Invisible by Design
Over the past two weeks I’ve been on the road – Normal, Illinois for a trucking conference, Chicago for a freight and supply chain conference, and Vancouver for a construction conference. Three very different industries, but one theme kept surfacing: buyers are ready to adopt AI, but they won’t tolerate clunky or confusing user experiences. In industries where every minute counts, the best tools are the ones that blend seamlessly into the workflow.
What struck me was how similar the adoption pattern sounded. Across logistics yards, construction sites, and dispatch offices, everyone said the same thing: we’re open to AI, as long as it doesn’t slow us down. The appetite for automation is high, but patience for friction is zero.
Usability Drives Adoption, Not Features
Buyers have no interest in training sessions. Almost every executive I spoke with had a story about trying to roll out a new system and watching adoption plateau after the first few weeks. Teams learn the one feature they need, and everything else sits dormant. Months later, leadership realizes they’re using maybe 10-20% of what they’re paying for. The problem isn’t the technology’s potential, it’s that employees don’t have time to re-learn how to do their jobs. In industries like construction or logistics, the workday is already chaotic. If a tool isn’t immediately intuitive, it becomes shelfware.
Even the most widely adopted tools face this challenge. In construction, for example, Procore is now standard across most firms, but many teams only use a fraction of its capabilities. It’s a platform that connects preconstruction, financials, field management, and analytics, enabling custom workflows, approvals, and integrations across the project lifecycle. It also serves as the central hub for collaboration, keeping every stakeholder aligned on the most up-to-date documents. Yet several construction General Contractors (GCs) told me that while Procore is now standard across their organizations, most teams use it primarily for one thing: document distribution. As the product has expanded, its interface has grown more complex – limiting deeper, day-to-day engagement.
If a market leader like Procore struggles with day-to-day engagement, the stakes are even higher for AI startups without a foot in the door. When founders pitch ambitious copilots or autonomous workflows, they need to remember that adoption hinges on immediate, intuitive usability. If it takes more than a few clicks or worse, requires training, it will be significantly harder to gain adoption. While it may be the C-suite that signs your Agreement, it’s the people using your product that will ultimately determine whether you succeed or fail.
Start With the Stack They Have
Another theme that kept surfacing in my conversations: buyers don’t want to rip out the tools their teams already know, even if those tools are imperfect. For example, a construction estimator may rely on Procore for document distribution, SharePoint for storage, and Excel for detailed cost modeling. Any new product that tries to force them to abandon Excel will almost certainly fail.
Therefore, AI needs to be the conductor, orchestrating across all of their internal systems to extract the necessary data and then structuring it in Excel (or whatever format they are comfortable with).
This is the key design challenge for vertical AI right now. It’s not about building standalone systems that demand new workflows; it’s about building orchestration layers that make existing workflows faster, cleaner, and more accurate. Success comes from fitting into existing workflows instead of forcing change. Only after trust and dependence are established can you pull users into a more comprehensive platform.
Accuracy is non-negotiable, but perception matters too
At one conference, I spoke with the owner of a large drayage carrier (over 300 trucks and 200 chassis). He explained that his team relies on offshore contractors to manually enter data into their TMS but acknowledged that the process was “far from perfect” due to frequent human errors. Even small mistakes, he said, can cascade into costly downstream issues: misrouted containers, compliance fines, or hours lost chasing corrections.
This is where AI can shine. It doesn’t have to be perfect, but it has to feel more reliable than the messy baseline. In most of these workflows, “better” doesn’t mean flawless; it means faster, more consistent, and less error-prone than the human process it replaces. That perception of reliability is built not just on accuracy, but on how clearly the system communicates what it’s doing. The interface plays a crucial role here: showing its work, surfacing confidence, and giving users quick ways to verify results. If checking the AI’s output takes longer than doing the task manually, adoption stalls. As one construction GC put it: “If I have to go back and check AI’s work, how much time is it really saving me?”
Design Is the Differentiator
One of the immediate hurdles to getting widespread adoption is building trust with end users. To do this, AI needs to:
Work inside the systems teams already use.
Surface answers with accuracy, context, and the option to drill deeper.
Allow easy verification.
This means interfaces need to balance speed with reassurance. This can look a lot of different ways (green-yellow-red states, inline confidence markers, side-by-side comparisons with the “old way”), and is best discovered by sitting next to your customers to see how they work.
Walking away from three conferences across three industries, it’s clear AI adoption will be won or lost at the interface layer. As agentic capabilities expand, that may change (if a tool truly functions as a “drop-in AI workforce,” usability might matter less than raw output). But for now, in a world where most products still compete on similar capabilities, design is the differentiator. For trucking dispatchers, construction estimators, or freight forwarders, it doesn’t matter how powerful the model is if the interface makes them feel slower, dumber, or less in control. In the end, adoption comes down to trust – and trust is built through quiet, dependable design.
The winning products will look unremarkable because the best design is often the one that disappears.