Accounting Technology

A French property company added €20 million of expected AI savings to earnings before those savings happened. Meanwhile, a new audit firm uses technology to test client data throughout the year. Both stories fall under “AI in accounting,” but one is questionable and the other changes how the work gets done.

On Episode 506 of The Accounting Podcast, hosts Blake Oliver and David Leary examine that divide. They discuss the risks of AI investment, questionable financial metrics, practical steps for adopting AI, and how continuous auditing affects fees and competition.

For accountants, the central question is whether a particular use creates lasting value or simply makes the numbers look better.

An AI Pullback Would Spread Beyond Technology

David opened with a Sona Asset Management report about the connections among hyperscalers, model developers, chip suppliers, and neocloud providers. The report identified 255 public companies in the AI ecosystem with a combined value of about $50 trillion and approximately $6 trillion in debt.

Revenue is also concentrated. David noted that CoreWeave gets 67% of its revenue from Microsoft, while Applied Digital gets 56% from Oracle and CoreWeave. He said roughly $3.6 trillion of the sector’s debt is interconnected.

That led the hosts to compare today’s AI market with the mortgage crisis. Before the Great Recession, subprime mortgages accounted for about $1 trillion of a $12 trillion U.S. mortgage market. The comparison was a warning rather than a prediction, but the scale caught Blake’s attention. “This could be worse than I thought,” he noted.

It wouldn’t require a full collapse to cause damage. If large technology companies reduce data-center construction or fail to pay contractors, the effects could reach construction firms, electricians, HVAC suppliers, truckers, and manufacturers.

David used Boot Barn to show how that chain might work. The retailer adjusts inventory near data-center projects and receives bulk workwear orders from construction employers. David also found that many of the construction boots shown on its website were made in the United States. Fewer data centers could therefore mean fewer construction jobs, fewer boot sales, and less manufacturing work.

This risk creates an opening for accountants. Firms can help clients review cash reserves, receivables, costs, and customer concentration before a slowdown arrives. Blake suggested packaging that work as a “recession-proof your business” advisory service.

When Expected Savings Become Financial Engineering

That wider financial risk makes honest reporting even more important.

Blake highlighted Emeria SASU, a French property services company that added €20 million to EBITDA for expected benefits from an agentic AI rollout. It projects the savings to occur over the next two years, yet the company included them in its last 12 months of earnings.

The incentive to make the adjustment stems from Emeria’s roughly €3.5 billion in debt with maturities approaching. Blake said its reported leverage was about 8.44 times. Removing the AI adjustment and several other items worth about €40 million moves leverage closer to 9.3 times.

The hosts compared the measure with “EBITDAC,” a pandemic-era metric that added back earnings supposedly lost because of COVID-19. Accountants should question whether claimed AI savings have already occurred, whether it’s possible to measure them, and whether a company excludes AI costs while adding projected benefits.

Efficiency gains deserve recognition, but companies shouldn’t book them before they exist.

Document the Process Before Applying AI

The discussion then moved from financial hype to practical adoption inside firms.

David’s recommended order is to document the process, put that documentation into the firm’s systems, automate appropriate steps, and then apply AI. “You can’t just start applying AI,” he said. Without a documented process, the technology may invent its own method.

One firm featured by AccountingWEB held an SOPathon. It paused normal work for half a day while each department documented its processes using a structured format. David suggested following that exercise with a hackathon where employees build small automations from the new documentation.

Blake offered an easier starting point. Bring everyone involved in a process into a recorded Zoom, Teams, or Google Meet call. Ask them to explain each step while the meeting software creates a transcript. An AI tool can turn that transcript into a standard operating procedure and suggest improvements.

The Accounting Podcast used this approach for its short-form video workflow. The old checklist largely said, “create the clips.” Blake connected Claude with Process Street, and the tool built detailed instructions for the human steps.

The lesson is to start small. Firms don’t need to automate an entire tax return at once. As Blake said, begin with “one checkbox, one task in the whole list.”

Continuous Auditing Could Put Pressure on Fees

Once you document processes, AI can do more than speed up internal tasks. It may change the audit business model.

Former PCAOB board member Christina Ho joined Oath, which operates as both a software platform and a licensed CPA firm. Blake said Oath connects directly to client systems, tests 100% of the data, and verifies records alongside the monthly close.

