Earmark CPE

Earn CPE Anytime, Anywhere

AI

Not All AI Is Created Equal and Your Next Software Decision Depends on Knowing the Difference

· ·

When Jeff Seibert ran consumer product at Twitter, he asked the finance team for his budget to throw a team event. They said they’d get back to him in 45 days. So he just ran the event without them.

That gap between real-time data and 30-to-90-day delayed financial reports was frustrating, and it eventually led Jeff to build Digits, a new general ledger designed from scratch for the machine learning age. After raising $100 million pre-launch, testing 2,000 monthly closes, and getting 80% of clients closed in under an hour, Digits launched in March 2025. Now, just over a year later, hundreds of accounting firms are onboarding thousands of clients onto the platform.

Jeff launched Twitter’s algorithmic timeline in 2016, and it was one of the first global deployments of machine learning. Now, the AI revolution Jeff helped launch is flooding the accounting profession with claims that are hard to verify. Every accounting software company seems to include AI in its marketing copy, promising everything from “fully automated bookkeeping” to capabilities that don’t add up under scrutiny.

In a recent Earmark webinar, host Blake Oliver and Rob Hamilton, Head of GTM at Digits, pulled back the curtain on how AI in accounting actually works. He was joined by Megan Reid, Product Specialist & Firm Enablement at Digits, who fielded questions throughout the session.

Every AI claim in accounting software isn’t real. But accountants who understand the four core model types (plus one common lie) will make smarter investments, automate the right parts of their workflow, and position their firms for a shift Rob sees coming by the end of 2026.

The AI hype problem (and one question to cut through it)

Before making any technology decision, you need a filter for separating real capabilities from marketing fluff. Rob offered a simple one that cuts through the noise.

He showed screenshots from multiple accounting software companies making bold AI claims. One promised “fully automated bookkeeping.” Another asked, “Do you do AI bookkeeping or do you use a dedicated team of experts?” The positioning has gotten so confusing that firms can’t tell what’s real anymore.

The confusion isn’t new. About five years ago, tech investor Naval Ravikant tweeted, “In most pitch decks, AI stands for Anonymous Indians.” For a long time, that was literally true. Services like Botkeeper rose and fell using offshore labor dressed up as automation. Today, “AI actually means we just bolted on and sent all of your data to ChatGPT,” Rob explained.

Here’s your filter: “AI is the same thing as machine learning,” Rob stated. “If someone is talking to you about AI and they’re not referring to machine learning as the underlying premise, it’s just BS.”

But this filter only works if you understand what machine learning actually is.

Traditional software is straightforward. You write code that tells the computer exactly what to do. It’s tedious to build, but rock solid once it works. Machine learning flips this completely. You feed the system thousands or millions of examples, and the model learns the patterns itself. As Jeff explained in a clip Rob played, “You give the computer the goal state—I want this outcome—and then the computer itself is learning how to do it.”

These models are neural networks. Thousands of hidden layers mimic how neurons connect, based on Google’s 2017 “transformer” research paper (the “T” in GPT). It’s a massive matrix multiplication problem where the system figures out how variables relate to each other.

But machine learning isn’t one thing. Different model types have different strengths and uses in accounting. Understanding these distinctions helps you avoid buying the wrong software and shows you exactly where AI can save time and where vendors are overselling.

The model types that matter (and one that doesn’t)

Rob walked through five categories that get lumped under “AI,” but understanding the differences is what separates informed decisions from expensive mistakes.

Generative models

Large language models (LLMs) are the ones you hear about most, ChatGPT being the prime example. GPT stands for “Generative Pre-Trained Transformer,” and these models generate the most likely continuation of whatever prompt you give them. Rob showed a useful application: turning bullet-point close notes into polished client emails. His advice is to write a “job description” for the AI once. Tell it who it is, give context, specify output format, add examples. Then just paste in different client notes as needed.

But generative models have serious limits. They’re “super eager” and always want to complete prompts, making them prone to hallucinations, or making things up that sound real. They’re bad at math because they generate text rather than calculate numbers. And they’re trained on the internet, not your specific clients. “The ways that it is hallucinating is stuff that maybe even is hard for humans to catch sometimes,” Rob warned. Always review the output.

Agents

These are LLMs with help. You give them a job description, a task, and tools, like computer programs they can use to generate reports, list accounts, or run calculations. The agent makes a plan, uses its tools, checks if the task is done, and loops until complete. Rob showed Digits’ agent answering “I want to hire 20 software engineers next year. Can I afford to?” with a data-backed response.

