KPMG cuts 10% of audit partners as AI still fails 1 in 5 accounting tasks

Attention: This is a machine-generated transcript. As such, there may be spelling, grammar, and accuracy errors throughout. Thank you for your understanding! Blake Oliver: [00:00:04] Kpmg and EY in the UK have started quietly demoting equity partners, which is a big change because getting to partner at a big four used to mean a job for life. Well, at least until mandatory retirement. And now they are demoting these partners from equity to salaried partners. David Leary: [00:00:24] Coming to you weekly from the OnPay Recording Studio. Blake Oliver: [00:00:31] Hello and welcome back to the Accounting Podcast, your weekly roundup of news in the profession. I'm Blake Oliver. David Leary: [00:00:36] I'm David Leary. Blake Oliver: [00:00:38] David who are our sponsors this week? David Leary: [00:00:40] Sponsors. This week we have cloud accountants, staffing, earmark on pay and fishbowl. Blake Oliver: [00:00:46] Are you tired of the endless search for qualified accounting talent? You're not alone. Growing accounting firms are struggling to find available and affordable team members when they need the most. Cloud accounting staffing has the solution with their revolutionary candidate portal. Unlike traditional staffing agencies that waste your time with sales calls, paperwork, and deposits, the Cloud Accountant Staffing Candidate Portal gives you instant access to highly vetted, qualified accounting professionals. No waiting, no hassle. Just top talent right now. What makes this different? Speed and simplicity. While other firms make you wait weeks or months with cloud account and staffing, you could interview someone as soon as tomorrow. Their boutique support ensures you're getting quality talent that's both available and affordable. Exactly what growing firms need. The candidate portal puts you in control. Browse live candidates, make selections on your timeline, and build your offshore team without the traditional headaches. To find, review and book interviews with potential team members, all in less than ten minutes, head over to The Accounting Podcast dot com slash The Accounting Podcast dot promo forward slash CPE. A s David, so much news to cover. This week, I want to talk about the least glamorous side of accounting that still can't be automated. We've got news about KPMG laying off 10% of audit partners in the US, and the best AI still fails at a fifth of accounting tasks. But I want to know first what's top of mind for you? David Leary: [00:02:06] Um, I think four stories that kind of tie together. So decimal is getting out of the, uh, Cass business and they're just going to have franchise, they're going to sell franchises of their tech. Um, collective is buying open ledger collective is one of these accounting firms with tech. H&r block wants to be more into the bookkeeping space and then, uh, fractional HR provider and, uh, called Austin Alliance. They want to get in the bookkeeping space. So there's a lot of jockeying and like, who's going to provide bookkeeping services or how they're going to deliver bookkeeping services. Blake Oliver: [00:02:38] Well, tell me about this decimal story. Um, I've heard of them before, but like, remind me who decimal is. David Leary: [00:02:44] Yeah. So decimal is an accounting firm with engineers. You know, I've kind of labeled these firms as this. They raised money. Um, they may have had multiple raises, but I think the latest I saw was like in 2022, they raised money. They actually. Do you remember KPMG spark? Blake Oliver: [00:02:58] Yes I do. David Leary: [00:02:59] That was that was wired. That from KPMG. And rolled those clients in. Blake Oliver: [00:03:03] Okay. Got it. David Leary: [00:03:04] Right. Um but they're going to pivot now. So so they raised money. They built decimal to provide accounting services. But what's the struggle with that? You've always said, right, you can't be a SaaS. You can't have SaaS valuations and raise money when you're a human service business. And we've seen the struggles, right? We've seen the struggles with Visor Tax bench scale factor, arguably bot keeper, even though they didn't really go directly to clients. Right. We've seen this over and over again. I'm probably forgetting a couple that have come and gone in De Niro, right? This model is really hard to do. Um, so what decimal is going to do now? They're going to pivot away from that. And they as of March 2026, they sold their services arm. So they're now not going to do any more client facing work. And they're just going to franchise out their tech, their workflows, their infrastructure to firm owners who basically want to have an independent firm but get that technology stack. Now, this is based. And so your franchise model is going to give you AI tools. It's going to give you automation Soc2 compliance, full operational system, sales, pricing, onboarding, QA. But it's very similar to pilot pilot, maybe about four months ago, maybe