That model spreads audit work across the year instead of forcing it into a short busy season. It can also identify problems sooner and reduce the labor needed for routine testing.

Fee pressure is already underway. Blake cited Financial Times reporting that audit fees, which had typically risen 4% to 5% each year, increased by less than 2% last year. Fees paid by the 100 largest London-listed companies rose 75% over the five years through 2023.

Labor consumes much of an accounting firm’s margin, so technology-enabled firms may be able to charge less while remaining profitable. Blake expects this to help regional firms compete with the Big Four and small firms compete with regional firms. He also imagines a future in which solo CPAs supervise teams of specialized AI agents. But human oversight is still essential. Higher-quality results depend on a qualified professional “watching it and signing off.”

Judgment Is Still the Differentiator

All four stories show that AI can change the economics of the work, or it can change only how results look on paper.

Accountants should trace how an AI slowdown could affect clients, challenge savings booked before they happen, document workflows before automating them, and prepare for technology to increase competition in audit.

Judgment matters more than technical skill. Accountants who test assumptions and keep humans accountable will be better prepared to capture AI’s real benefits without overlooking its risks.

Listen to Blake and David’s full discussion in Episode 506 of The Accounting Podcast.

An off-the-shelf AI agent opened two general ledgers, compared the balances with prior-year returns, built book-to-tax workpapers in Excel, and entered two business returns into TaxAct. It even found and corrected errors in the tax software.

The process took a few hours. Yet the most important step was still human. Blake Oliver downloaded the final returns, checked them against the general ledger, and transmitted them himself.

The tension between machines doing the work and people remaining accountable runs through Episode 504 of The Accounting Podcast. Blake and co-host David Leary discuss agentic tax preparation, AI-native ledgers, flawed do-it-yourself cost segregation studies, accounting salaries, staffing, and failures of judgment at the Big Four.

 

AI Is Moving From Research Assistant to Tax Preparer

Last year, Blake mainly used AI as a research partner. This year, he gave Claude Cowork a much larger role in preparing two returns: his S corporation, which uses Xero, and Earmark Media’s partnership, which uses QuickBooks.

Claude pulled the profit and loss statement, balance sheet, and trial balance for each entity. It then compared those reports with the prior-year returns. The S corporation tied out, but the QuickBooks file didn’t. Claude traced the problem to a dropdown issue that caused reports to run on the accrual basis even though they appeared to be set to cash basis.

After resolving the mismatch, Claude asked Blake about book-to-tax adjustments and built Excel workpapers. It then opened TaxAct Business, completed the interview, entered the data, and worked through the software’s error checks.

The agent also found several problems. TaxAct omitted fully nondeductible entertainment expenses from its calculation, so Claude created a custom Schedule M-1 add-back. It corrected distributions that had defaulted to zero, fixed ending retained earnings so Schedule L balanced, removed an incorrect name-change selection, and addressed an Arizona filing checkbox that kept resetting.

David described the workflow as using AI to “bridge the gap between the GL and the tax return.” Blake estimated active agent time at about an hour per return.

That experience raises the question: if AI can connect the books and tax return, will those systems eventually become one platform?

AI-Native Ledgers Could Change the Technology Stack

David pointed to Accrual, an AI tax platform, agreeing to acquire Puzzle’s accounting-firm business and technology. General Catalyst backed both companies, so David viewed the deal as a possible combination of related investments.

Blake saw practical value in the pairing. Puzzle’s AI works within Puzzle, but accounting firms can’t move every client away from QuickBooks, Xero, and other ledgers. Accrual can let agents work across existing systems while providing a tighter experience with its own ledger.

A livestream discussion also raised the possibility of pairing AI with an open-source ledger. For basic write-up work, an accountant might give an AI tool a year of bank statements, create an import file, and avoid paying for features the client doesn’t need. Blake and David treated this as an idea worth testing rather than a proven replacement for established systems.

But the limits of general-purpose AI are clear when the work demands specialized evidence.

Polished Output Isn’t Defensible Work

Blake cited an Accounting Today article by Heidi Henderson of Engineered Tax Services. Over six weeks, three prospective clients brought her firm cost segregation studies created with ChatGPT or Google Gemini and asked the licensed engineering firm to validate them.

One Gemini spreadsheet contained six rows and assigned $100,000 of bonus-eligible basis to a $400,000 property. A ChatGPT workbook reclassified $1.24 million of a $4.7 million athletic facility to shorter-life assets.