Guardrails are critical. Microsoft’s early agent “started asking people on dates in the chat,” Rob noted. “You don’t want your accounting agent dispensing dating advice.” Agents work well for updating schedules, running quality checks, and answering analytical questions, but they’re slow and need careful boundaries.

Predictive models

These got Rob visibly excited, and for good reason. These models take an input and predict an output from known options. When the model sees a $5 Starbucks charge, it considers the client’s location, history, and chart of accounts. For a local client, it’s meals and entertainment. For a New York client in California, it’s travel. A $157 Starbucks charge is probably an event, regardless.

What makes predictive models perfect for transaction categorization is they can’t hallucinate; they only choose from existing options. They’re deterministic (same input, same output), include confidence scores, and run fast and cheap once trained.

Digits built a “layer cake” of predictive models:

  1. Client-level (learns each business)
  2. Firm-level (encodes your best practices)
  3. Global (trained on 180 million transactions worth nearly $1 trillion)
  4. An LLM fallback for completely new transactions

The result was over 97% accuracy, compared to standalone LLMs that plateau below 80%, which is about the same as outsourced bookkeepers.

Document extraction models

These combine OCR with layout-aware language models that understand document structure. Previous tools used Amazon’s Mechanical Turk, which relied on humans manually extracting data and took hours. Modern extraction models work in seconds. Digits’ bank reconciliation automatically pulls PDF statements, matches transactions to the exact spot in the PDF, and generates audit reports.

Data analysis

This model is where Rob pulled the rug out. Financial reporting and analysis is “actually just math. It’s not ML.” Computers have done statistical analysis for decades. Could you build an agent to do it? Sure, but it would be slow, expensive, and probably wrong. “If anyone says their AI does reporting and statistical analysis, please ask them what they’re talking about.”

Here’s how the right models map to your month-end close:

  • Book transactions: Predictive models (with LLM fallback)
  • Reconcile statements: Extraction models plus matching algorithms
  • Update schedules: Agents
  • Review and correct: Agents with quality checklists
  • Analyze and report: Statistical analysis plus agents for questions

“Shoehorning an LLM in to solve a problem and just sending a bunch of information is fundamentally incorrect,” Rob emphasized. Each step needs the right model. No single AI approach handles everything.

The 2026 prediction

Understanding model types is just the foundation. The urgency comes from how fast everything is converging.

“Across a large client set in different industry types, it’s highly likely that the month-end close process is looking to be completely automated by the end of 2026,” Rob predicts. Even his “95% automated” hedge probably sounds aggressive. But his logic follows directly from the technology.

If predictive models hit 97% accuracy on transactions, extraction models automate reconciliation in seconds, agents handle schedules and quality control, and statistical analysis covers reporting, then manual work drops to a fraction. Rob’s goal is to see accountants doing “1/20th of the work you’re doing today.”

He acknowledged limits. Construction firms with complex job costing might not hit that threshold. But for firms serving professional services, cash-basis businesses, and straightforward accrual clients, the automation curve is steep.

An AI-native firm will focus on value instead of tedium. Deeper industry expertise. Stronger client relationships. Higher margins. You’re not reviewing every transaction, you’re supervising the system and handling exceptions. Those hours saved on reconciliation become advisory time clients actually value.

But this is also a competitive necessity. “AI won’t replace you. Someone who’s good at using AI is going to,” Rob said, quoting a common warning. And he was direct about the stakes. Firms that don’t adapt will face “a cascading effect on business models” as early adopters pull ahead.

For overwhelmed firms, which Rob acknowledged includes most firms, he offered practical starting points:

  • Map your processes first. If you use workflow tools like Karbon or Keeper, you’ve probably documented your steps. If not, start there. You can’t identify where AI fits until you know what you’re actually doing.
  • Start small and low-stakes. Don’t tackle your biggest challenge first. Try drafting emails, testing categorization, or visualizing data. Build your intuition gradually.
  • Get hands-on with new tools. Rob mentioned being impressed by Claude Opus, which could build HTML dashboards from his data (something he couldn’t do as a non-engineer). The specific tool doesn’t matter; hands-on experience builds judgment.
  • Know your business before choosing where to start. As Rob put it, “You need to know the details of your business to know where you can start and where the right places to poke and prod are.”

