There’s a running joke among accounting providers about “automated” bookkeeping services: that they’re a great source of cleanup projects.
With the recent unforeseen closure, and mere days-later acquisition, of Bench.co, I’m reminded of the continued misperceptions about where, exactly, software and AI are presently in the accounting space, and how much necessity remains for a human touch.
For a bit of background, Bench.co was released to the public in 2013 as a combination software platform and bookkeeping provider. Geared toward small businesses, it offered only very straightforward, cash-basis accounting. (Options like payables and receivables were not available, nor sales tax filing, etc.) Eventually, year-end tax filing services were added as well.
The company boasted $100M in VC funding, and positioned itself very much as a software platform first. However, customers expressed mixed reviews of the final product they received. Per customer reviews on TrustPilot, though the software seemed to function as designed, employees rotated frequently and service was inconsistent. However, it still came as a shock when the platform was abruptly taken offline on December 27th, 2024, and clients were emailed that services would be ending immediately. (Clients were told that their data would be accessible until March of 2025, and were advised to plan on filing extensions for their 2024 taxes.) It appeared that Bench’s roughly 600 employees would be left jobless shortly after Christmas.
A mere 3 days later, it was announced that Employer.com would be acquiring Bench, ensuring continued services for their customers and employment for their staff. Of course, the breakneck pace of these announcements have done little to assuage client concerns about the accuracy and availability of their data for the upcoming tax season, and many are looking to alternatives.
However, I will always advise caution in entrusting business financials to, what are, primarily, software companies. In my experience, software-focused “automated” bookkeeping rarely produces a valuable final product. As I mentioned in my first sentence, new clients often come to us for help in cleaning up the messes these providers have created. Though I have never been “behind the scenes” at somewhere like Bench or Quickbooks Live, I have seen the books they prepare. In short, the issue often comes down to a continued undervaluing of the importance of accounting knowledge.
It is easy to train an AI or clerical-level employee on the data entry of recording cash-basis expenses, or in reporting deposits to an income line. And, for a time, financials can even appear accurate, as many business owners typically focus on their Profit & Loss Statement. It is only when errors start to snowball that problems become obvious.
I have seen books with 7-figures worth of “uncleared” transactions dating back years, resulting in massively over-stated expenses which in turn triggered an IRS audit. On the other side, I have seen revenue duplicated when invoice payments and their matching deposits were recorded as two separate transactions (again, sometimes going back for an extended period). Balance Sheets are often a mess, and any transactions which do not affect cash are lost entirely. (Something that becomes relevant when, for example, a business owner misses a deductible expense for the theft of an equipment asset.)
I believe strongly that software and AI are powerful tools and I enjoy using those tools to make my work more streamlined, and I look forward to what future technological developments will bring. But the current trend to try to remove the accounting professional from the equation is, to me, similar to entrusting a 7-year-old to operate a jackhammer because “the machine does all the work”.
In the future, I am sure that more and more accounting tasks will be successfully delegated to AI-driven programs. However, I do feel there will always be a need for a trained individual, with a strong understanding of fundamental accounting principles, to manage the process.