“The target is $1 million in client fees per accountant,” Blake Oliver says, opening Episode 119 of the Earmark Podcast. Then he asks: if software runs the close and accountants handle the review, that efficiency is worth serious money. So who gets it? The firm, the client or you?
His guest, Jessica McKellar, studied computer science at MIT, co-founded Pilot as CTO and now runs it as CEO. Pilot is a bookkeeping, tax prep and CFO advisory firm. It closed nine years of books, roughly 187,000 month-end closes, before it started selling its Meridian software to other firms. Pilot sponsored the episode.
Pilot uses autonomous closes in pursuit of an internal goal of $1 million in client revenue per accountant. Jessica expects if production work can be done in software, it often will be. The opportunity left is judgment, review and client relationships, a shift software engineers have already lived through. On the episode, they discuss how the autonomous close works, Blake’s pushback, and the engineering parallel.
Taking the Accountant Out of the Iron Man Suit
The revenue goal depends on one shift. Software has to stop helping accountants close the books and start running the close itself, all the way through to review.
Pilot didn’t start there. In 2016, Jessica’s co-founder clicked around QuickBooks Online while she hovered over his shoulder, writing the first version of the software against the QuickBooks Online APIs. The first-generation tools were “the Iron Man suit for accountants.” They were effective, but the experience involved constant “context switching or handoffs between a person and the software.” With that kind of workflow augmentation, she says, moving firm-level metrics can be “surprisingly difficult.”
Over the last 18 months or so, foundation models advanced to the point where “you can actually take the accountant out of the Iron Man suit” and put an AI agent in it. As long as the data inputs are in place, she says, the close runs on its own through to review. Once firms get comfortable, the close kicks off at the start of the month and “you wake up to completed books.”
For Jessica, the payoff is predictability. “It reduces the work, but it reduces the variance.” That means fewer capacity mismatches and fewer Sundays spent covering for a sick colleague.
Blake’s Skepticism, Tested
In Blake’s view, blockers, accuracy and the audit trail stand in the way.
“I’ve always been skeptical of the claims of fully autonomous month-end close,” he says. In his experience, missing client information comes up in almost every task.
Jessica separates true blockers, like no access to the primary checking account, from judgment calls. Her example is a client paying Gusto for the first time when Meridian has no payroll access. The agent parks the activity “somewhere reasonable” and flags it for the reviewer. It can’t do the granular breakout until it gets payroll access or someone uploads a payroll journal. The reviewer then decides whether the issue should hold up the close.
On accuracy, Blake is blunt: “90-something percent is often not enough. It needs to be 100%.” Jessica says the math isn’t left to the AI. “There’s like deterministic software that is doing the math of rolling forward the supporting schedules.” The agent verifies everything ties out, and it flags a broken recon. A full audit log shows why each transaction was categorized, along with monthly cuts of the recons and schedules.
There are still rough edges. Xero doesn’t expose bank reconciliation through its API, so bank feed matching has to happen in a browser.
Software Engineers Already Lived This
Jessica’s engineering background gives her prediction weight. “This is the biggest change to software engineering as a profession probably ever,” she says. The old way of “artisanally crafting and reviewing and deploying code” is going away quickly, and many engineers are “in mourning” for skills they spent careers building.
What replaces it? “It’s not about producing the code anymore. It’s about having good taste, good judgment, knowing what to build.” Some engineers find that exciting. For others, it’s “a source of deep existential stress.”
She draws the accounting parallel without softening it. “If that work can be done in software, it’s often going to be.” The judgment, strategy and client work left behind is a real opportunity, but “it’s not going to be for everyone.”
Blake describes the split he already sees. People with judgment and experience become dramatically more productive. He puts himself at “10X.” People without that judgment risk having their jobs automated away. He also points out that this seems to be the first technological revolution to really hit white-collar workers, after earlier ones automated blue-collar jobs.
Who Actually Gets Paid?
Jessica believes “accountants are going to start making more than software engineers in some cases.” Her claim comes with a condition, though, “if some of that efficiency is getting accrued back from a compensation perspective.” She notes that the margin firms reclaim can go to the client experience, to employees or simply back to the firm.
Blake sees the clearest path in the firm of one. “If I can generate a million in revenue and not have to hire anyone,” that solo practitioner would out-earn nearly every engineer outside companies like OpenAI, Anthropic and Meta.
What lasts is human work. Jessica points to “the part that’s being a therapist for your clients,” and to accountability. Clients want to talk to a human when something goes wrong. Work that requires a license or credential stays too.
Blake adds an observation of his own. Accounting ranks near the top of automatability charts, yet new-grad hiring is “totally flat” while other professions are down 10% to 20%. He wonders whether that reflects slack from the talent shortage or something the analyses miss. Jessica’s answer is the durable relationship between business owners and an expert they can rely on.
The $1 Million Accountant Is a Role, Not a Gift
Software may make the $1 million number possible. It doesn’t hand it to anyone. Getting there means moving toward review, judgment and client relationships as more production work shifts to software.
“It’s never been a better time, to my mind, to be an accountant,” Jessica says, but also, “I’m not exactly sure what’s going to be happening for us all five years from now.”
Firm owners and staff need to consider how they’ll share the efficiency gains. Listen to the full episode for Jessica’s walkthrough of the autonomous close and her conversation with Blake about AI’s effect on the profession.