The “wait and see” window is closing. Firms that develop AI literacy now by asking questions about models, data handling, and use cases, will be ready for the rest of 2026 and beyond.

Your next move: Better questions, smaller steps, faster action

Let’s turn Rob and Megan’s insights into actionable takeaways:

  • Not all AI is equal. Four real model types plus one fake (data analysis) get lumped together. When vendors pitch “AI-powered reporting,” you now know to dig deeper.
  • Each close step needs a different model. Predictive for transactions. Extraction for reconciliation. Agents for schedules. Statistics for reporting. Anyone claiming one solution does everything deserves scrutiny.
  • Predictive models beat LLMs for categorization. Layered architectures that learn your clients and firm patterns dramatically outperform chatbots. Bigger isn’t always better.
  • Ask vendors the hard questions. What model type? Where does data go? Are you training your own models or sending financial data to third parties? This is due diligence.
  • The tipping point is closer than you think. Whether Rob’s 2026 prediction proves exactly right or directionally right, the trajectory is clear. Understanding these distinctions now positions you to take action.

For those interested in going deeper, Rob mentioned resources like the AI Native Accounting Foundation and the AI-Native Accounting podcast hosted by Kacee Johnson, where industry leaders discuss the latest developments.

The accounting profession is at a real inflection point. Smart firm leaders will develop the literacy to ask smart questions, experiment in the right places, and redirect time from tedium to advisory work clients value.

Rob noted this might be one of the few professions with such clear AI use cases, putting accountants at the forefront of innovation. That’s an opportunity to shape how technology serves the profession, not the other way around.

Watch the on-demand webinar for complete details, including live demonstrations, security architecture specifics, and audience Q&A covering nonprofits, inventory clients, and platform migrations. The future of accounting is being written now. Make sure you’re part of the conversation.

AI Agents Now Complete Tax Returns Start to Finish While the Government Can’t Even Audit Its Own Books

· ·

The US government just declared itself insolvent. AI agents are completing tax returns without human intervention. And the accounting profession is caught between these two massive disruptions.

In Episode 481 of The Accounting Podcast, hosts Blake Oliver and David Leary opened with a bombshell that somehow flew under the mainstream media radar. The Treasury Department’s own financial statements show the US is $42 trillion in the red, and that’s before counting Social Security and Medicare obligations. They then dove into an equally seismic shift with guest Kenji Kuramoto, founder of Acuity and newly appointed Managing Partner in Residence at AI company Basis, exploring how artificial intelligence is transforming every corner of the accounting world.

Deficit Spending Just Keeps Going

“It’s official. We are insolvent,” David announced at the start of the episode, referencing the Treasury’s 2024 financial statements. They show $6 trillion in total assets against nearly $48 trillion in total liabilities. That $42 trillion hole doesn’t even include the $88 trillion in unfunded Social Security and Medicare obligations sitting off the balance sheet.

“Imagine a family making $52,000 that owes $1.3 million in a line of credit,” Blake said, putting the crisis in household terms.

Making matters worse, the Government Accountability Office issued a disclaimer of opinion for the 29th consecutive year, essentially saying it can’t even verify the accuracy of the numbers because the Department of Defense has never passed an audit.

“This is the reason a huge number of people voted for Trump,” David said. “They wanted to stop deficit spending, and it just keeps going.”

Meanwhile, AI Is Eating the Accounting Profession

While the government’s books are falling apart, AI companies are racing to automate the work of keeping everyone else’s books together. TaxGPT announced an AI agent capable of completing 1040 returns from start to finish without a preparer touching a keyboard or mouse. The tool works with existing web-based portals and tax prep software, pulling in W-2s, 1099s, and other source documents, then having a review agent double-check everything.

“Why go after tax pros?” David asked. “Just get in bed with the portal companies and go after TurboTax.”

Kenji, who recently joined Basis after selling Acuity and taking a year off, described watching AI agents handle complex accounting work that made him come out of retirement. “I saw an agent handle complex payroll entries like booking the GL entry, creating an accrual because the pay period didn’t align with month-end, posting the reversing entry for the following month, and building a complete set of work papers,” he said. “I saw this last year, and I was like, wait, what?”

