Case Study
THE PROBLEM
As invoice volume grew, the accounts payable process at this industrial-kitchen equipment manufacturer stayed tied to paper. Invoices and packing slips were kept in physical folders, filed by location rather than digitized into any system.
Finding a document meant knowing exactly where it lived and going there in person. If the record was at another plant, that meant travel. If the person who needed it wasn't on site, someone else had to locate the paper, scan it, and email it over.
This became especially costly when a single document needed to be reviewed by multiple people over several weeks. Each review cycle meant repeating the same search, the same trip, and the same scan, for a piece of paper that had already been found once before.
WHY IT MATTERED
None of this showed up as a single dramatic failure. It showed up as friction, repeated over and over. Every reviewer who needed a document paid the same cost the last person did: locating it, physically retrieving it, and getting it into a form someone off site could actually use.
Information that should have taken seconds to pull up instead depended on someone's memory of where a folder was kept, and their availability to go get it. As invoice volume increased, this pattern didn't get easier. It just repeated more often, across more documents, with no digital record to fall back on.
THE SOLUTION
Arumai Standard deployed an AI agent to sit on top of the accounts payable inbox, digitizing packing slips, three-way matching them against purchase orders and invoices, consolidating every related document into one digital record, and generating the purchase invoice itself, a step the accounts payable team had previously done by hand.
Getting an AP automation deployment to work in daily use took three things:
The visible work was the agent. The real work was the data foundation and the exception handling underneath it, and that is usually where in-house attempts stall. That distinction held because Arumai Standard is led by an advisor who has run AI from inside manufacturing operations, not advised on it from the outside.
THE OUTCOME
The change is qualitative, but it is direct. Documents that once required a physical trip to a specific plant location can now be viewed digitally and immediately, by anyone who needs them, wherever they are. Each transaction now has one record instead of a scattered paper trail, with the packing slip retained alongside it so questions about what was actually received can be answered without searching for paper.
Repeated physical trips to re-locate the same document for successive reviewers are no longer necessary. The accounts payable team no longer creates purchase invoices by hand.
The team recovered 25 hours per week that had been going into manual invoice processing. That figure was measured after deployment. The time people used to spend locating documents was not formally measured, but the searches and trips it removed were a daily source of delay.
WHAT THIS SHOWS
This AP automation deployment did not start with AI. It started with digitizing paper records into a structured, searchable foundation, and only then applying an AI agent to automate the matching and data entry work sitting on top of that foundation. For ETO manufacturers, including those building equipment for industrial kitchens, weighing where to start, this is the order that works: build the digital foundation first, then deploy AI on top of it.
FAQ
Timelines depend on how ready your existing records are, since the work centers on pulling every packing slip, purchase order, and invoice into one trusted structure, teaching the agent to handle exceptions instead of failing quietly, and validating results against real invoices until the team trusts them. In this deployment, that groundwork came before any time was recovered. Book a call to talk through what that groundwork would look like for your operation.
No. The agent took over the matching, digitizing, and invoice creation work the accounts payable team used to do by hand, but people still handle the exceptions the agent flags. In this deployment, that shift freed up 25 hours per week the team had been spending on manual processing.
Three-way matches don't always line up cleanly, so the agent flags the mismatch and routes it for review rather than failing quietly or forcing it through. Getting that exception handling right was one of the three things this deployment had to solve before the accounts payable team trusted the output.
This ETO manufacturer recovered 25 hours per week that had been going into manual invoice processing, measured after the deployment went live. Results depend on your starting volume and how paper-based your current process is. Book a call to talk through what's realistic for your team.
Book a Call
If your team is dealing with paper-based processes that slow down accounts payable or any other part of the plant, we'd welcome a conversation about where to start. No commitment required.
No commitment. No pitch deck. Just a conversation.