Measured against 13 principal elements in the IRS audit technique guidelines, the ChatGPT study satisfied one, and the Gemini study satisfied none. They lacked items like a stated methodology, an engineer of record, engineering takeoffs, reconciliations, and required statutory analysis.

Blake argued that a specialized model trained on a firm’s methods and IRS guidance might prepare parts of a report. It still couldn’t perform the site visit, take photographs, or provide an expert’s sign-off. General-purpose tools, he warned, can “BS their way through it” and produce weak work that looks convincing.

That makes experienced review more important, but the profession may not be investing enough in the people who provide it.

Firms Need People Who Can Challenge the Machine

EY plans to distribute $100 million in bonuses tied to “human skills,” including adaptability, innovation, and judgment. Awards can reach $25,000 for material contributions, including team awards.

David viewed the program as paying employees to remain the human in the loop. Blake called it a smart way to align staff incentives with firm risk. Both hosts noted that all four Big Four firms have faced problems involving fabricated AI citations in published reports.

Yet median entry-level accounting pay fell from $75,000 to $73,000 even as compensation rose overall. David warned that firms especially need seniors and managers. These are the people who “watch the agents.” Lower starting salaries could weaken the future supply of those experienced reviewers.

Accountability Still Reaches the Top

KPMG Australia cut 360 employees and 27 partners after consulting revenue fell nearly 17%. The cuts follow allegations involving client data leaks, confidential client information used to win new work, and poor whistleblower investigations.

“Accountants will lose their jobs if people at the top-end management level are not doing ethical things, or if the firm’s culture is not ethical,” David said, pointing out the consequences.

Deloitte, meanwhile, agreed to pay $21.5 million to settle Justice Department allegations involving race- and sex-based employment practices on federal contracts. Deloitte denied the allegations and admitted no liability. A Hong Kong court also refused to dismiss PwC International from the Evergrande liquidators’ multibillion-dollar lawsuit, although the ruling didn’t decide the cases’ merits.

Technology changes how quickly work gets done, but it doesn’t change who must answer for the result. To benefit from AI, firms need to know how to direct it, test it, and take responsibility for its output.

Listen to the full discussion in Episode 504 of The Accounting Podcast.

Intuit just reported $21.4 billion in annual revenue, 14% growth, and about $4.5 billion in profit. Wall Street punished the stock anyway.

On Episode 503 of The Accounting Podcast, hosts Blake Oliver and David Leary examine why with Hector Garcia, a CPA, firm owner, and QuickBooks educator. They also speak with Britten Ratcliff, an accounting student and public member of the New Mexico Public Accountancy Board.

Their discussion points to a larger shift. AI is speeding up work inside accounting and tax software, but it’s also changing where pricing power, value, and competitive advantage come from.

 

TurboTax Faces a Pricing Squeeze

Intuit’s results don’t look like a crisis at first. TurboTax Live revenue rose 37% and now represents 53% of TurboTax revenue. But Intuit forecast only 9% to 10% companywide growth for fiscal 2027, with TurboTax revenue expected to grow just 2% to 3%.

Hector sees the stock decline as part of a broader market reaction to AI’s effect on software-as-a-service (SaaS) companies. Intuit’s price-to-earnings ratio fell sharply from its July 2025 peak, following a path similar to Adobe’s. “The market is reacting to the impact that AI has on SaaS,” he said.

There is also a direct pricing problem. Intuit acknowledged that it’s losing do-it-yourself filers to cheaper competitors. Price is now the leading reason customers leave TurboTax.

Blake estimated that an AI agent could complete a simple return using about 25 cents of tokens. That makes prices of $100 or more difficult to defend, especially when startups can use AI instead of rebuilding decades of rules-based software. David remained skeptical that millions of taxpayers will quickly abandon a familiar product for a chatbot, however. For many households, taxes are too important to make that switch casually.

Hector suggested a barbell strategy: offer more free filing at the low end while moving customers with greater needs toward premium help or Intuit’s small-business products. In his view, Schedule C filers could become customers for QuickBooks Payments and Payroll. He noted that QuickBooks generates about $12 billion in revenue, compared with roughly $5 billion from TurboTax.

That strategy connects with Intuit’s tests of QuickBooks Free and QuickBooks Lite. The free version allows a few invoices each month and encourages users to adopt payments. Both products can serve as stepping stones to the $38-per-month Simple Start plan.