The flood of AI announcements kept coming throughout the episode:

  • Ramp launched an accounting agent that auto-codes transactions down to the line-item level on invoices, claiming to save finance teams 40+ hours per month
  • Xero announced a multi-year partnership with Anthropic to integrate Claude AI directly into its platform
  • Canopy launched a bookkeeping module with AI that continuously reviews books and flags issues in real time
  • Double (formerly Keeper) released AI Journal Entries that can handle complex, repetitive entries from source documents
  • BILL announced agents for invoice coding, W-9 collection, and automated vendor payment responses

“Everyone thought we were boring,” Kenji said. “Look at this. All these Y Combinator companies spinning up and fundraising announcements and agents everywhere. Come on. Exciting.”

The Skills Gap Is Already Here

The shift is showing up in real time in hiring data. In 2023, only 18% of accounting job postings mentioned AI skills. Now it’s 30%, a 67% increase.

“The real-world requirement is probably 50%,” David argued. “People are behind on updating their postings.”

But a better question is what happens to the business model. Kenji described how at Acuity, the bottleneck was always people. Plenty of companies needed help with their books, but you couldn’t hire enough accountants to serve them cost-effectively. AI agents break that constraint. One highly efficient bookkeeper might handle 45 to 60 clients today. “Will one person eventually be able to handle 200 clients?” David asked.

The threat isn’t just from other firms. An article on Payments.com found that everyday taxpayers are already using ChatGPT and Gemini to do their taxes before ever talking to a professional. The reason is “speed and simplicity,” David explained. “AI can explain tax concepts, organize the documents, and suggest deductions. These are things they’re not getting from their tax professional.”

Are Tokens the New Billable Hour?

As AI cuts the time needed to complete work, firms are scrambling to figure out how to price their services. Bloomberg Law reported that PwC, KPMG, and RSM are all exploring alternatives to hourly billing.

“This may be the thing that finally gets us there,” Kenji said about moving away from billable hours. “If I just used AI to help me get my work done and I’m cutting down my billable hours, I’m losing revenue.”

“You can bill for tokens,” David suggested, offering a provocative alternative.

He then vented about Earmark’s own token consumption across multiple platforms, including Claude, GitHub Copilot, Retool, ChatGPT, and more. “Two days ago, an automation stopped working,” he said. “We spent five plus people hours trying to increase our tokens and get the automation working again.”

The problem is, token costs are opaque and growing. David introduced two terms gaining traction: “token anxiety,” or not knowing what you’re being charged for, and “AI FinOps,” managing AI costs across platforms.

“There’s an opportunity here for firms to become a token expert and offer it as a service,” David suggested.

Blake’s take was more pragmatic. “It’s better than timesheets, that’s for sure.”

The Window Is Closing

The government that sets the rules can’t even audit its own books while declaring itself insolvent. Meanwhile, AI agents are automating core accounting work at a pace that makes the shift from paper to computers look gradual.

“These agents are actually now becoming a component of our workforce,” Kenji said. “You’ve got accountants and you’ve got agents. This is the future state we’re moving into.”

For practitioners, it’s clear that the tools to dramatically expand your capacity exist right now. But so does the threat of clients going straight to AI and bypassing your firm entirely. The window to adapt is open, but it won’t stay that way for long.

As Blake noted about current AI pricing, “When Uber was new, everything was really, really cheap.” The subsidies won’t last forever. To thrive, firms need to figure out the new economics now, whether that’s value pricing, token billing, or something else entirely. Those that don’t may find themselves as obsolete as the government’s ability to balance its own books.

Listen to the full episode for the complete discussion, including deeper dives into specific AI capabilities and Kenji’s firsthand perspective from inside an AI-native company.

Fake Auditor Conclusions, Fabricated Board Minutes, and the Growing Cracks in Accounting’s Trust Infrastructure

· ·

A compliance startup allegedly sold hundreds of companies fake SOC 2 reports complete with made-up auditor conclusions and board meeting notes that never happened. In Florida, legislators nearly abolished the state’s Board of Accountancy entirely. And AI companies now run ads that sound exactly like QuickBooks marketing copy.

These are just some of the topics Blake Oliver and David Leary tackled in their latest episode of The Accounting Podcast. The hosts dug into stories that show the systems meant to ensure trust in accounting face threats from multiple angles.

The (Alleged) SOC 2 Scandal

“This is wild,” Blake said, thanking a listener for sending him a detailed investigation about Delve, a VC-backed compliance startup. The company allegedly created fake SOC 2 reports at scale, using what the hosts described as a disturbing playbook.

