Plant Floor to Cloud: The Podcast

88% of manufacturers say they have the right AI expertise in the building. 22% are actually funding it. Everyone’s satisfied with the results — and the bill is quietly coming due somewhere they aren’t looking.

This episode’s thread: we’re buying AI as a technology purchase when it’s really a capability purchase. The software shows up on the delivery date. The capability to support it has to be built — and almost nobody is budgeting for it.

I walk through a quality early-warning model we built that caught a real field-failure event in back-testing — and still never made it past proof of concept. Not because it was wrong, but because nobody could explain how it was right.

In this episode:

• One question: what percentage of your last AI approval went to training?

• The story: we could check the answer, not the reasoning

• The telephone game — how model findings degrade across your org chart

• The Defensibility Test

• The Requisition Rule and the Named-Person Standard

• Quick hits and the industry pulse

• Coming up next month: the AI projects that actually paid

Plant Floor to Cloud is the monthly podcast companion to the weekly newsletter — mid-market manufacturing, practitioner-level detail, zero vendor hype.

Subscribe: https://plantfloortocloud.com

Go hug your ERP.

Newsletter Links:
We Shipped the AI Chatbot Everyone Wants. Then We Killed It in Eight Weeks. (August 4, 2026)

Somebody Has to Own the Kill Switch. Almost Nobody Does. (August 11, 2026)

The AI Got It Right. We Couldn't Tell If It Was Lucky. (August 18, 2026)

Lucky #13? 88% Say They Have the Talent. 22% Are Paying to Build It. (August 25, 2026)

RSM Survey:
RSM Middle Market AI Survey 2026 manufacturing snapshot 

A reordering agent ran quietly for six weeks, drafting purchase orders a human clicked to approve — right almost every time. So they took the human out of the loop, on a Friday. Over the weekend, a malformed price file turned dozens of individually-plausible orders into a Monday-morning mess nobody had actually signed off on. The model didn’t make a mistake — it did its job perfectly. It just wasn’t the job that mattered.

That one mix-up — mistaking the demo for the job — is the thread running under nearly everything that went sideways in this July’s newsletter.

From that single story, four ways to catch the gap before it catches you: the Leash Test, the Funding Test, the Friction Test, and the Plateau Test. Plus six Quick Hits on AI committees, digital twins, and outcome-based pricing loopholes, and a fresh cut of RSM’s 2026 manufacturing AI survey that validates the whole argument at industry scale.

No vendor hype. Ever.

Source newsletter issues referenced
"Four ways implementations go sideways" (Jul 7)
"Your AI dashboard is green. Your deployment might be dying." (Jul 14)
"The Weekend The Agent Did Its Job Perfectly" (Jul 21)
"B2B eCommerce stopped being a science project" (Jul 28).

RSM Middle Market AI Survey 2026 manufacturing snapshot 

“MSC Industrial Supply ecommerce grows as Q3 sales exceed $1 billion,” Digital Commerce 360

“Global Industrial's digital investments power retention as Q1 sales grow 9.2%,” Digital Commerce 360