Automatic translation of incoming order e-mails into ERP-ready orders
Verkooijen Veem's order administration replaced manual PDF entry with an AI dashboard that prepares every order for its ERP, cutting average processing time to about five minutes.
Verkooijen Veem is a Belgian logistics company that brokers transport, storage and repackaging of plastic granulate. Positioned between the ports of Rotterdam and Antwerp, it serves international clients across Europe and Asia. We worked with its order administration department.
Challenge
Verkooijen Veem receives 50 to 70 orders a day by email, mostly as PDF attachments. Two employees entered each order into the ERP system by hand. The details they needed were scattered across large documents, often over several pages. Each order took at least 15 minutes. As volumes grew, the team could no longer keep up within working hours. That put a key client promise at risk: every order is processed within two hours. The deadline was missed a few times a week, which hurt client relationships.
Solution
We built a web dashboard that reads order emails and their attachments. An AI model extracts the details the ERP needs and fills in the right fields. Employees check the result and load the order into the ERP in one step.
- Drag-and-drop intake: employees drag order emails from Outlook into the dashboard and mark them as inbound or outbound.
- AI extraction: reads the email and its PDF, Word or Excel attachments and structures the data.
- Review screen: shows every field, so employees can correct details before sending.
- One-click ERP entry: sends approved orders straight into Boltrix, the company's ERP.
An employee checks every order before it enters the ERP, and the tool runs in Verkooijen Veem's own Azure environment.
From feasibility test to daily tool
First prove the extraction works, then build around the team.
- Feasibility study On a client test dataset, we measured how accurately OCR models filled the required ERP fields. The study, planned for four weeks, confirmed the tool was feasible.
- First version We built a working skeleton and demo to test the flow.
- Fine-tuning We refined the tool around the team's specific requirements as they surfaced.
- Go-live We deployed in the client's Azure environment and fixed early issues quickly through short feedback loops.
Pieter de LangeAI Engineer
Mart KempenaarAI Engineer
Noah van PottenAI Engineer
Hamza LamrabetAI EngineerImpact
Order administration now spends far less time per order. Instead of searching PDFs and typing data, employees check prepared fields and send the order on. An average order takes about a third of the time, and complex orders save up to half an hour. The team now handles the daily volume much more consistently within the two-hour client deadline. Three people in order administration use the tool every day. Verkooijen Veem is now exploring the same solution with its partners.
More cases
At AdamI, we transform businesses by using the technology of artificial intelligence
