Fractional CFO

Who you are

• You think in reconciliations. Every number ties to something outside the system that produced it, or it is not a number yet. You form a working model of an unfamiliar ledger fast, find where the model is wrong, and say so in writing
• AI is how you work now. You have let a machine draft the mechanics and then caught what it got wrong, and you can describe exactly how you verified it. You write plainly, because the readout you hand a founder is the product. You can own a decision at 10 to 15 hours a week, which means you drop the re-performing and keep the judgment
• You have probably made a close fast and kept the before-and-after numbers, or built a 13-week forecast a CEO ran cash on. Or you refused to sign a number, and you can say what happened next. A CPA is a plus. We care about the artifact and the reasoning more than the firm on your resume
• If your resume is unconventional but the work above is yours, apply anyway

What the job involves

• A controller-grade seat with CFO judgment, 10 to 15 hours a week. You sign the monthly close of a profitable $22 million company whose books are kept by machines, own its 13-week forecast, and write the one-page readout the founder runs on
• Product.ai is the verified truth layer for shopping: when a person or an AI agent needs to know what is actually true about a purchase, we answer with proof. SimplyCodes is the first proof at scale, the code verification service whose robots run real checkouts so shoppers only see codes that actually work
• Earns about $22 million a year. Profitable. Bootstrapped. Founder-owned since 2009. No outside investors. No board. Fewer than twenty operators, outbuilding companies 10x our size
• Today the close is signed in-house, and this seat makes the signature independent. An in-house ledger anchors every transaction to a bank or card statement, and a month closes only when the categorized rollup re-derives the bank-verified cash change to the cent. AI agents ingest the statements, post the reconciliations, and flag the exceptions. A company that sells verified truth keeps its books by the same rule, so a number counts as true only when it ties to a source outside the system and a person independent of the machine signs it
• The machine already fills the preparer seat, so you are the CPA-grade eyes on the close. You own the rules the machine runs on, decide what it could not, prove nothing was dropped, and put your name on the month. You also own the 13-week forecast and the operating budget, and you write the one-page readout the founder runs the company on
• You work directly with the founder and with the operator who runs day-to-day finance and administration. You hand clean financials to our outside tax CPA firm. You hold no payment authority, by design
• A statement-anchored ledger. Every month reconciles to the cent against bank and card statements, and the close signs only when the categorized rollup re-derives that cash change exactly. It is a custom, in-house ledger, and we are not migrating it. Comfort inside a client-built system is the scarce skill
• Machine-kept books with a human signature. AI agents do the mechanics, so they ingest, match, post, and flag. You work in a review dashboard and in written rulings. The agents post, you rule on the exception queue, and the ledger records your call. The signature is an internal management sign-off, and our outside tax CPA firm remains the external preparer
• Affiliate commission revenue. Commissions accrue when a network confirms a sale and reverse when the shopper returns it. Settlement comes on the network's calendar, so revenue recognition and the cash forecast both depend on that lag
• A standard general ledger kept in sync for the tax CPA. We keep a standard general-ledger package in sync from our ledger. Our ledger is the source of truth, so the mirror never originates a number. You run the sync and verify each month's balances against the statement anchors
• An owner-operator structure. One owner today, plus a team ownership program. The books must keep its capital accounts, distributions, and grants clean from the first month
• The monthly close and its signature. Stage the statements, make the categorization rulings, reconcile to $0.00, and sign. You own the exception ledger, so every rejected or corrected item is accounted for and every rule change is authorized by you. Statement and payroll-journal intake runs on your clock. If you and the founder disagree on a ruling, your dissent goes into that ledger in writing, and if you refuse to sign a month, the month stays open until the exception is resolved. The bar is a month that closes without the founder doing the mechanics and a number he trusts enough to act on
• The cash picture. The engine refreshes the cash numbers daily, and you own the 13-week forecast built on them, its assumptions, and what it tells the founder. The operating budget is yours too. You watch and flag the cash floor, the reserve target, and the card-balance rule; the founder moves the money. You set the spend targets and report each month where spend stands against them
• The outside handoffs. You hand our outside tax CPA firm a clean year-end package on time, and you run the quarterly reconcile the firm signs against the mirror. Counsel and advisors get fast, exact answers from you. The mirror stays in sync
• The founder's monthly readout. One page. Cash, margin, spend against budget, and what changed, written so the founder can act on it in five minutes
• The close, the readout, and the mirror sync are monthly. You refresh the forecast on your own cadence. Counsel questions come in bursts, and the year-end package is once a year. That is how the seat fits in 10 to 15 hours
• The craft you must already own: month-end close discipline, bank and card reconciliation, revenue recognition judgment, and hands-on 13-week forecasting. Comparable experience we accept: a controller or CFO seat at a $10 to $50 million company, or a fractional practice with real closes behind it. A fund whose books you kept yourself counts too
• What you will grow into here is signing books a machine prepared. You will design the checks an AI system cannot fake, rule on an exception queue instead of keying entries, and authorize changes to automation you did not write. That is the controller seat of the next decade, and few people have sat in it yet

Compensation

$125 - $175 / hour

Application process

• Async video screen. Brief and self-recorded: about 15 minutes, whenever works for you
• A structured call with the operator who runs day-to-day finance. The same questions for every finalist, recorded with your permission, so the comparison is fair. One of them is the question this seat turns on: what would you need from us before you put your name on a month?
• Conversation with the founder. How you model a ledger you did not build, and what you would check before you sign
• Paid working session. One past month of our real books, names masked, with errors planted in it. For about a day, at your stated rate and against the clock, you work in our tools. We score what you catch and count what you flag that was not wrong, and all of it comes before any live access or bank feed
• Complete a brief video questionnaire to apply for this role. It takes 12 minutes to calibrate initial compatibility. Your only required preparation is reading the Guide linked above. Beyond that, just bring clarity of mind and a readiness to share who you are

Location

Los Angeles, Remote from US