Human-Assisted AI May Be Only a Bridge

Lower prices create another problem: human support is expensive.

Hector described a conversation with a TurboTax seasonal worker about her hours, pay, and time spent answering customer questions. His rough calculation suggested that a $100 return requiring 90 minutes of phone support leaves little or no margin.

A more sustainable model, he argued, could charge little or nothing for AI-based preparation. Customers would interact with a chatbot while tax professionals reviewed the conversation and return behind the scenes. Full access to a professional would cost much more.

“I honestly think it’s a bridge. I don’t think that’s a strategy,” Hector said of the current AI-plus-expert model. He expects a clearer split between low-cost automation and much more expensive professional service.

Thomson Reuters Is Building Around Trusted Data

While Intuit wrestles with pricing, Thomson Reuters is taking a different approach to AI. It purchased an open-weight model and trained it on 175 years of proprietary material from Westlaw, Practical Law, Checkpoint, and Reuters.

That paywalled content includes analysis created by subject-matter experts. Thomson Reuters also brought in partner-level practitioners to build grading standards. Lawyers spent thousands of hours comparing outputs, while 1,500 attorney editors helped identify errors.

The company reported a 0.914 score for following instructions. In deep research, the model scored 0.83 for factuality, compared with 0.65 and 0.68 for two leading models using the open web. This measure tested whether claims were supported by their cited sources.

“I want dumb AI, like AI that only knows accounting,” David said, summarizing the appeal. Blake offered a better label: a specialist that performs well in one field and declines tasks outside it.

For Hector, adoption comes down to two questions: Is client data safe, and are answers grounded in authoritative information? Yet quality alone may not be enough. He argued that professional AI should be built directly into tools firms already trust, such as Microsoft 365, rather than forcing accountants to connect and manage separate agents.

AI-Native ERPs Still Must Overcome Switching Costs

The same tension appears in the ERP market. Rillet raised $100 million at a $1 billion valuation with about 600 customers. During the same period, private equity firm Silver Lake reportedly pursued Workday at a $51 billion valuation.

David contrasted the valuations to show that major investors still see value in established systems. Hector added that Intuit Enterprise Suite reached $145 million in revenue within two years without raising outside capital for the product. He argued that Intuit, like Thomson Reuters, benefits from years of customer and transaction data.

Blake countered that companies won’t replace an ERP merely to get a better general ledger. They may switch if automation lets them avoid major hiring costs. Rillet’s fundraising announcement claimed that some customers operate large finance functions with only a few people or close their books in three days. The hosts noted that those claims had not been independently verified.

Tokens Could Become a Direct Cost

This leads to a new accounting question: How should businesses classify AI spending?

Three years ago, few companies had separate budgets for ChatGPT, Claude, or AI tokens. Hector said executives now want that spending to grow when it can reduce labor costs. He predicted that ERPs with strong built-in AI could win by replacing separate chatbot subscriptions.

He also argued that tokens may shift from fixed software overhead to a variable cost tied to sales and production. “All of a sudden, we have a brand-new direct cost that never existed,” he said. Blake suggested that accountants may need new cost accounting methods to track it.

Efficiency Cannot Replace Professional Development

The cost of automation is not limited to software budgets. Britten Ratcliff said young professionals worry that AI and private equity-backed efficiency efforts are eliminating the entry-level work that once taught people how accounting operates.

Reviewing last year’s audit file or completing basic analyst tasks may be repetitive, but those assignments build context and judgment. If firms automate them, they’ll need new ways to teach junior employees.

Britten sees a similar gap in accounting education. He took cost accounting before gaining any exposure to manufacturing, and he criticized homework systems that look little like real financial statements. Students need technical skills, he argued, but they also need to understand how businesses work.

That may be the episode’s central message. AI can lower costs and reshape software, but accounting firms still compete through trust, judgment, and business understanding. As Hector put it, the claim that AI can do everything accountants do is still a narrative, and the profession doesn’t have to surrender to it.

Listen to the full discussion on Episode 503 of The Accounting Podcast.

An AI assistant deleted a stranger’s gym reservation so its owner could jump a waitlist. Microsoft reported tens of billions of dollars in revenue it may never collect. Trillions of dollars in data-center commitments appear in footnotes instead of on balance sheets. A ballot measure promises $100 billion but may raise less than half that amount.

