Lemvigh-Müller Uses AI Agents to Automate 100,000 Supplier Order Confirmations
The Danish wholesaler is using AI to process the emails and PDFs that sit outside its EDI network, showing how distributors can automate established workflows without replacing their core ERP systems.
Lemvigh-Müller is using AI agents to automate one of the most persistent manual processes in wholesale distribution: processing supplier order confirmations.
The Danish distributor of steel, plumbing, heating and electrical products sends approximately 175,000 purchase orders to more than 2,000 suppliers each year. While some confirmations arrive through EDI, around 60% are still received as unstructured emails and PDF documents.
These documents previously had to be reviewed manually and compared with purchase orders. Changes to pricing, quantities or delivery dates could take hours or days to reach the company’s systems.
Lemvigh-Müller has now developed an automated workflow using SAP Business AI, working with NTT DATA Business Solutions. The system uses three specialised agents to manage different parts of the process.
One agent receives and sorts supplier emails and attachments. A second extracts prices, quantities and delivery dates from the documents. A third compares that information with the original purchase order in SAP and determines whether the confirmation matches or requires human attention.
The company expects the solution to automate more than 100,000 confirmations annually. Reported results include a touchless processing rate of more than 90% and approximately 98% matching accuracy. The project moved from its initial tests into production in around ten weeks.
The project is significant because it addresses the large amount of B2B activity that remains outside fully integrated digital channels.
EDI has automated transactions between established trading partners for decades, but connecting every supplier is rarely practical. Distributors continue to receive quotations, confirmations, invoices and delivery documents through email, often in inconsistent formats.
AI provides another way to process this information. It does not replace EDI or the ERP. It creates a bridge between unstructured documents and the systems that already manage purchasing, inventory and customer orders.
That connection also has implications for B2B eCommerce and customer experience. When supplier changes are identified and recorded more quickly, distributors can provide customers and internal teams with more accurate information about availability, pricing and expected delivery dates.
The value is not simply the time saved in procurement. Better upstream information can improve the quality of the experience delivered through websites, customer portals, sales representatives and service teams.
The project also exposed areas where master data needs improvement, including information related to Incoterms. AI can accelerate the process, but the reliability of the outcome still depends on the quality and consistency of the underlying business data.
Lemvigh-Müller expects the system to release capacity equivalent to three or four full-time roles. The company says the objective is to redirect employees toward complex exceptions and supplier discussions rather than reduce headcount.
The same approach could eventually extend to other documents and workflows, including invoices and order management.
For manufacturers and distributors, the practical relevance is clear. Useful AI projects do not always need to begin with a large transformation program or a customer-facing assistant. They can start with a repetitive process where the inputs are unstructured, the rules are understood and the business value can be measured.
Customers may never see the agents working. They will experience the result through faster responses, more reliable information and fewer order problems.