Bev Okoro is a Procurement Operations Lead at a mid-market packaging and fulfillment company in the Midwest. She oversees agent-generated purchasing decisions — shipping routes, vendor payments, inventory reorders — approving upwards of 300 per day. Before operations, she spent six years as an EMT. She agreed to this conversation on the condition that we note she is, in her words, "a composite of every ops person who has ever stared at an approval queue at 3 PM and felt their soul leave their body." We have honored this request.

You came to operations from emergency medicine. Do those feel like different jobs?

Bev: They feel like the same job wearing different pants. In the ambulance, you're triaging. This patient needs attention now, this one can wait, this one looks stable. My approval queue works identically. Vendor I recognize, amount looks normal, description matches the pattern. Next. The difference is that in the ambulance, when you triage wrong, you find out in minutes. In my queue, you might never find out. Or you find out six months later when someone in finance pulls a thread and the whole sweater comes apart.

Walk me through a typical day with the queue.

Bev: Mornings are when I'm actually good at this. Coffee's working, brain's online. I'll catch things. A reorder quantity that looks high for the season. A vendor I don't recognize submitting through a category they don't usually touch. I'll flag maybe five or six items before lunch. That's my real oversight window. Maybe two hours.

After lunch it shifts. The queue refills. By 2:30, 3 o'clock, I've got a hundred-plus items and my brain has decided that everything from Allied Packaging is fine because everything from Allied Packaging has always been fine. I'm not reviewing at that point. I'm scanning. There's a difference, and I know there's a difference, and I click "approve" anyway.

You've described a threshold rule you use. Only scrutinizing above a certain dollar amount. When did that start?

Bev: God, I don't even remember. It wasn't a decision. Okay, so one week early on, the agent was generating maybe 200 items a day and I was trying to actually review each one. Read the description, check the vendor history, verify the quantity against recent trends. I was getting through maybe 60 before the queue backed up so badly that my manager asked if the system was broken. It wasn't broken. I was just doing what the process document said to do.

So I started letting the small stuff through. Under a certain amount, known vendor, standard category? Approved. And nothing bad happened. Nothing bad happened the next week either. Eventually the threshold became the job. The formal process is still in a SharePoint somewhere. I read it once.

Research on decision fatigue suggests quality degrades measurably over a day. Judges grant parole at 65% in the morning and nearly 0% right before breaks.¹ Does that match your experience?

Bev: Here's the thing nobody tells you. I don't experience it as degradation. I experience it as efficiency. I feel faster in the afternoon. I feel like I've got the pattern down. But faster and better aren't the same thing, are they? I was a faster EMT in year four than year one. Some of that was genuine expertise. Some of that was that I'd stopped considering possibilities I should have been considering.

There's a concept in the automation bias literature called "weak automation bias," where you don't ignore contradictory evidence so much as stop looking for it.²

Bev: That's me. That's exactly me. I'm not ignoring red flags. I'm not seeing them because I've stopped looking at the part of the screen where they'd appear. The agent's been right, what, 99% of the time? More? At some point your brain just... why would you argue with that? It starts to feel like checking whether gravity still works.

What about the audit trail?

Bev: (pause)

Yeah. The audit trail.

I did not spend eight hours doing comprehensive review. I spent maybe two hours doing real review and four hours doing whatever it is I do in the afternoon. Pattern matching. Vibes-based approval. Whatever you want to call it.

And I know that if something goes wrong in that afternoon batch, the log shows I approved it. Not "Bev batch-approved this during a low-energy window without detailed review." Just "Approved. Bev Okoro. 3:47 PM."

Does that keep you up at night?

Bev: Not every night. Some nights.

Have you considered raising this with your manager?

Bev: (laughs) And say what? "Hey, you know that oversight function I'm supposed to be performing? I've been doing a different, worse version of it for eighteen months"? That conversation ends one of two ways, and I don't love either of them.

The alarm fatigue research says healthcare workers hear a thousand alarms per shift and somewhere between 72 and 99 percent are false.³ You know what those workers do? They turn the volume down. That's not negligence. That's survival. But try explaining that to a review board after something goes wrong.

What do you think you're actually providing? As the human in this loop?

Bev: (long pause)

Plausible deniability? No. That's too cynical. I catch things. I do catch things, in the morning, when I'm actually looking. And I have judgment the agent doesn't have. I know that a vendor who's been reliable for three years just got acquired, and the new parent company has a reputation. The agent doesn't know that. I do.

But am I providing what the process says I'm providing? Comprehensive review and informed approval of each item? No. I'm providing a morning filter and an afternoon rubber stamp and a name on a log that makes everyone feel like a human checked.

I used to save lives, you know. Now I click "approve." Some days those feel equally important. Most days they don't.

Footnotes

1. Danziger, S., Levav, J., & Avnaim-Pesso, L. — "Extraneous factors in judicial decisions," as cited in https://en.wikipedia.org/wiki/Decision_fatigue ↩

2. Springer Philosophy & Technology — "What is Wrong With Automation Bias?" (April 2026): https://link.springer.com/article/10.1007/s13347-026-01090-9 ↩

3. EBSCO Research — "Alarm Fatigue," citing 2023 study in Preoperative Care & Operating Room Management: https://www.ebsco.com/research-starters/environmental-sciences/alarm-fatigue ↩