and mid-sized brokerages are more exposed and have less runway to figure it out, but there is a whole ecosystem of tools developing to help you out. Small means flexible, so be flexible.
Third, and this is the point I want to sit with: notice what the Block story is not. It’s not a story about a new business model. Block is still a payments company; Cash App is still Cash App. The product did not change. The cost structure changed. That distinction matters. The first-order effect of AI in transportation will be margin expansion and headcount rationalization within existing business models, not the wholesale disruption of how freight moves or how brokerages create value. Trucks will still move. We are in the transformative chapter. The true revolutionary chapter is still being written. AV is coming, and you’ve got time to figure out how to play it.
Which brings us back to those four holes in a desk in a Chicago suburb in 2008. The typewriters were gone, but only just, 37 years after the PC made them obsolete. AI will not wait 37 years to reshape freight. The pace will be faster. But the long tail is real, the adoption curve is real, and the gap between the leading edge and the median operator is going to be wider and more consequential than in any prior technology wave.
The long game is still a game. Play it accordingly.
