Pamarco, a maker of precision parts for the printing and packaging industries, says AI software now handles about 40% of the tasks in its monthly close. The close is the routine of finishing and checking the books at month end. The figure comes from a Microsoft case study published in September and reported by PYMNTS.
What changed
Pamarco runs the AI agents inside Microsoft Dynamics 365, a business software system that tracks money, orders and invoices. The agents code invoices, run close tasks and put together board reports. The company has seven legal entities in the United States and Europe and serves customers in almost 70 countries.
Glyn Wilson, Pamarco's chief information officer, said the agents could take on 50% to 60% of close tasks by year end.
How it works
Each close task has an owner, a due date and a list of steps that must finish first. The agents check for open tasks every hour. When a task is due and the earlier steps are done, an agent does the work, updates the status and emails the results.
- Each agent has a set workload, set permissions and a measurable time savings goal.
- Employees still handle exceptions and approvals.
- Task owners keep responsibility for judgment and review.
Freight bills show the payoff. Most of Pamarco's invoices without purchase orders are freight bills. The company reports about a 95% success rate coding them automatically, even when several tax codes apply. After it bought a business in Germany, fewer than 5% of that entity's invoices needed manual coding during validation.
What the 40% does not mean
Automating 40% of tasks is not the same as cutting 40% of accounting work. Tasks vary in size and time. The case study does not say which tasks the agents run or how long they took before. Wilson also said the company does not think of AI as a way to reduce headcount, but as a way to get more from its employees.
The same approach reached reporting. A sales director who spent about five hours a month preparing sales analysis for the board now gets an automatically generated presentation on the sixth working day.
Why it matters for other businesses
The lesson is where Pamarco started: repeat tasks with clear rules, with people still approving. The hard part is connecting systems. In a PYMNTS Intelligence survey of 100 U.S. CFOs in July 2026, 90% named integration complexity as the top obstacle to automating more, up from 67% in 2025. The firms surveyed had $250 million to $2.5 billion in yearly revenue.
There is interest in going further. In a separate October 2025 survey of 60 CFOs, 42% at companies with at least $1 billion in revenue expected high-impact results from AI that coordinates work across several finance systems.
What happens next
Pamarco plans to extend the agents to sales, quoting, order entry, customer response and vendor communications.
The takeawayStart AI in finance with repeat tasks that follow clear rules, and keep people in charge of exceptions and approvals.