According to a Substack series Blake reviewed, Delve’s platform pre-populated everything from policies to evidence and even independent auditor conclusions. The company then allegedly routed these pre-written reports through audit firms that simply rubber-stamped them.

“These are allegations. I have not independently verified any of this,” Blake was careful to note. “This is a very in-depth Substack report by an anonymous poster. So we should take this with a grain of salt.”

But the details were alarming. The investigation claimed to find board meetings that never happened, security simulations that were never performed, and trust pages showing controls as “implemented” before any actual work was done. Companies had written policies claiming they had mobile device management, VPNs, and intrusion detection systems, even though they had none of these.

The author analyzed 322 public Delve trust pages and found that 321 showed the exact same SOC 2 control set, which seems odd for supposedly customized compliance programs.

“The logo you get is an AICPA logo, right? You’re getting a stamp of approval from the AICPA,” David said, cutting to the heart of the problem. “Is the AICPA checking on all these badges that are on company websites?”

Blake explained how easy it would be to game the system. “All you have to do is find a firm willing to sign off without actually doing the work,” he said, comparing it to the BF Borgers case in Colorado, where a CPA firm was caught signing off on audits it never performed.

“This is the problem with assurance,” Blake continued. “If you have a few bad actors willing to just sign off, sign off, sign off, they can make a lot of money. And how do they get caught? And if they get caught, what happens?”

Florida Almost Killed Its Board of Accountancy

While fake compliance reports threaten the profession from within, Florida’s legislature almost destroyed a key piece of regulatory infrastructure from the outside.

House Bill 607 would have eliminated the Florida Board of Accountancy along with other professional licensing boards as part of a sweeping deregulation push. The Florida Institute of CPAs called it “the most serious threat to the profession in decades.”

“How do you regulate the CPA in Florida?” Blake asked, explaining the stakes. Without a Board of Accountancy, there’s no enforcement mechanism, no oversight, and no one to investigate bad actors.

The bill moved quickly through two committees before being stopped. But victory came at a cost. To focus on defeating the bill, FICPA had to table its own effort to create alternative pathways to CPA licensure that would have allowed candidates to qualify with 120 credit hours instead of 150.

The irony wasn’t lost on the hosts. Florida was the first state to implement the 150-hour rule. Now, while about 30 states have approved alternative pathways, efforts to defend against total deregulation have sidelined reforms.

“We want to streamline licensure, but we don’t want it to go away,” Blake said. “We’ve got folks who want too much regulation, and then we’ve got folks who want no regulation. There’s got to be a middle ground here.”

David predicted this won’t be the last such attempt. “I imagine we’re probably going to see more pushes for this because people are going to want the big, huge AI companies to have their AI do CPA work without a license in the way.”

When AI Ads Look Like QuickBooks Ads

Speaking of AI companies, David discovered something unsettling through a targeted LinkedIn ad. Anthropic is marketing “Claude for Finance” using language that sounds exactly like traditional accounting software.

The ad promised to handle recurring financial workflows, organize receipts into clean spreadsheets, build quarterly revenue models, and cross-reference documents for month-end close.

“Third-party app developers and accountants and CPAs that use Claude essentially trained the model so they could just take everybody out of the middle,” David explained. He compared it to how the iPhone camera evolved. At first, you needed third-party apps for filters and editing. Now it’s all built in.

The hosts also discussed a Wall Street Journal article about how regular people are already using AI for tax work. Examples ranged from using Copilot to model Roth conversions to having AI explain confusing IRS notices. One person used Gemini to value charitable donations for their tax return.

“The takeaway is they’re avoiding getting an accountant or a tax professional,” David said bluntly.

But the technology isn’t perfect. One user found that Grok gave wrong answers about capital gains tax until he rephrased his question. A retired tax preparer tested ChatGPT on an IRS volunteer certification exam, and ChatGPT failed.

This leads to what David called the “fact check tax,” a term from Anthropic’s own survey. “An assistant that sounds sure but is often wrong forces you to treat everything as suspect. Instead of freeing attention, AI creates a permanent fact-check tax.”

The Bigger Picture

These stories paint a picture of a profession under pressure from multiple directions. Fake compliance reports undermine the attestation model. Deregulation efforts threaten the licensing framework. AI platforms are positioning themselves as replacements rather than tools.

