How AI is changing procurement and supply chains

Behind every manufactured product is a complex supply chain heavily reliant on human knowledge, as well as data across email threads, documents, and enterprise software. It’s why direct procurement — securing physical materials and components needed to build products — has lagged behind business functions such as HR, finance and payroll when it comes to automation. According to data from Boston Consulting Group, a management consultancy, only 35% of procurement teams have adopted AI. Yet the need for better supply chain management is crucial, with 64% of manufacturers reporting that they spend 10% or more of their manufacturing budget reacting to supply chain disruptions, relying on last-minute production changes or emergency sourcing, according to inventory optimisation platform LeanDNA. “I think we all feel deep frustration sometimes with having to deal with the procurement side of the business,” says Robin Van Aeken, CEO of procurement and supply chain automation company Magentic, which is working to solve the problem by deploying AI agents to manage the procurement and supply chain process for its customers. “Everyone knows they want to get to this stuff working autonomously [...] but we have to take them on a human-in-the-loop journey where the agents are learning from the users until we get there,” he adds. In an interview with Fred Ellis, partner at VC firm N47 , Van Aeken unpacks how the company’s AI agents work in practice, and what automation means for the future of procurement and supply chain. Modernising procurement and supply chain with AI Direct procurement is complex and localised, says Van Aeken, adding it’s traditionally been owned by manufacturing plants, and by senior category managers who have known their three suppliers for 30 years. Advertisement Everyone knows they want to get to this stuff working autonomously. Because of this reliance on personal knowledge and relationships, AI often isn’t used in the direct procurement process. The main barrier is the inadequacy of legacy Enterprise Resource Planning [ERP] systems — software organisations use to manage daily business processes, such as finance, HR, supply chain, procurement and sales. These systems are designed to record financial outcomes, but not the reality of how outcomes were achieved and how decisions were made. “If I have a three-month negotiation [and I need] to understand the correct price of an item, what will be stored [on the ERP] is the final price that's paid,” Van Aeken says. “All of the negotiation, that context, the supplier performance, everything that went into it is not captured.” Van Aeken says that modernising this function is critical for a business to grow. "This is true in European manufacturing where we are competing so aggressively with cheaper, lower-cost manufacturing.” Magentic's core technology relies on the fact that AI agents can process and reason across unstructured data. The opportunity now is for them to deliver this reasoning ability across supply chains. What Magentic’s AI agents do Magentic’s AI agents extract data from sources, such as emails, documents, files and CAD drawings — digital models representing physical objects, buildings or machine parts. The procurement team of the future finds more win-wins with suppliers, does more value engineering, and pre-empts more disruptions. They can then use this data to verify pricing and purchase orders, and secure cost savings directly within enterprise systems. Rather than relying on human managers to manually verify every detail of a transaction, Magentic’s agents process these verifications autonomously. The technology can be adapted to the unique supply chain of a customer and the company works with enterprises across the automotive, oil and gas, and food and beverage industries. To build trust with these clients, Magentic typically starts new relationships using an outcome-based pricing model— charging only a percentage of the savings generated during a pilot phase — before transitioning to a customised subscription model once daily operations are fully automated. For one client, for example, Magentic addressed “the time it takes to go and do tasks like the checks on a purchase order,” says Van Aeken. “We were doing 1m purchase orders for a client a year that we were running agentically.” A purchase order is an official document sent by a buyer to a supplier detailing the exact items, quantities, and agreed prices for a specific purchase. By analysing these orders, Magentic’s AI agents are able to verify pricing accuracy, evaluate if the current supplier is the best choice, identify opportunities to renegotiate prices and reroute orders if a cheaper, pre-existing contract exists. We’re strongly opinionated that procurement teams are going to change. Magentic’s agents also identify and close ‘value leakage’— the hidden loss of capital that occurs when a company overpays or fails to negotiate savings due to process inefficiencies, human error, or missed opportunities across its supply chain. According to research by World Commerce and Contracting, value leakage costs enterprises, managing contracts and supply chains, on average around 9% of their annual turnover. "We're able to look across all of your spend, and say, ‘here's little areas where you could close off bits of value leakage and get some profit back for your company,’” Van Aeken says. The future of procurement Magentic's approach is to generate savings for its clients in a fraction of the time it usually takes when implementing new enterprise software. “With one of [our customers], in the first three months of going live, we saved them $1.2m. And in the six months after that, we've now hit $4.2m,” Van Aeken says. Speed is critical to building momentum within large organisations, he adds. “From contracts being signed to having $4.2 million saved, within a year, and using AI in some of the very low margin areas [...] it's massive.” For decades, procurement professionals have been slowed down by administrative work, and automation will allow them to shift their focus to more strategic work. Advertisement “We’re strongly opinionated that procurement teams are going to change,” Van Aeken says. “Procurement for too long has had to deal with work that isn't high value enough. "The procurement team of the future finds more win-wins with suppliers, does more value engineering and pre-empts more disruptions, all of which make procurement much more valuable." Van Aeken hopes that, instead of being viewed as an administrative hurdle, procurement will grow as “a more strategic function, a partner to the CFO, a partner to the manufacturing and to the engineering as well, as opposed to being too much of a cost centre.”