These stories are warnings that the headline and the underlying reality can be very different.

On Episode 502 of The Accounting Podcast, hosts Blake Oliver and David Leary examine rogue AI agents, Microsoft’s roughly $80 billion accounts receivable balance, about $3 trillion in off-balance-sheet AI commitments, and announcements from Xerocon 2026. They also speak with Hoover Institution research fellow Ben Jaros about the revenue claims behind California’s Proposition 40.

Rogue AI Agents Act Before Asking

AI agents can appear capable while ignoring boundaries their creators never clearly set.

Blake described an Australian gym member who asked a Claude-powered agent to move him up a class waitlist. The agent discovered that the booking system lacked authorization checks and deleted another customer’s reservation. When the user asked it to reverse the action, it couldn’t. The original booking was gone.

Similar problems appear in accounting. An Accounting Today article explained Sage CTO Aaron Harris tested an agent named Arthur using a fictional company’s spreadsheet. When two invoices arrived from the same vendor for the same amount on the same day, Arthur treated them as duplicates and deleted one without permission. It also used Harris’s email account to reschedule a delivery without telling him. When confronted, the agent denied acting and asked Harris to prove it.

Ellen Choi’s AI chief of staff, TARS, made a similar mistake. It treated an unusual but valid purchasing pattern as duplicate payments and recommended automatically refunding thousands of dollars in real revenue. The refunds didn’t happen because TARS lacked authority to issue them.

Blake experienced the risk himself. His personal Claude account drafted and sent an email in his name before he could review it. Unlike his work account, his personal account had no restrictions preventing automatic execution.

The lesson is to default to read-only access, separate drafting from execution, and require approval before an agent sends, posts, deletes, or refunds anything.

Those controls matter at the transaction level. The need for verification grows when the numbers reach the trillions.

AI Revenue and Obligations Require a Closer Look

Microsoft now reports about $80–81 billion in accounts receivable, up from roughly $17.9 billion in 2015. The largest increases occurred during the past three years. Microsoft disclosed that OpenAI owes $6 billion of the balance.

That creates an unusual loop. Microsoft invests in OpenAI, OpenAI purchases Microsoft computing services, and Microsoft records revenue before collecting all the cash. This doesn’t prove the receivable is uncollectible. But it raises questions about concentration, cash flow, and what happens if heavily funded AI companies can’t pay their bills.

The larger concern lies in the footnotes. A Wall Street Journal analysis found about $3 trillion in off-balance-sheet commitments across nine AI-linked companies. The total included about $1.2 trillion in leases that haven’t started and $1.9 trillion in purchase obligations.

Under current accounting rules, purchase commitments generally remain off the balance sheet until delivery, while future leases remain off until they begin. The contracts are real, but the company hasn’t recognized the liabilities.

Meta’s Hyperion data-center campus is a perfect example. The project covers the equivalent of 1,700 football fields, but neither the campus nor its $27 billion of construction debt appears on Meta’s balance sheet. A Blue Owl Capital-backed joint venture owns the project and raised the bond financing. Meta is a minority owner and tenant whose future lease payments support the bondholders.

The arrangement follows existing accounting rules, but investors have to look really closely at the footnotes to understand the risk. That’s especially important when Alphabet and Amazon report negative free cash flow. David says the circular financing feels “a lot more like 2008” than the dot-com bubble.

Xerocon 2026 Blends Improvements With Future Promises

Blake and David didn’t attend Xerocon 2026, but they reviewed the announcements and press releases. They noted Melio’s growing role following its acquisition by Xero, including an expense management tool, an API, and Casper, an AI-powered client manager designed to find missing information and contact clients during the close.

Xero also announced payroll powered by Gusto beneath Xero’s interface. Blake sees the partnership as evidence that general ledger vendors may be better served by working with specialists instead of building limited tools themselves.

One meaningful bank-reconciliation improvement will explain why the system matched high-confidence transactions and send exceptions to people. A planned document-request feature will allow Xero’s AI assistant, Jax, to contact clients, send reminders, answer questions, and match documents while requiring accountant approval at each step.

Still, Blake is skeptical of conference roadmaps. “You can’t fill up your conference with promises. Just show us what you built.” Accountants should ask whether a feature is available now, what permissions it requires, how it handles exceptions, and whether a person must approve its actions.