As Blake noted about AI, “It’s going to be really hard for Intuit and Xero to keep up unless they’re just plugging into ChatGPT or into Claude. How can their own AI chatbots keep up with what these companies are doing, and how fast they’re developing?”

For accounting professionals, these are challenges that require attention and action. Listen to the full episode of The Accounting Podcast to hear Blake and David discuss these stories and more. 

The Privacy Excuse for Not Using AI in Accounting Just Lost Its Last Leg

· ·

Blake Oliver needed to file a City of Los Angeles business tax return for his last remaining bookkeeping client. Instead of spending 30 minutes clicking through websites and copying numbers, he gave Claude Cowork a single instruction: “Search my email for info about the account and help me file it on the city website.”

What happened next, documented on a recent episode of The Accounting Podcast, shows exactly where the accounting profession stands with AI adoption. The AI agent searched Blake’s email, found the tax notice, extracted the business details from a PDF, logged into the city website, navigated to Xero to pull gross receipts, filled out the return, and drafted the client confirmation email. Total human involvement: one correction when it pulled accrual instead of cash basis numbers.

“This is a task that might take 15 to 30 minutes if you filled out a time sheet. Claude just did it,” Blake told co-host David Leary during their weekly news roundup.

The Numbers Show AI Closing In Fast

OpenAI didn’t just release another model update with ChatGPT 5.4. It specifically targeted the kind of work that fills an accountant’s day. As David read from OpenAI’s announcement, the company “put a particular focus on improving GPT 5.4’s ability to create and edit spreadsheets, presentations, and documents.”

The benchmarks back up that focus. Using something called GDPval—which measures performance on real-world knowledge tasks across 44 occupations—ChatGPT 5.4 now beats or ties industry professionals 83% of the time. On spreadsheet tasks specifically, it jumped from 68% accuracy to 87% in a single generation.

“It’s getting close to that 90% success now on everything,” David observed. For context, that means if you give an accountant and this AI the same spreadsheet task, the AI will match or beat the human’s performance nearly nine times out of ten.

Real Accountants, Real Work, Zero Software Costs

While Blake was experimenting with Claude for business tax returns, a developer went further. Fed up with TurboTax, he used Claude to complete a 42-page federal return plus two trust returns, all at zero software cost beyond his AI subscription.

His approach was surprisingly low-tech: Downloaded blank PDFs from the IRS, have Claude fill them out, then print and mail. The biggest challenge wasn’t getting the AI to do the calculations or understand the tax code. It was trying to make it work with the IRS’s online fillable forms. So he skipped that part entirely.

“The comments were like, ‘Can I quit doing my returns tomorrow? I’ve been waiting for this my whole life,'” David said, describing the reaction from tax professionals who saw the developer’s work on social media.

The timing is notable. These experiments happened during tax season, when practitioners are supposedly too busy to explore new tools. Yet here’s a developer replacing TurboTax with Claude, and Blake casually using an AI agent for client work.

The Privacy Excuse Just Disappeared

Most firms claim they can’t use AI because client data is too sensitive. This week, Zapier offered a solution to the privacy problem.

Its new AI Guardrails can detect over 30 categories of personally identifiable information, redact sensitive data before it reaches AI systems, block workflows when it detects problems, and identify attempts to manipulate the AI. You insert it as a step in any workflow, and it sanitizes the data before AI ever sees it.

“If you have client data being passed through Zapier into any AI tool, go add this step to your workflows,” David advised listeners.

Blake was even more direct about the implications. “I totally see this being a huge tool for accounting firms, because we have all this information we want to use with AI. But a lot of it is too sensitive. That’s the main reason most firms aren’t doing anything with it.”

Beyond AI: The Week’s Other Bombshells

While AI dominated the discussion, Blake and David covered several other major stories that accounting professionals need to know about, including:

The Botkeeper Collapse Gets Messier

In an interview with Accounting Today, CEO Enrico Palmerino claimed the company went from healthy to dead in eight days. But Blake uncovered how Botkeeper engineered its financials by selling their bookkeeping clients to a firm called Benchmark Cloud Accounting. It then had that firm buy a multi-million dollar Botkeeper license. “That is how you turn service revenue into SaaS recurring revenue,” Blake explained.

Iran’s Drone Economics

The cost disparity between Iran’s drones and America’s million Interceptor missiles raises questions about the financial sustainability of current military strategies. “We’re spending $3 million to shoot down something that costs $20,000 to $50,000,” Blake pointed out.

