IT Brief US - Technology news for CIOs & IT decision-makers
United States

BeNeering CEO: Procurement AI is about long term journey for ROI

Thu, 19th Nov 2026 (Today)
Jake MacAndrew
JAKE MACANDREW Interviews Editor

Some procurement leaders are still waiting for their AI investments to pay off, BeNeering founder and CEO Christoph Moll said, citing conversations with Chief Procurement Officers. He said companies looking to automate purchasing need to define their buying rules and check the quality of the historical data their tools will use.

The company, whose platform serves more than 400,000 users worldwide, builds software that sits alongside SAP systems. It works with the data held in those systems and adds an AI layer intended to automate parts of purchasing and sourcing.

Companies now have AI strategies and budgets, and many are working towards what SAP calls the autonomous enterprise, in which fewer people are needed to run complex processes, according to BeNeering's CEO Christoph Moll.

Returns, however, are slower to arrive, according to discussions Moll has had with clients. Chief Procurement Officers have told BeNeering they are not yet seeing a payback on their AI spending, and Moll put the share of cases still at the investment stage.

"There will be return on investment for sure, but it's not from today to tomorrow. It takes time," said Moll. "You have to do background work in advance, so you have to prepare that you can define clear expectations on how AI should help you. I think this is also the problem with companies and people. If you cannot address what you expect from AI, it will not help you, because it cannot understand your mindset."

Moll advises procurement leaders to treat AI as a journey rather than expect results within weeks. They should involve their staff and assess which use cases justify the investment, potentially with advisory partners. As programs grow, they will also need to consider how to orchestrate multiple AI agents and how to use historical ERP data and knowledge bases. Communication with suppliers during sourcing and order acknowledgement is a further consideration.

Data quality

Another hurdle is data. Moll said companies must decide which sources AI should treat as trustworthy. One example is the purchase order history held in SAP, which could be used to suggest suppliers based on the previous years of dealings, or to classify a requirement by category such as facilities, tools or services.

"If your SAP data is bad, I mean your historical purchase order data, because nobody was taking care to classify the old orders and they are assigned to wrong categories, then your prediction by AI will also be wrong," said Moll.

Large SAP landscapes add further complexity. Moll said big German DAX-listed customers often run between three and four SAP back-end systems, and some run 20 or more. BeNeering does not consolidate this data into a single source. Instead, its platform connects multiple back ends and routes each user to the correct one. An employee in Brazil, for example, would be connected to the system responsible for Brazil, and someone in a mobility division to the mobility system. Users who work across several systems can switch between them.

The same logic applies to its AI. Because one supplier can carry a different number in each SAP system, the AI is designed to work with inconsistent identifiers rather than wait for the data to be harmonised. Moll said harmonising data across systems at a large customer can take years and cannot be done in two or three months.

Next phase

Moll sees the next step as letting business users build small applications inside the procurement platform by prompting AI, rather than raising requests with IT or development teams.

Many procurement teams, he said, run processes in complex Excel spreadsheets that deviate from standard practice and cannot be moved into off-the-shelf software. A user could load such a spreadsheet and ask AI to build an app to manage the data.

"For smaller apps, business people, analysts or analytical people can describe what they need, and then you get this software developed by AI. You're not coding it anymore," added Moll. "This would be also appreciated a lot by customer if you can build also your apps in other environments. If you go to the hyperscalers, you can build apps with AI, but to build it into a procurement platform, into an existing platform - this is this what I would see also as the next level that is coming."