The people most afraid of AI are usually safer than the people who aren't worried.

Here's the story the AI consulting industry likes to tell: automation takes the boring, repetitive work off your plate so your skilled people can focus on what they were trained for. The dishes get done by themselves. The humans do the cooking.

It's a comforting story. It's also mostly false.

Automation doesn't chase the lowest-value work. It chases the highest-cost repetitive work. Those are not the same thing. The $18 an hour admin doing data entry is less likely to be automated than the $180 an hour senior engineer assembling proposals at 11pm. The economics don't favor automating cheap work. They favor automating expensive work.

Walk through a mid-market professional services firm with this lens and you'll see it everywhere. The senior engineer pulling nights and weekends to respond to an RFP isn't doing low-value work. They're doing high-value work that the firm has been absorbing for years because the only alternative was not bidding. A proposal automation system is not coming for the admin assistant's job. It's coming for the senior engineer's evenings.

The same pattern holds for compliance. A P.Eng signing a drawing after a rushed review isn't performing low-value work. They're performing a legally binding act that requires licensure. An automated compliance check doesn't replace the license. It replaces the review the licensed professional used to do manually, badly, under pressure. Whether that makes the role more valuable (better reviews, faster signoff, fewer missed issues) or less valuable (why pay a senior engineer for a task the software handles) is a leadership call, not a technology one.

Senior designers who spend most of their day pointing people to data buried across four systems are not doing low-value work. They're doing knowledge work disguised as wayfinding. When that wayfinding becomes searchable, the role shifts. What they were paid for yesterday isn't what they're paid for tomorrow. That's a real change.

The people most confident that AI won't affect them are usually the ones doing knowledge work they've been performing so long they've stopped seeing it as a process. They assume the machine will come for the line workers, not for them. Then they watch someone demonstrate a RAG system that answers the questions they've been the sole source of answers for, and the ground shifts.

This is not an argument for or against automation. It's an argument against the specific lie that AI only affects "low-value" work. The lie is not accidental. It's strategic. It lets vendors sell into organizations without triggering the professional class's self-preservation instincts. It lets leadership green-light projects without having to answer hard questions about who's affected. It lets everyone involved pretend the thing they're doing is politically neutral when it isn't.

What happens in practice: the work most worth automating is the work most senior, most expensive people hate doing. Proposals. Compliance reviews. Data wayfinding. Calculation tools nobody else can maintain. Reporting into systems nobody trusts. These are the things that burn good professionals out, and they're also the things with the highest return if you automate them.

So the honest framing is this. AI is coming for high-cost repetitive work first. That often means the most expensive roles in your firm are the ones most affected, not the least. Acknowledging this out loud is the first step toward a transition that doesn't feel like a betrayal.

The people worried about being replaced are sometimes right. The people not worried about it are often wrong. The answer to that isn't reassurance. It's planning.

A firm that modernizes with honesty about who's affected, and what redeployment actually looks like, ends up with a senior engineer who's not doing proposals at 11pm anymore. A firm that modernizes while pretending nobody's affected ends up with a senior engineer who's not doing proposals at 11pm anymore, and who's also being quietly managed out because "the work changed."

Both firms saved money. Only one of them is still a good place to work in two years.

The senior people worth retaining can tell which firm they're in. They're watching the signals. How leadership talks about the transition before it happens tells them more than what happens afterward.