I Completed a $7.6M Cash-to-Accrual Conversion in 3 Weeks. Here's What I Learned.

A few weeks ago I wrapped up one of the most complex engagements I've taken on as a solo practitioner.

A PE-backed behavioral health company needed their books converted from cash basis to accrual basis — fast. New investors were doing due diligence. The timeline was 3 weeks.

The scope was 15 months of accrual entries covering revenue and expenses.

Here's what made it complicated:

The company used two billing systems that overlapped during a mid-year transition. The parent company's accounting team was on the other coast. Revenue was $7.6M across the conversion period with month-to-month swings from $242K to $957K. And payroll ran through a shared ADP profile across three separate facilities — so every pay period had to be untangled before I could calculate what belonged to which entity.

I did the whole thing solo. And I used AI as my primary analytical tool throughout.

Not as a gimmick. Not as an experiment. As a working partner that processed billing data, built formula-driven Excel workbooks, calculated accruals, and helped me move at a pace I couldn't have matched on my own.

I want to share what I learned — because whether you're a controller, a firm owner, or an accountant thinking about how to do more with less, the lessons apply.

The engagement at a glance

The deliverable was a complete set of accrual-basis workpapers with journal entries ready to post in Sage Intacct.

That meant:

A revenue model built from scratch — analyzing 76 patients across two billing systems, calculating a net collection rate from a 9-month lookback period, building a monthly AR roll-forward, and producing 16 formula-driven journal entries. The opening accounts receivable balance alone required tracing 370 individual payment transactions back to their original dates of service.

An expense accrual package covering payroll (34 bi-weekly pay periods allocated across 15 calendar months by working days), prepaid insurance (8 policies with different terms amortized monthly), management fee smoothing (with a rate change mid-year), PTO liability calculated from individual employee salaries, depreciation on fixed assets, and AP cutoff analysis.

38 journal entries total. 800+ formulas. Zero hardcoded amounts — change one assumption and everything recalculates.

What actually worked

Here's the part most people are curious about — what does it actually look like to use AI on real accounting work?

It's not magic. It's delegation.

I didn't hand AI a trial balance and say "convert this to accrual." That's not how it works. What I did was break the engagement into discrete analytical tasks and delegate each one with clear instructions and source data.

Upload the billing files. Tell it what you need. Review the output. Correct what's wrong. Move to the next piece.

The power isn't that AI gets everything right the first time. It doesn't. The power is that corrections take seconds instead of hours. When I caught that the payroll stub was calculated using calendar days instead of working days, one sentence fixed the methodology and the entire workbook rebuilt in under a minute. That correction — calendar days to working days — changed the 12/31 accrual by $3,534. On a traditional engagement, that's an hour of rework. With AI, it was 30 seconds.

The client's director of accounting sent a single email answering seven different questions spanning payroll intercompany allocations, insurance policy corrections, management fee rates, AP dispositions, and more. I uploaded the email and said "analyze all of this and update the workbook." One pass. Six tabs updated. Fourteen open items resolved.

That's the real efficiency gain — not any single calculation, but the compound effect of moving faster on every iteration.

What AI can't do

I want to be honest about this because I think the industry conversation around AI in accounting is heavy on hype and light on reality.

AI cannot exercise professional judgment. When the calculated rate produced a negative ending AR, I had to assess whether that was a data problem, a methodology problem, or a timing issue — and present options to the client. AI processed the math. I interpreted what the math meant.

AI cannot manage client relationships. When the CFO questioned why two tabs in the workbook didn't reconcile, I had to explain the difference between patient-level totals and calendar-month totals in a way that was clear without being condescending. AI drafted the email. I shaped the tone.

AI cannot catch what doesn't smell right. When a patient showed zero payments but I knew from context he'd been paid, that was 28 years of accounting intuition. When payroll spiked from $50K to $95K per period in November, I knew to ask about intercompany entries before assuming it was growth. AI processes data. You interpret it.

And AI absolutely cannot protect you contractually. Documenting that you're the executor and not the methodology owner. Making sure the client certifies their data. Tying the timeline to data receipt, not a calendar date. That's your job.

The bottom line

This engagement would have taken me 80-100 hours working solo with traditional tools. With AI handling the analytical heavy lifting, I completed it in roughly half that time — with better documentation, more formula-driven workpapers, and a cleaner audit trail than I would have produced manually.

AI didn't replace me on this engagement. It made me faster, more thorough, and more confident in the final product.

That's the real story. Not that AI is coming for your job. It's that the practitioners who learn to work with it are going to run circles around the ones who don't.

Want the full breakdown?

I put together a 7-page case study with the actual process, the prompt patterns that worked, what went wrong, and the complete metrics from the engagement.

Download the free case study here →

No sales pitch. No course to buy. Just the playbook from a real engagement so you can see what's possible and apply it to your own work.

Stephanie Vasquez is the owner of Essential Accounting LLC, a fractional controller and financial infrastructure firm based in Santa Clarita, CA. She has 28 years of experience in accounting and operations.