Rillet's $100M reads as proof mid-market ERP is rip-and-replace, mostly at the cheap end

Split the 600 accounts the way Koop splits them and the displacement story narrows. He says half of Rillet's customers came off Intuit, 30% off NetSuite and Sage Intacct, and 20% off Oracle, SAP, Workday and Microsoft products ^[9]. Applied to 600 customers, that is roughly 300 wins from QuickBooks-class books, 180 from the mid-market suites, and 120 from the tier-one vendors ^[20][24][21]. Companies leave Intuit when they outgrow it, and always have. The genuinely new part of the claim is the 120, plus the unquantified note that many of the clients added since last summer are public companies [c6b].

That is why the EY alliance is the more consequential line in the same board update. Rillet says it has an alliance to introduce AI tools to the audit firm ^[6]. In mid-market finance the expensive part of replacing a general ledger is not implementation, it is explaining the new ledger to an auditor halfway through a fiscal year. Three months ago Rillet shipped a governance feature that lets accountants see and audit every decision an agent made, including which numbers it pulled and how it calculated them ^[18]; Koop says the difficulty was compressing agent data into a form a human can read [c18b]. That work is aimed at the audit file rather than at month-end close.

The shortage argument has a ceiling written into it. Koop attributes the demand to the US accountant shortage ^[8], but rules for public companies currently require that every transaction an AI agent makes be approved by another human ^[19]. An agent that needs a signature per transaction moves where the work happens without removing the person, and that requirement binds hardest on exactly the public company accounts Rillet has been adding [c6b].

As a financing event, this is mostly repricing by people already inside the company. Rillet has raised $200 million since leaving stealth two years ago, which makes this round half of everything it has ever taken ^[3][22]. Iconiq was in the $70 million Series B last summer and led this one, with Seth Pierrepont taking a board seat and saying the deal "wasn't a cold start" ^[10][12]. Sequoia led the Series A ^[11]. Forty-eight hours is a fact about what existing investors already knew, not about price discovery.

The growth figure needs the same discount. An annualized revenue rate that doubled in a single quarter ^[5] compounds to sixteen times over a year if it holds ^[23], and nobody underwrites that; the defensible read is one strong quarter off a small base. What Rillet has shown, to its own customers as much as to Oracle's, is that a general ledger can be pulled out and handed to a two-year-old vendor ^[7][3]. That cuts both ways, and it will turn up in retention before it turns up in the next valuation.