KPMG and Polymarket

Anonymous accounts on the prediction market Polymarket have been suspiciously successful at betting on earnings for companies audited by KPMG- (and only KPMG) audited companies. The amounts are small so far, but as David noted, “Are they doing it on the real derivative markets as well?”

Record 401(k) Withdrawals

Vanguard reports hardship withdrawals have tripled since 2020, jumping from under 2% to 6% of participants. Despite positive business sentiment, individual financial stress is climbing.

What This Means for Your Firm

General-purpose AI agents can complete multi-step workflows across email, accounting systems, and government websites. The privacy barriers that kept firms on the sidelines now have concrete, deployable solutions. The capability exists. The safety tools are live. The only question is timing.

Blake’s Claude experience offers a preview of the emerging division of labor. AI handles the execution, humans provide the judgment. The AI pulled the wrong basis for the numbers. Blake caught it. That’s where professional value lives now, not in the clicking and copying, but in knowing what the AI doesn’t know to check.

The message might seem poorly timed to practitioners overwhelmed by tax season. But accountants are eager for tools that eliminate the drudgery, even in the thick of deadline pressure.

Listen to the full episode to hear Blake walk through his Claude workflow step by step, get David’s take on what ChatGPT 5.4’s benchmarks really mean, and understand why the Botkeeper story matters for anyone considering AI-powered bookkeeping solutions. The episode reflects a profession at an inflection point—not in some distant future, but this week.

A Private Equity Insider Explains What Happens After Your Firm Gets Acquired

· ·

Devin Mathews has a 14-year-old dog that had never been sedated for a dental cleaning—not once in 14 and a half years. Then a private equity firm bought his veterinary office. Suddenly, the dog needed his teeth cleaned twice a year, at $1,000 a pop.

“I never knew my dog needed so many dental services,” Devin tells Blake Oliver on the latest episode of the Earmark Podcast. “It’s the upsell opportunity.”

This small anecdote captures the anxiety spreading through the accounting profession as private equity floods in. What really changes when PE takes over? Who benefits? Who gets hurt? And what about artificial intelligence? Is it going to make all these PE investments obsolete?

Devin brings an insider’s perspective with an outsider’s freedom to speak plainly. As a partner at ParkerGale Capital with 30 years in private equity, he knows the playbook. But since his firm invests in software companies, not accounting firms, he can share what really happens without affecting any deals. He regularly fields calls from employees at freshly acquired firms trying to figure out what just happened to their careers.

The Satisfaction Gap: Partners vs. Everyone Else

Blake starts with data that sets the tone for the entire conversation. An Accounting Today survey found that over two-thirds of partners at PE-backed accounting firms say they’re satisfied with their experience. But only about 15% of non-partner employees feel the same way.

“Let me get this straight. The partners who got paid in the transaction are ecstatic because they have terminal value,” Devin says. “And the rank and file who probably didn’t even know the business was for sale, their lives have completely changed, and expectations have gone through the roof.”

That’s the core issue. When PE acquires an accounting firm, the capital is there to buy out the existing owners, not fund operations. Partners cash out. Staff wake up to a Zoom call announcing new ownership and dramatically different expectations.

Most employees don’t realize how much analysis has already happened before that announcement. “Some 26-year-old in New York has run all that math,” Devin explains, “and they literally know you and your business better than you know it.” PE firms have sorted every employee by bill rate and utilization. They’ve hired McKinsey or Bain to benchmark everything against industry standards. They know exactly where they can push harder.

The first target is the person who gets paid a lot but doesn’t bill many hours. “This guy has been around for a long time,” Devin describes. “Maybe they’re great at business development, but they’re just not billing the hours anymore.”

How PE Actually Makes Money in Accounting

The mechanics are straightforward: bill more hours, raise bill rates, get more efficient, and make acquisitions. When revenue equals people times hours times rates, those are your main levers.

But Devin acknowledges the challenge with professional services. “The assets walk out the door every night.” Push too hard, and those assets can walk across the street to one of the 44,600 CPA firms in North America that aren’t owned by private equity.

Blake raises a critical point based on his own experience as a manager at a top-25 firm. The traditional path of working your way up, becoming a partner, and receiving ownership and profit distributions disappears under PE ownership. Instead, you get RSUs or phantom equity that only pay out if there’s an exit event.

