CUSTOMERS
How a global technology leader deploy digital workers to resolve pricing leakage end-to-end
Industry
Industrial technology
Cases
Value leakage prevention
Technology stack
SAP, Azure, Snowflake
Manufacturing runs on complex, multi-tier supply chains where prices are negotiated hard, only to drift. A purchase order often match its supplier confirmation line for line, and still both somehow sit above the price agreed weeks earlier because no one updated the price record in between. Across hundreds of purchasing categories, that difference adds up to millions in lost value for companies each year.
For a global technology leader with almost €80B in revenue, chasing down that leakage had become a major, recurring part of daily work. The procurement teams were manually investigating deviations across contracts, systems, and price lists, then spending hours resolving each one.
Some existing processes and SAP workflows could detect deviations between two documents, such as a PO and a supplier confirmation, but then left it to the buyer to resolve. That meant spending hours tracking down the relevant details and working out how to resolve the issue, often across 200+ cases per week.
They came to Magentic with a clear ask: automate the entire process end-to-end, from identifying the root cause to finding the evidence, proposing a resolution, and even drafting supplier communications to correct the mistake, with a clear audit trail.
Putting digital workers to work
We started a pilot to prove what our digital workers could do. The proof of concept runs directly against the company's live Data Cloud and Snowflake environment and within existing Incoterm and MOQ workflows, with ERP data replicated daily.
The workers cover both direct and indirect purchases, checking for pricing issues, mismatches, payment terms, MOQ, and scale issues across 165 purchasing categories within one division alone. They then take action inside Teams, with the evidence attached and the draft ready to send.
Eight weeks from kickoff, the digital workers have already found a potential overcharge pattern surfaced across 33 purchase orders, with one proof of concept example flagged a supplier-backed price roughly 34% above the expected tier.
These cases are presented to the buyer as a final approval check, not as issues on a dashboard, with the company already looking to automate these checks without a human in the loop.
What's Next
Broader functional user testing ran through June and July with an initial cohort of seven to eight users, followed by a formal pilot start in early July. The company set a goal of running the experience inside its own environment by August, with contract system integration through a third-party contract platform following later in the year.
Early signals from the pilot have been encouraging: buyers are resolving cases faster and with more confidence, and the team is already looking past the initial rollout toward scaling the approach across additional categories and business units. What started as a pilot is shaping up to be a blueprint for how the company tackles price leakage going forward.
