
National AP · In active build
Multi-billion North American GC
Customer Story
From a five-stage
manual cleanup,
to a self-managing pipeline.
How a major North American GC is rebuilding their Accounts Payable function around a Human Emulator that runs intake to JD Edwards E1 voucher posting, with real-time VIM ↔ E1 sync replacing the manual status-update layer the Project Accountants do today.
The Challenge
What the existing process looked like.
Vendor invoices arrived at INDPIMIACCPAY@pcl.com. PDFs were auto-ingested into VIM, where OCR interpreted vendor name, project number, purchase order, and amounts. In practice, VIM frequently misread invoices and assigned them to the wrong project. Any PDF (including non-invoice documents) was automatically ingested. Corrected invoices returned by vendors were flagged as duplicates and could not be re-uploaded without manual deletion.
Project Accountants manually reviewed newly ingested invoices, downloaded them into the appropriate project folder, and updated VIM statuses. There was no synchronization between VIM and E1, which meant no real-time visibility of voucher numbers and payment status. The administrative effort to track lifecycle stages was significant.
Audit-side, every clean invoice still required a manual MATX lookup for the MRR, a manual filename marker for partial vs. full receipt, and a manual audit pass against quantities, pricing, billing entity, freight, eco fees, and PST treatment. None of it was wrong. It just didn't scale as the firm took on more.

National AP · In active build
Doc: AP Process R2, National AP modernization
The Build
Five stages. One Human Emulator. No babysitting.
The Ruh AP Human Emulator owns the entire pipeline, intake to posting, with real-time VIM ↔ E1 sync replacing the manual status-update layer entirely.
01 · Intake & Routing
Distinguishes invoices from non-invoice PDFs at the mailbox. Auto-routes to the correct project and Project Accountant based on vendor, PO, and project context. Detects corrected re-submissions and supersedes the original automatically, no manual deletion. Applies the strict naming convention and files into the project's New Invoices folder.
02 · VIM Processing & PA Assignment
Replaces the manual VIM status update layer entirely. Statuses simplified to focus on Purchasing Issues and Vendor Issues. Real-time sync between VIM and E1 means voucher numbers and payment status update without a PA touching the keyboard.
03 · Receiving & Audit Validation
Pulls receivers from E1 using the PO. Reads invoice line data and retrieves the associated MRR / packing slip from MATX without manual search. Runs the full audit pass, quantities, pricing, billing entity, non-receiver charges (freight, eco fees, reels), PST treatment. Routes exceptions with full reasoning to the right PA.
04 · E1 Posting & Payment
Creates batches, assigns voucher numbers, renames PDFs by voucher number, runs ReportsNow batch check reports, and posts in JD Edwards E1. On error, voucher is reversed and a Teams notification is sent to the originating accountant. On success, batch moves to Posted Batches and VIM status updates to 'Entered into E1' automatically.
Bring your version of this story.
Same Human Emulator architecture, scoped to your workflow.
Field Note
Real-time VIM-to-E1 sync replaces the manual status-update layer entirely. The PA stops being the integration.
Engagement scope · AP R2
The Stack
Fits the tools the team already trusts.
Runs inside the customer's tenant. Token pass-through to the customer's model provider. No data leaves their environment.
ERP & Voucher
JD Edwards E1 · ReportsNow
OCR & Intake
VIM · Outlook (INDPIMIACCPAY@pcl.com)
Receiving & Match
MATX · Project SharePoint
Comms & Audit
MS Teams · Audit trail per agent
Guardrails
Where we explicitly keep humans in the loop.
Runs in tenant
Customer's AI license, customer's network. No data egress.
Human-in-loop on payment
Autonomous payment release is explicitly out of scope. Always a human sign-off.
Vendor master untouched
No vendor master data creation or modification. ERP migration out of scope.
End of case study
Run the same play at your firm.
Same Human Emulator architecture. Runs in your tenant. On your AI license. Against the tools your team already trusts.