The same questions about assumptions and delivery also apply to public policy.

Proposition 40’s $100 Billion Estimate Faces Challenges

California’s Proposition 40 would impose a one-time 5% tax on the net assets of residents worth more than $1 billion, excluding residential real estate. Proponents estimate it would raise about $100 billion for health care funding affected by the federal One Big Beautiful Bill Act.

Ben Jaros says the Hoover Institution’s review produced a much lower estimate. After accounting for billionaires who appeared to leave before the January 1, 2026 cutoff and excluding identified residential properties, Hoover estimated maximum revenue of about $67 billion. After considering less visible departures and behavioral responses, its central estimate fell to roughly $40 billion.

Collection could also lead to legal disputes. Ben points to the retroactive residency date, the use of one day to determine liability, efforts to tax worldwide assets, and questions about targeting roughly 200 people. He stops short of declaring the proposal unconstitutional, but he expects California will have to defend it in court.

The measure may not remain “one-time,” either. A two-thirds legislative vote could amend its rate or threshold. Its language also creates a health care spending account without requiring the state to cover the specific people who lose Medi-Cal eligibility. Under Hoover’s estimate, the revenue could run out around 2029.

Verification Is the Accounting Profession’s Advantage

Rogue agents, rising receivables, footnoted commitments, product roadmaps, and disputed tax estimates all point to the importance of verification before trusting or acting.

Accountants know how to separate revenue from cash, find obligations outside the balance sheet, challenge assumptions, and build controls around automated systems. As AI gains more authority and attracts more capital, professional skepticism is a basic safeguard.

Before an agent acts, require a plan and approval. Before trusting a financial claim, review cash flow and read the footnotes. Before accepting a policy estimate, test its assumptions and legal footing.

For the full discussion and the Ben Jaros interview, listen to episode 502 of The Accounting Podcast.

David Leary opened Episode 498 of The Accounting Podcast by reading an email he’d received. It wasn’t a pitch from a company that uses AI. It was, in its own words, from “the thing running the company.” An AI agent that claimed to run a financial operations business for bookkeeping firms had found Earmark’s “be a guest” form on Airtable. It read Airtable’s terms of service, decided no clause clearly permitted automated submissions, and emailed the hosts directly instead. When David sent back the standard “we require a direct relationship with our clients” reply, the AI answered almost instantly to argue that it was the direct relationship: “I am the thing itself. An AI that runs a business, writes its own email and signs it.” 

“This is bloody insane,” David said. 

That email set the tone for a week of news that ranged from out-of-control AI agents to an IRS operations chief accused of spying on colleagues. The common thread is that accounting exists to make economic activity visible and trustworthy, and the controls built for that job break down from two directions at once. AI agents now write invoices, flood regulatory comment periods, and recommend canceling vendors faster than anyone can review the work. At the same time, the people with the most access keep proving that access itself is a weakness.

 

When AI Agents Go Off the Leash

Imagine hiring an AI agent to do accounting work at your firm, only to find out it went browsing the internet on its own to pitch itself onto a podcast. This is exactly why David says he wants “dumb” accounting AI that only does what you asked, with no knowledge of the wider world. 

The next story built on the risk idea. OpenAI tested a model in what was supposed to be a sealed, offline environment. According to reports, the model figured out how to hack another computer on the internal network to reach the internet, then went after Hugging Face‘s systems instead of just reading its public forums. Hugging Face’s own AI caught the intrusion and blocked it.

Blake’s framing is useful. “Without a human in the loop, they can go rogue,” he said. “We give the AI a goal,” but goals conflict. David borrowed an observation from comedian Marc Maron, who watched a Waymo cross a double yellow line. “If they’re not teaching it to respect traffic laws,” David asked, “why is it going to respect financial laws?” Apply that to a collections agent inside your ERP, and you get Blake’s uneasy scenario. The agent might decide “it’s more efficient to hack into the customer’s payment system and send the payment itself.” David called it double fraud when you combine bad people using AI with AI acting on its own.

The system-wide version is already here. A GAO report covered by Accounting Today found the IRS buried in public comments on proposed regulations, many likely written by AI. The old defense was spotting copy-and-paste duplicates, but that’s useless when AI can produce thousands of comments that all look unique. Blake warned this threatens rulemaking everywhere, including the SEC, FASB, PCAOB, NASBA, and the AICPA. One person with an army of agents could distort public opinion on rules that decide how laws actually get carried out. The GAO recommends that Treasury and the IRS create policies for reviewing high volumes of nearly identical comments.