“You drive home after that speech,” Devin imagines, “and you say to your spouse, ‘Remember that path I had to be a partner? That’s gone. Now I only get paid if there’s an exit.'”

Good PE firms try to address this through transparency and communication. Devin describes his ideal post-acquisition speech: acknowledge the surprise, address the fear directly, and promise that resources will match the higher expectations. “Expectations of ten out of ten, resources of ten out of ten. That’s a great combination.”

But the reality hits later on. “I walk you through the PowerPoint and it sounds really great,” Devin says. “Then six months later you’d be like, ‘Where’s Devin? What happened?'” The speech is easy. Delivering on it is where most PE firms struggle.

AI Changes Everything, But Not How You Think

The conversation takes an unexpected turn when they start talking about artificial intelligence. AI is threatening the entire economic model PE investments depend on.

Devin’s firm is all-in on AI. Every Friday is “DIY Friday,” and everyone spends two hours experimenting with AI tools, trying to replicate workflows, and testing what works. They pay for all the major models. They’re true believers.

But the results aren’t quite what they’re hoping for. “A lot of times it takes me twice as long to review and audit what the AI built than it would have taken me to build it on my own,” Devin admits. The tools hallucinate. It’s “wildly confident about things it knows nothing about.”

The key insight is that AI is a “ceiling raiser, not a floor raiser.” It makes experienced professionals amazing. It makes entry-level people only slightly better because of domain knowledge. An experienced accountant can spot what the AI got wrong and fix it. An entry-level accountant doesn’t know enough to catch the mistakes.

“An entry-level developer, like an entry-level accountant, doesn’t have enough domain knowledge or experience to see what the AI did wrong,” Devin explains.

“It’s a great time to be mid-level or experienced. It’s a bad time to be entry-level,” Blake observes, noting the cruel paradox. The routine tasks that used to train new accountants, like sampling, confirmations, and rolling forward work papers, are being automated first.

The legal profession offers a preview. Harvey, an AI platform for law firms, raised about half a billion dollars. It claims to work at the level of a fifth-year Harvard Law associate. Every major firm supposedly uses it. “But Kirkland and Ellis isn’t charging me any less than they used to,” Devin notes. The efficiency gains aren’t reaching clients. Firms capture them as profit.

Your Options Are Better Than They Appear

So what should accounting professionals actually do? Devin has specific advice.

First, if you’re at a PE-backed firm, demand transparency. “Show me Bain Capital’s value creation plan,” he suggests. “How did they underwrite this, and how are they going to get there?” Understanding the plan helps you align your work and see your compensation trajectory. If they won’t share it, that tells you something, too.

Devin identifies three problems that plague most firms before PE arrives. There’s no clear plan (or too many plans), no communication about why leaders make decisions, and nobody understands how compensation works. Good PE ownership can fix these issues. Bad PE ownership makes them worse.

The good news is, PE has bought only about 400 of the 45,000 CPA firms in the US. “There are 44,600 CPA firms in North America that aren’t owned by private equity,” Devin points out. “And you can leave.”

Devin has advice for early-career accountants facing pressure from private equity and AI automation: Start your own firm.

“Get really good with the AI. Way better than the 35-year-old or the 45-year-old. And start your own firm. Be an AI-first accounting firm, and you own all of it.”

It’s never been easier, he argues. You can set up an LLC on LegalZoom. You can reach clients directly through LinkedIn or YouTube. The barriers that kept young professionals locked into traditional firm hierarchies are crumbling.

“It’s pretty obvious most adults have no idea what they’re doing, and they’re mostly full of BS,” Devin says with characteristic bluntness. “So don’t wait for your turn. Go get it.”

The Bottom Line

The traditional accounting career ladder is being dismantled. PE is removing the path to partnership, and AI is removing the entry-level work that trains new accountants.

But professionals who understand what’s actually happening have more options than they might think. There are still 44,600 independent firms. Starting your own practice has never been easier. And if you’re staying put, you can at least demand to see the plan and understand where you fit.

As Devin puts it, “Why do you need to wait in line and have some private equity firm tell you how you get to run your business? Go find some other people who believe in it the way you do, and go build the firm in your image.”

For the full conversation, including Devin’s stories about his own podcast journey and more details on how PE firms evaluate acquisition targets, listen to the complete episode of the Earmark Podcast.