longer, six months ago, rolled out local partners. And that's kind of the same thing, even though it's not really a true franchise model. But if you if you're a small bookkeeping practice, cash practice, you can, uh, join pilots, local partners and just get all their tech. So instead of having. Right. So it's kind of, they're seeing this model happen more now. And so is the old model of like the bench model, the scale factor model. Now that decimals pivoted and feels like pilots pivoting is this over this, like we're just going to be an accounting firm and we're going to automate everything because we have lots of engineers. Is officially dead. Blake Oliver: [00:04:53] I don't know if it's dead yet, but there's still a gap. There's a gap between what you can automate and what you can't. And I have a story here from cfo.com that illustrates that gap and explains, I think might explain why these tech companies are moving to this model. So, David, the model you are talking about is AI powered human humans finish the rest of the work. Like that's why they're doing this because they, they can't do it themselves. Or they could, they could do it themselves, but they, they would then be a services business. And that's not what a tech company, you know, wants to be. They can't get the valuations they need. Right. Yeah. David Leary: [00:05:32] Because I think before it was we're going to get engineers and code our way to efficiency. Now there's new players like kick.ai who basically took over the bench client. I think who their bet is, we're going to use AI to bridge that gap. Blake Oliver: [00:05:45] Right. David Leary: [00:05:45] But it's the same model but different tech. Blake Oliver: [00:05:48] Yeah. And the problem is that the AI is just not there yet. And we know this from a study, a benchmark that a company called Dual Entry put together. And what they did is they evaluated 19 AI models across 101 real accounting workflows. And the top performer is Claude Opus 4.7. That makes a lot of sense. That's the one that is really hot right now. Everyone's switching over to Claude from ChatGPT because it's so good, but it still only achieves 79% accuracy on real world accounting tasks. So it's still failing at 1 in 5 of these tasks. And that's a problem because one out of five is not acceptable. Those errors cascade through financial statements. And if you add. David Leary: [00:06:38] An employee and ten hours of the week, right, approximately ten hours of the week, 8 to 10 hours of the week was just wrong. You'd probably freak out if it's a human employee, right? Blake Oliver: [00:06:49] Well, and then, you know, you got to clean up the mess, right? So it the efficiency is actually even lower. It's not like you're automated 80% of the work because you got to go clean up that other 20%. That's messed up. So examples of the, the stuff that AI was tested on include transaction classification, journal entry creation, bank recs, month end close work. And the other AIS are ranked in here. Right. Claude Opus 4.7 did the best 79%. Gpt 5.4 from open AI is slightly behind 77%. Gpt four, for comparison, scored about 40% on the same task set. So it's gotten a lot better, but we're still pretty far from 100% right. Going from 80% to 100%, I think is going to take a while. And so that's why in accounting, you just haven't seen AI taking over yet. Um, there's there's just too many mistakes for us to rely on it without safeguards. So, you know, I think that's like, it's fascinating. Um, and I've got another story here about one of those tasks that is definitely not going to get automated by AI anytime soon. And that's thanks to the Controllers Council. They published a story called The Least Glamorous Side of Accounting that still can't be automated. Any any guesses as to what that might be? Uh, never, never. Not going to be automated. It's it's not glamorous. David Leary: [00:08:28] Not glamorous. Can't be automated. Oh. Blake Oliver: [00:08:32] All right. I'm gonna I'm gonna give it to you. David Leary: [00:08:34] I just don't have I don't, I just don't have it. Blake Oliver: [00:08:36] Well, it's because you weren't an auditor, David. And, uh. David Leary: [00:08:39] That's true. Blake Oliver: [00:08:40] Physical inventory. Observation. David Leary: [00:08:43] Yes. Inventory retail at the mall. So I'm familiar. Blake Oliver: [00:08:46] Okay. So you know, you did you ever participate in an inventory count? David Leary: [00:08:50] Yes. I've had where, you know, the auditor came in and helped count. Yeah. I've done this and seen it. Yeah. Blake Oliver: [00:08:58] So regulations require the inventory accounts, you know, are done by people in person. And so there's some great examples in this story. Um, climbing into 90 foot grain bins to make sure the grain is in there counting rocks in snake infested quarries, spending all night manually counting frozen vegetables after barcode scanners fail in the cold. Here's a quote one auditor said after visiting a pig farm. It didn't m