The Tax Guy Who Quit Told Me More Than Any Labor Report

One of our best tax guys left last month. Not for another firm, not for partner track somewhere shinier. He left to join an AI company's tax research group—to help build the LLMs that will do tax work.

I figured it was a payday. Tech money. Stock options. The usual story.

So I sat him down before he left. "What's the number?" I asked.

"I'm not chasing a payday," he said. "I want to understand what cutting edge looks like."

That conversation told me more about the future of professional services than any McKinsey report on AI adoption. Because by the time the data shows up in workforce studies, the best people have already repositioned themselves.

The Studies That Aren't Actually Contradictory

Two pieces of research landed the same week. The European Central Bank published findings showing AI's effect on US jobs and wages remains muted. No employment crater. No wage collapse. The numbers look... fine.

That same week, The Wall Street Journal reported that record numbers of top graduates are skipping traditional employment entirely to call themselves "founder." They're building, not joining.

Most people read these as contradictory. Nothing's happening vs. something's happening.

They're not in tension. They're sequential.

The market reprices in two steps, not one: first in ambition, then in statistics. The talent moves before the spreadsheets catch up. My guy didn't frame his decision as fleeing a sinking ship. He framed it as moving toward where the value is being created. But the signal is the same either way.

The Wrong Question Gets The Wrong Answer

We keep asking: "Will AI replace jobs?"

That's not how disruption works. A job title rarely disappears overnight. What erodes is leverage. The raise shrinks. The rate you can charge stops climbing. The scope of what clients will pay you to do narrows. The work survives. The pricing power doesn't.

I watched this exact pattern play out in software engineering twenty years ago. When offshore development hit scale, senior developer jobs didn't vanish. But the premium for writing straightforward code collapsed. The developers who saw it coming moved upstream—into architecture, into product, into the problems that still commanded pricing power.

The ones who didn't? They're still writing code. They're just getting paid less for it, adjusted for inflation and expectations.

Kodak's Revenue Looked Fine Until It Didn't

Here's what makes this pattern hard to spot: the financial indicators lag by years.

Kodak's revenue in 1999 was $14.1 billion. In 2000, still over $13 billion. The business looked defensible. Film wasn't dead. The income statement said everything was manageable.

But if you'd been watching where Kodak's best engineers wanted to work, you'd have seen the signal much earlier. They were leaving for digital imaging companies. Not because Kodak was failing—because they could read the trajectory. The smart money, the ambitious talent, the people who had options—they moved toward where the value creation was heading.

By the time it showed up in revenue, it was too late to reposition.

I've now watched this movie four times: enterprise software moving to cloud, media moving to digital distribution, retail moving to mobile-first, and trading floors moving to algorithmic systems. The pattern is always the same: the talent spots the shift in leverage before the financials do.

What Losing Pricing Power Actually Looks Like

My tax guy wasn't running from tax work. He's running toward understanding how AI models learn tax logic—because that's where the next decade of leverage gets built.

Here's what he understood: routine tax research is about to lose pricing power. Not disappear. Lose leverage.

A client used to pay $X for an associate to spend six hours researching a technical question. When an LLM can surface the same answer in six minutes, the client still needs judgment about what to do with that answer. But they won't pay the same rate for the research itself.

The senior judgment? Still valuable. Maybe more valuable, because now you can exercise it on ten situations instead of one. But the pyramid under it—the leveraged model where partners bill out associates doing research—that pricing model is under pressure.

The associates who see this coming have three moves:

1. Move upstream to the judgment layer faster than the traditional timeline

2. Move sideways into the firms building the tools

3. Move out to build their own leverage somewhere else

My guy picked option two. Not because tax is dying. Because he's repositioning himself on the side of the shift that's gaining pricing power, not losing it.

The Leading Indicator You're Not Watching

So here's the boardroom version, the question you should ask Monday morning:

Where do your best young people want to go?

Not where they're going for money—people chase comp in every direction. Where are they going when they take a lateral move, or even a step back in cash, because they see something in the trajectory?

When your high-performers start choosing leverage somewhere else over climbing your ladder, they're telling you which of your roles is about to lose pricing power. They're reading the map. Usually 12-18 months before it shows up in your billing realization rates or client price sensitivity.

I'm watching where RSM's early-career talent looks when they get recruited. Not just who leaves—that's lagging data. Where they're curious. What side projects they start. Which internal initiatives get the talent volunteering vs. the ones we have to staff with mandates.

The pattern I'm seeing: people want proximity to where the models get built, not just where they get deployed. That's the signal.

The Uncomfortable Question

Here's what I don't know: whether I'm reading this right.

Maybe my tax guy is chasing a mirage. Maybe the LLMs hit a capability wall and tax research stays human-leveraged for another decade. Maybe I'm overindexing on one data point because it confirms my priors.

But I know this: the people with the most options tend to be right about where value creation is heading. Not because they're smarter. Because they have the freedom to act on pattern recognition without waiting for certainty.

Are you tracking where your best people's attention is going? Not where they are today—where they're leaning?

Because by the time the labor studies catch up, by the time it's obvious in the billing data, the people who could have repositioned your practice will have already repositioned themselves.

I lost one to an AI company. He wasn't chasing money. He was reading the map.

Are you?

What to do Monday morning: Pull your list of high-performers under 35. Not to retention-plan them—to interview them. Ask what they're curious about. Where they see leverage being created in five years. What they'd build if they could. Listen for the pattern in their answers. That's your leading indicator, not your attrition rate.