This is part two of The Squeeze, and it is about accounting advisory pricing — specifically, the question your fee is built on, and why AI just changed it. Last week, in part one of this series — "When the close costs $49" — I left you with a promise: there's something the four-way squeeze on the monthly close can't touch, and it has a name. Here it is. In that same room of small-firm owners, one owner cut in with the question that actually matters: if AI is doing the books, who's overseeing it — prompting it, checking it, driving the ship? The moment she asked, the ground shifted. That question is a different question than the one our profession has answered for 200 years.
The question every prior shift asked
Think about what desktop software changed, then the cloud. Each time, the debate was the same: who does the work? You, your staff, an offshore team, the client in their own ledger. The work was a fixed quantity of effort, and the only variable was whose hands did it. Pricing followed the hands — more work, more hours, more fee.
AI breaks that. The categorizing, the reconciling, the first draft of the tax return, and the first draft of the advisory report are starting to do themselves — and the direction isn't subtle.
In a June 2026 benchmark, Digits clocked transaction creation at 34 seconds for a human against 0.53 to 5.22 seconds for AI — between 6.5 and 64 times faster. It's Digits' own test, so weigh it accordingly, but the order of magnitude is the point. And a bookkeeper has maybe 40 productive hours in a week; the software has 168, and it never tires. The work isn't the scarce thing anymore — or it soon won't be.
So "who does the work" stops being the interesting question. The new question — the one that owner put on the table — is who's accountable for the work being right.
Rearview mirror, dashboard, windscreen
Here's the way I've started picturing it. For 200 years, our profession has been the rearview mirror. We tell you what already happened — last month's books, last year's return. Accurate, necessary, and entirely about the past.
We've been moving, slowly, to being the dashboard: what's happening right now. Real-time numbers, a close that's never more than a day stale. That's better, and AI makes it normal rather than heroic.
But the fee is migrating again — to the windscreen and the GPS. Not "here's what happened" or even "here's where you are," but "here's what's ahead, here's the hazard, here's the road I'd take." Nobody pays a premium for the rearview mirror once the mirror installs itself. And it's nearly there — AI already drafts much of the bookkeeping and will soon prepare the bulk of straightforward tax returns, the rearview work that anchors your fee today.
Be clear about what the windscreen actually is: it's the advisory layer — the guidance, the analysis, and the business improvement you give the client. It's built on the hindsight sitting in their numbers, sharpened by AI tools that spot patterns and model options faster than you could by hand. It gives the owner visibility into where they stand against their goals, and holds them accountable to the plan. The rearview is still the raw material — the value is in what you help the client see through the windscreen, not the mirror.
Knowledge stopped being the moat
The uncomfortable corollary: knowing the rule is no longer worth much. "Can I deduct this in my state?" used to be a question only you could answer. Now your client asks an AI and gets the same answer you'd give. Knowing the rule is free.
What isn't free: knowing this business well enough to catch when the confident AI answer is wrong for them specifically, and being the person whose name sits on the filing when it matters. That's not knowledge. That's judgment and accountability — and they're scarce in exactly the spots where AI is most confident and most occasionally wrong.
So is quality control the answer? Not quite
It's tempting to say the new job is quality control — checking the AI's work. That's half right, and the half that's wrong matters. AI is getting better at quality control too; before long it may catch routine errors more reliably than a tired reviewer at 6pm. If your pitch is "I check the machine's work," you've picked another race against the machine.
The durable version is sharper. It's not that you check better — it's that you're accountable, and you can show your checking. A logged, inspectable record of what the AI got wrong and what you corrected is something a client can see and an auditor can test. "Trust my judgment" is what every competitor says. "Here's the correction trail with my name on it" is something the $49 service structurally cannot offer.
What this means for your team
This reframes the bookkeeper, the tax preparer, and the staff accountant — it doesn't erase them. The role moves from producing the work to monitoring the system, reviewing the exceptions it flags, and owning the corrections. One person can hold far more clients that way — the constraint stops being how fast they process and becomes how well they judge. If you're a staff member reading this, that's your career: be the person who catches what the machine misses and can prove it, not the person who races it on speed.
So the product was never the close. It's the windscreen, the accountability, and the correction trail behind both. The question is how you package and price that when the cheap compliance tier is collapsing around it — and that's exactly what I'll solve next Friday in part three, "Stop selling compliance on its own."
If you want help making that move — repricing around the windscreen instead of the rearview — that's the work we do in AI Practice Transformation. The next cohort starts July 10. Find out more at theaiaccountant.ai/transformation.