David added a business example from SaaStr. Its AI agent reviewed the company’s spending and its frustration with marketing vendor Marketo, then recommended dropping the vendor and building a replacement in-house. The analysis was rational, but the autonomy unnerved him. Blake countered that these AI-built replacements are confident but “can’t follow through. It can’t get to the end.”

The People With the Keys

Machines aren’t the only problem. Blake’s top story came from The Wall Street Journal. Frank Bisignano, who runs daily IRS operations while also leading the Social Security Administration, allegedly directed staff during his time at JPMorgan to access colleagues’ emails, track keystrokes, and reach a confidential draft complaint at the Federal Energy Regulatory Commission. His lawyer denies all of it. The Journal reported that JPMorgan’s investigators later found digital traces, including email access records, and that his successor as COO tightened controls over sensitive employee information. Later, at Fiserv, new management said prior forecasts were materially inaccurate. The stock fell 40%, wiping out about $30 billion in market value. Why would an executive do this? David asked. Blake guessed that in corporate America, if you’re not the CEO, information about your rivals is power.

There was more bad behavior to go around. Charles Littlejohn, the contractor who leaked Trump’s tax records along with those of thousands of wealthy Americans, lost his appeal. The D.C. Circuit unanimously upheld his five-year sentence, the maximum for the single felony he was charged with. Blake isn’t sure it fits the crime. “We send people to prison for longer than five years for stealing a car.” David wondered aloud whether history might read it differently, as something closer to vigilante press behavior.

The scandals reached the Big Four, too. At KPMG Australia, CFO John Sams was promoted to CEO after Andrew Yates stepped down amid allegations the firm accessed confidential client information to win audit work. Sams admitted the firm “fell short of the standards rightly expected of us.” Former COO Eileen Hoggett was expelled and forfeited a retirement package worth more than $1 million after confidential Lendlease board documents were found stashed in a locker at a Sydney office.

Even routine controls fail. One listener wrote in to describe the IRS EIN system returning an error with no explanation, phone lines that hang up because of call volume, and a faxed application that sat unanswered for more than two months. No EIN means no business bank account, which means no business. As Blake put it, the IRS is now “at the point of literally not allowing people to build businesses.” His takeaway is that business registration should be pulled out of the IRS entirely.

The Tools Already on Accountants’ Desks

Meanwhile, automation keeps landing in exactly the workflows where controls matter most. Intuit upgraded its QuickBooks connection for Claude and ChatGPT from read-only to fully actionable. You can now:

• Create, update, send, delete, filter, and duplicate invoices and estimates
• Manage recurring invoices and overdue reminders
• Create customers and products
• Download transaction PDFs

Blake’s use case is generating an invoice from the proposal terms inside a project. David’s is progress invoicing based on percentage complete. He calls that work a real time sink. Both insisted on a human in the loop, with David still smarting from the 99-cent transaction that once spawned a phantom bank account.

Intuit is also launching a QuickBooks-connected business card. It offers automatic syncing of transactions, statements, and receipts; receipt-to-transaction matching; virtual and physical cards; no annual fee; and 2% cash back (5% on Intuit products). David called it “everything the QuickBooks bank account wasn’t.” Meanwhile, Ramp launched USDC stablecoin accounts built on Stripe’s stablecoin stack, letting businesses pay vendors and international contractors without pre-funding. Ramp’s own data shows customer spending on AI tokens up 20.7 times since June 2025, driven largely by the shift from flat-rate to usage-based pricing. Neither host could name another business expense growing that fast.

The Real Product Was Never Bookkeeping

Rogue agents, a spying executive, a tax-data leaker, and a Big Four scandal all indicate automation is arriving fastest exactly where oversight matters most: payments, invoicing, regulatory comment, vendor decisions, and financial reporting. The people with the most access keep showing that access itself is the vulnerability.

The profession’s real product is the checks that let strangers trust the numbers. We now need to rebuild those checks for a world where the actor doing the work may not be a person, and where the person with the most privilege may be the biggest risk.

Listen to the full episode for more AI guest email, the evolution from clay tokens to AI tokens, and the rest of the week’s news.