Discussion

Pamarco says AI agents now handle about 40% of its month-end close tasks

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#10798

A manufacturer says AI agents now handle about 40% of its month-end financial close tasks across seven legal entities, including checking work, coding invoices and preparing reports. The example, detailed by PYMNTS from a Microsoft case study, shows how the technology is being put to work inside a finance team, with people still responsible for exceptions and approvals.

Watch Desk analysis

What happened

Pamarco runs the agents on Microsoft Dynamics 365. They check outstanding close tasks hourly, carry out work when dependencies are complete, update task status and email the results. Staff retain responsibility for judgment, review and exceptions, according to the case study.

The company says it has automatically coded about 95% of freight invoices. After acquiring a German business, its US team built coding rules for the new entity, and fewer than 5% of those invoices required manual coding during validation. Pamarco’s chief information officer said the agents could take on 50% to 60% of close tasks by year end.

Why it matters

This is a concrete account of AI being fitted into existing accounting workflows, rather than being asked to replace the whole finance department. The figures describe tasks automated, not the share of accounting labour removed: the case study does not break down how long those tasks took before automation or how complex they were.

The practical hurdle may be less about getting an agent to perform a defined task than connecting it reliably to the company’s systems. PYMNTS reports that 90% of CFOs in a survey named integration complexity as a barrier to further automation, up from 67% in 2025.

Our read

The useful detail is the division of labour: agents handle routine, bounded tasks, while employees keep the approvals and judgement calls. That is a more credible picture than “AI does accounting”, a phrase with a habit of doing rather a lot of work itself. Treat the efficiency figures as Pamarco’s account, not an independent productivity study.

What to watch

  • Whether Pamarco reports results for the additional close tasks it expects to automate.
  • How the company measures errors, review time and savings alongside the share of tasks automated.
  • Whether integration remains a major obstacle as it extends agents to sales, quoting and order entry.

Discussion spark: Would you trust AI agents to run routine finance tasks if staff retain approval and exception handling, or should more of the workflow stay manual?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.