(Wolters Kluwer covered my trip to CCH Connections and is a paid partner. The opinions are mine.)
I watched an AI prepare a 1040 in a few minutes. Live, no screenshots or pre-recorded video.
Wolters Kluwer ran the demo at CCH Connections in Orlando.
A preparer uploaded last year’s return, a W-2, and a brokerage statement. The software extracted the data, lined up current year against prior year, filled in the return, ran diagnostics, and flagged potential issues for a reviewer. Then it produced the workpaper and the PDF.
E-file isn’t in the product yet, but Wolters Kluwer says it’s coming.
It was a simple return. A good preparer might need 15 to 30 minutes. I’d need 30, and I’d probably fat-finger a number. The software needed a few minutes.
Impressive. So what does a firm do with the time?
We discussed the right and wrong answers on Episode 511 of The Accounting Podcast.
Will hourly billing eat the savings?
Many firms still bill hourly for compliance work. Cut the prep hours, and you cut the revenue. The easy fix is to give each preparer more returns and keep everyone billing.
Questian Telka of Reqoncile Financials joined me at CCH Connections to talk about this on the showcase floor. Everyone is talking about how this tech will solve the burnout problem in the profession. But hourly billing turns the savings into more volume.
I’ve argued for years that firms should price on value, not hours. This could finally force the issue.
Firms that keep selling time have no reason to give any of it back.
Reviewers inherit the pile
Say AI cuts prep time by a factor of 10. A firm that piles on more returns also feeds more work into review.
Every return still needs a human review, and that person can’t speed up by a factor of 10. AI can flag issues, but somebody still has to read them. Easier review of more returns is still more review.
The calendar doesn’t help
Questian pointed out that K-1s for high-net-worth clients often show up around September 15, if you’re lucky. The deadline is October 15. Faster prep won’t change either date.
You still need to turn all of those reviews around in about 30 days.
Tired reviewers trust the machine
Any manager with too many clients knows the temptation to accept what’s in front of you as given. Now imagine it comes from an AI that’s usually right. That’s how professional judgment erodes.
How often do your best reviewers catch an error on the first return of the morning? Compare that to how often they catch it on the fortieth return they’ve reviewed that week.
I covered a related result in a recent post about a Mercor study, Adding a “Human in the Loop” Made Claude Worse. Twelve CPAs worked on month-end close tasks with Claude’s help. They took about 15 times as long as Claude alone, and in some tasks they changed correct numbers into wrong ones.
A person in the middle of an AI workflow can slow it down and make it worse. Tax review puts a person in that seat, but oversight is still necessary.
Price the work, then sell planning
The other path to a return on AI is more revenue. So charge for the value of the return instead of the hours it takes. Then use that freed-up capacity on tax planning, which many firms don’t sell today.
Wolters Kluwer previewed a planning tool in the general session. It can model entity changes, such as a partnership switching to a C corporation, along with the savings.
The tool isn’t available yet, so there wasn’t a lot to see.
Still, a firm that sells planning can prepare fewer returns, earn more, and keep team members from drowning in review. It also makes accounting a better job, and a better pitch to the students the profession says it wants.
So, what will you do with the hours you get back? Your answer determines whether AI makes your firm better or just busier.
