A few months back I sat in on a planning meeting with a distribution client working through next quarter’s reorder points. Their Operations Manager had three spreadsheets open next to the ERP, cross-checking numbers by hand because she didn’t fully trust any one of them on its own. At some point she said, half joking, “I don’t need the system to think for me, I just need it to stop giving me homework.” That line has stuck with me more than anything I’ve read about AI in distribution this year, because it’s a much better description of what’s actually happening on warehouse floors than the “AI will transform your supply chain” headlines suggest.
Here’s what I mean: MHI’s 2026 industry report with Deloitte came out a while back, and one number jumped out at me: nearly half the supply chain leaders surveyed now see AI as significantly disruptive to their operations, up 25 points from the year before. Read on its own, that sounds like the setup for a story about job losses. But the same research also ranks the talent gap as the number one problem distribution companies are dealing with right now. So you’ve got leaders bracing for disruption at the exact moment they can’t find enough people to run the warehouse they already have. The National Association of Wholesaler-Distributors, which ACE Micro belongs to, put out its own guidance on this last year that says pretty much what I’d tell you if you asked directly: AI in distribution isn’t built to replace the people doing the work, it’s built to take the rote parts off their plate. Something doesn’t add up if you take the “AI replaces workers” framing at face value.
What I’ve actually seen change, talking to people who run distribution operations, is closer to what happened when power tools showed up on job sites. Nobody stopped needing a carpenter once drills replaced hand screwdrivers. What changed is which parts of the job ate up the day. A buyer I know used to spend two or three hours a week just assembling demand data before she could even start making a call on a seasonal item. Now the pulling-together part happens inside the system, and she still makes the same judgment call, she just gets to it faster. A customer service rep doesn’t stop handling weird orders because AI reads a scanned PDF and drops it into the ERP. They stop being the one who types it in line by line, and start being the one who catches it when the AI gets it wrong.
That last part is the one people underestimate. AI reading a messy order or flagging a reorder isn’t the same as AI being right about it. There’s a researcher at MIT’s Intelligent Logistics Systems Lab, Matthias Winkenbach, who’s tracked this same pattern across a bunch of warehouses: less time executing, more time on oversight and exception handling. That tracks with everything I’ve seen. Someone still has to catch the mistake, and that job, watching for the exception instead of doing every step by hand, is a different kind of tiring than the old version, but it’s not nothing. This is where a lot of the “AI empowers workers” messaging goes a little soft, honestly. It’s not empowering in some abstract sense. It’s more that the parts of the job that used to be pure grind get smaller, and the parts that actually need a person get more concentrated.
That concentration matters more in some corners than others. A distributor moving pallets of shelf-stable goods can live with an occasional bad AI suggestion. One moving lot-controlled medical supplies or industrial parts can’t, not without it turning into a compliance problem instead of just an inconvenience.
Where this gets complicated is when the system underneath doesn’t support any of it. I’ve talked to more than a few Microsoft Dynamics GP shops lately who want the AI-assisted forecasting or the automatic exception flags that show up in Business Central’s Copilot, and it just isn’t there in Dynamics GP, because it was built in an era where the point of the software was recording what already happened, not suggesting what to do next. That’s not a knock on GP, it did what it was built to do for a long time, but it’s not the platform this next stretch of work is happening on, and I don’t think that gap closes with an add-on. It’s one of the clearer signs a GP shop has outgrown the platform, not a dramatic one, just a real one.
I don’t have a tidy conclusion here, other than that the honest version of this story is a lot less dramatic than either “AI takes your job” or “AI saves distribution.” It’s mostly about which parts of a workday get smaller and which parts get more concentrated, role by role. That’s worth more attention than it’s getting.
President and VP of Business Development
read full bio