# Tech Is Dumb Out of the Gate > Automation doesn't eliminate the need for the people who know the work. It depends on them. The warehouse worker who trained the system and got promoted is... Published: 2024-11-12 · Updated: 2026-06-23 · Topics: AI & Machine Learning, Digital Transformation, Growth Leadership, Team Building & Culture, Product Strategy & Development · Author: Alan Wizemann There is a warehouse worker I keep thinking about, long after the project that introduced me to his story has ended. Before we brought automation into the pick-pack and shipping operation at one of our warehouses, his job was to sort broken cases by hand. That is the kind of work that sounds simple until you watch someone do it well – when a bar orders two bottles of one thing and three of another, somebody has to pull those items and stage them correctly for the delivery. He had been doing it long enough that he carried a mental map of the entire area in his head, knowing exactly where to position each product category so that his path through the warehouse took the fewest possible steps. Over years of doing the work by hand he had developed staging sequences that no system we were bringing in had any concept of. When automation arrives in a situation like that, the assumption most people make is the obvious one (that his job is gone). What actually happened was almost the exact opposite. He became the subject matter expert who trained the system, and the routing logic, the staging algorithms, the decisions about how to organize the pick paths all got better precisely because they were built around what he already knew. He did not just help configure the initial setup either. He was then put in charge of redesigning the staging areas and managing the whole operation, so his career advanced because of a change that looked, from the outside, like exactly the kind of thing that eliminates the people doing the work. I tell that story because it captures something that gets consistently missed in conversations about automation and AI in the workplace, which is that technology is dumb out of the gate. A new system, no matter how sophisticated it looks in the demo, simply does not know how the work actually flows in your specific environment. It does not know that certain products get picked together because of how a delivery route happens to be laid out. It does not know the shortcuts a veteran employee has developed over a decade of doing something that looks simple from the outside but very much is not. That knowledge lives in the people who do the work. If you bring in automation without making them part of how the system learns, what you get is something that technically functions but performs well below what was actually possible – a machine running on someone else's assumptions instead of the hard-won ones already sitting in the building. The framing I see in most headlines about AI in the workforce is wrong in a specific and, I think, revealing way. When a large company announces that it has replaced thousands of employees with AI, the coverage treats the announcement as evidence of what the technology can do. I read it almost the other way around. Those employees did not stop having value the morning the press release went out. The company made a decision about how to use the capability AI had created, and it chose to reduce headcount rather than redirect that freed-up capacity toward the things that actually require human judgment. That is a choice, and it is worth naming it as one, because it is not an inevitable outcome of the technology simply existing. The question we ask when we bring technology into an operation at Southern Glazer's is not how many roles we can eliminate but something closer to its inverse: what are the people doing right now that they should not have to do, and what would they do with that time if we took it off their plate? When we brought in document processing automation to handle a high volume of support emails and paperwork that used to require people to read and route everything manually, the honest truth was that the people doing that work were overqualified for it. Their experience and judgment were spent on a task a system could handle, and freeing them from it was never a cost-cutting exercise so much as a decision about where their skills were actually most valuable. This matters especially in a relationship business, which is exactly what we are. Southern Glazer's runs on the relationships our sales team has with customers and the ones our delivery drivers build with the accounts they serve every single week. So every hour a salesperson spends on data entry or order management that a platform could just as easily handle is an hour they are not spending with a customer. The automation is not replacing the relationship at all; it is protecting the time that makes the relationship possible in the first place. What the warehouse worker's story illustrates is something I now see consistently across every automation project we run, which is that the people who do the work closest to the ground are the ones who actually know what the work requires. The drone that checks shelf inventory in our warehouses is more accurate because the people who used to do those checks by hand helped calibrate how it should operate. The machine learning that reads planograms is better because the people who used to scan them already knew what the edge cases looked like long before the model ever encountered one. There is always, in my experience, a people component to automation working the way it is supposed to. That holds not only at the start during configuration but continuously, as the system keeps running into situations the original training data never covered. The human in the loop is not a limitation on the technology; it is the thing that makes the technology worth the investment at all. So when people ask whether the work can be automated, I think they are asking a question that has already been answered – a surprising amount of it can. The question that actually matters, the one that separates a company that uses this moment well from one that merely follows the headlines, is whether you use that capability to take the plate away from the person holding it, or to clear it so they can finally do something more with what is left. --- Canonical: https://alanwizemann.com/articles/tech-is-dumb-out-of-the-gate