OpenAI's Enterprise Push: What the Press Releases Aren't Telling You

Does This Matter for Industrial Automation?

Here's the thing. The press releases talk about healthcare, life sciences, financial services. Notice what's missing? Manufacturing. Warehousing. The dirty, loud, physical work.

Now, I called an old colleague at Siemens last week (won't name him, he's still employed) and asked what they're seeing. His take was that the language models are genuinely useful for documentation, troubleshooting guides, and training materials. Where they fall apart is anything that needs to understand physical context. A robot arm doesn't care how eloquent your instructions are if the gripper can't handle the part tolerance.

That said, the ServiceNow integration could matter. A lot of maintenance workflows, spare parts ordering, technician scheduling, that's all ServiceNow territory. If GPT-5 can actually make those workflows smarter, that's not nothing.

What About the Jobs Platform?

This is the part that caught my attention. OpenAI says they're launching certifications and a jobs platform to "connect workers with jobs, training, and certifications."

I've seen this movie before. Back in the PLM software boom, every vendor had their certification program. Some were worthless paper, some actually meant something. Too early to say which way OpenAI's will go.

What I do know is that the skilled trades are already short-staffed. If these certifications help maintenance techs learn to work alongside AI systems, great. If they're just another checkbox for HR departments, we haven't solved anything.

The Skeptic's Take

Look, a million business customers is impressive. But "business customer" could mean anything from a Fortune 100 company to a three-person startup with a ChatGPT subscription. The number by itself doesn't tell us much about depth of adoption.

OpenAI also announced some leadership changes, which the tech press covered extensively. I don't have much to say about that. Companies reorganize. What matters is whether the products work.

The honest assessment? GPT-5 probably is meaningfully better than GPT-4. The enterprise integrations probably will make some workflows easier. But the gap between "AI-powered enterprise workflows" in a demo and reliable automation in a facility with 15-year-old PLCs and spotty WiFi remains unclear. Nobody's publishing case studies on that yet.

What I'm Watching

I want to see actual deployment numbers in manufacturing and logistics, not just "business customers." I want to see the ServiceNow integration handle a real maintenance workflow, not a scripted demo. And I want to see what those certifications actually teach.

Until then, I'm filing this under "promising but unproven." Which, in fairness, is better than most of what comes across my desk.

Aisha Patel · 1 hour ago · 7 min