Month-end close prepared by an AI agent and reviewed by CRCFO staff
Client name withheld
Background
This engagement was the first in which CRCFO used an AI agent to prepare a client’s month-end close. In 2026, the client’s chief executive asked CRCFO to use AI in delivering the company’s finance and accounting services. The client’s books remain in its existing QuickBooks file, and the AI agent works directly in that file. No migration was required, and the client did not need to learn a new system.
How the close works
An AI agent, operating on a close platform connected to the client’s QuickBooks file, performs the production work: recording and classifying transactions, reconciling accounts, and preparing the financial statements. CRCFO staff trained the agent on a month that the team had already closed manually. From that completed close, the agent learned how this client’s transactions are classified and recorded, and it applies the same treatment each period. The agent records every action it takes, so its work can be reviewed and verified. It flags any item outside its training for a staff member to resolve.
Accruals
The client’s books are kept on the accrual basis. When a vendor bill arrives after month end for work performed in an earlier month, the agent records the expense in the month in which the work was performed and reverses the accrual when the bill is posted. For example, legal fees billed in mid-September were accrued to August, and a consultant’s bill covering mid-August through mid-September was allocated between the two months. The agent also records a recurring monthly payroll accrual based on an established pattern, and it records other accruals when a staff member instructs it to do so. Annual subscriptions paid in advance are recorded as prepaid expenses and expensed over the months they cover.
Each month, the agent’s close checklist also identifies recurring expenses that appear to be missing and transactions that may relate to more than one month. The agent cannot accrue a cost for which it has no bill, established pattern, or instruction. As with a manually prepared close, the reviewer remains responsible for identifying work performed but not yet billed.
Review and approval
A member of CRCFO’s accounting team performs the close review on this engagement. Each period, the reviewer examines the agent’s work and resolves the items it flags. During implementation, the reviewer worked directly with the platform provider’s onboarding lead. The engagement CFO approves the financial statements before they are released to the client.
Results of the August close
- The August close was completed and submitted for CFO review.
- In the reviewer’s assessment, the agent completed most of the work.
- A few items required staff input before the close could be completed.
Those items were resolved through the exception process, which is designed to route such items to staff. The agent gathered, matched, and assembled the data, and CRCFO staff performed the review, exercised judgment, and approved the results.
Management reporting
In addition to the close, the engagement provides management reports that the client uses to make decisions. An AI model analyzes the client’s complete transaction detail for each period rather than a sample. From that detail, it prepares budget-to-actual reports, a consolidated report of subscription and recurring spending, and contractor spending reports. All three reports address needs that the client’s team identified on the kickoff call.
Because the AI model can analyze the complete transaction detail at once, CRCFO can respond to a question the client raises in a meeting with a full analysis while the related decision is still open. Each such analysis identifies outliers, traces figures to source records, and includes CRCFO’s recommendation. Without this capability, questions of this kind are often answered only after the next close.
Each report was built once and is refreshed each period with current data, so the client receives a consistent report each month without the report being rebuilt.
Data privacy
CRCFO uses an enterprise AI platform. Under the provider’s commercial terms, client data entered into the platform is not used to train the provider’s models. Client data is used only to deliver services to that client.
Visit www.crcfo.com to discuss an AI-prepared close for your company and to learn more about our financial services for technology companies or to schedule a consultation.