It’s the third of the month. Bank statements have posted. Payroll exports and loan schedules are ready, and the close checklist is about to consume another week of staff time.
Now imagine Meridian already imported the activity, reconciled the balance sheet accounts, booked the accruals, and rolled forward the supporting schedules. Instead of doing the production work, you open a short review list.
That’s not hands-off bookkeeping. Someone still needs to investigate a few exceptions, like a payroll liability that never cleared, an unusually large vendor payment, or a prior-period problem a client introduced.
In Episode 159 of The Unofficial QuickBooks Accountants Podcast, host Alicia Katz Pollock talks to Jessica McKellar, Founder and CEO at Pilot, about Meridian, Pilot’s AI accounting platform for firms. Pilot began pursuing software-driven bookkeeping automation around 2016. It tested the technology through its own full-service accounting practice and refined it across more than 8,000 businesses and hundreds of thousands of month-end closes before publicly launching Meridian in June.
The experience taught them that the breakthrough isn’t just the AI; it’s the accounting structure underneath it.
The breakthrough took a decade, not one prompt
Jessica and her co-founder started Pilot “pre all the AI stuff” with the idea that they could substantially automate bookkeeping with software. If they could create a reliable autonomous month-end close, they could build other useful services on top of it.
The goal went beyond efficiency. “We want every small business owner in America to have access to high-quality finance and accounting support,” Jessica says. For many owners, that support was out of reach.
Pilot is a technology company offering full-service bookkeeping, tax, and CFO advisory services, and it became Meridian’s test lab, automating recurring work and directing complex accounting tasks.
That history means Pilot didn’t design Meridian around hypothetical workflows. Its team gained experience by serving real clients every day. As Alicia notes, that’s different from vendors that rush autonomous ledgers to market before testing them at scale.
Meridian is also modular, ledger-agnostic, and industry-agnostic. It can serve as a broad technology platform for a solo firm or provide production automation within a larger firm’s existing systems.
Accounting context provides the guardrails
That foundation leads to the key technical difference: Meridian isn’t a generic AI agent waiting for an accountant to explain accounting from scratch.
“You don’t have to input a bunch of accounting context into the system,” Jessica explains.
The platform has a built-in understanding of:
- Supporting schedules and rollforwards
- Balance sheet reconciliations across reporting periods
- Vendor categorization histories
- Firm-level and client-level preferences
- Capitalization and materiality thresholds
- Bad debt policies
- Accounts that should clear on a regular cycle
Consider a bank reconciliation. Meridian can import activity, compare it with the expected closing balance, identify missing items, and reason through a remaining discrepancy. It can also recognize when the problem comes from a prior period and may require a conversation with the client.
New bookkeepers may reconcile bank and credit card accounts without realizing that payroll and sales tax liabilities also need attention. Alicia has found cleanup files carrying a $23 liability balance for five years because no one noticed it. But some errors are far more serious, like a broken revenue recognition schedule that overstates revenue for two years.
Jessica compares Pilot’s decade of development to building an “Iron Man suit for accountants.” During roughly the past 18 months, large language models became capable enough to operate inside that structure. The system can now do the work, check it, and show its work to the accountant responsible for the client.
An autonomous close still depends on good inputs
With those guardrails in place, the monthly workflow is much simpler.
First, the firm defines its standard checklist and policies. These may include capitalization thresholds, bad debt treatment, accrual methods, allocations, and QuickBooks class requirements. Firms can upload existing checklists or sample workpapers instead of rebuilding every step manually.
Next, Meridian needs complete source data. Firms can continue using QuickBooks Online bank feeds, connect through Plaid, or upload statements. Other inputs may include payroll exports, loan schedules, and point-of-sale data. Automation can’t finish a close when required information is missing.
Once you meet those dependencies, Meridian can run automatically or at the push of a button. It handles transaction imports, reconciliations, categorization, payroll allocations, accruals, revenue recognition, fixed assets, and schedule rollforwards.
The accountant then receives a review summary showing what Meridian did, what needs attention, which liabilities are open, and what questions may need to go to the client. The accountant can review that summary with the balance sheet, profit and loss statement, workpapers, and reporting package.
“All I have to do is a review,” Jessica says. That review still requires professional skepticism and responsibility. It’s not a rubber stamp.
Automation makes more client requests practical
Once routine production work shrinks, firms can reconsider projects that once required too many hours.
Through a chat interface, an accountant can ask Meridian to refine a chart of accounts, move vendors from operating expenses to cost of goods sold, restate historical SaaS revenue from contract PDFs, or apply QuickBooks classes to about 5,000 past transactions. The system also updates the related schedules, documentation, and supporting evidence.
These tasks were always possible, but they were often too expensive for the client or unprofitable for the firm. Automation changes that equation. “It lets you say yes to clients more,” Jessica says.
“The value is no longer data entry. The value is now opportunity,” Alicia says.
You still land the close
For QuickBooks professionals, the practical lessons are:
- Build automation on documented policies and reliable source data
- Use purpose-built accounting controls, not prompts alone
- Keep human review as the final control point
- Apply the time you recover to analysis, advice, and client requests
Autonomous accounting doesn’t remove the accountant, but it moves work away from mechanical production and toward review, judgment, and advisory services.
After all, the name is Pilot, not Autopilot. The system may fly the month-end close, but you still read the instruments, resolve the exceptions, and sign off.
Listen to the full episode to hear Jessica and Alicia explain the complete workflow on The Unofficial QuickBooks Accountants Podcast.
Alicia Katz Pollock’s Royalwise OWLS (On-Demand Web-based Learning Solutions) is the industry’s premier portal for top-notch QuickBooks Online training with CPE for accounting firms, bookkeepers, and small business owners. Visit Royalwise OWLS, where learning QBO is a HOOT!