"We want to make sure that AI can understand what the right decision is."

AI is rapidly making its way into procurement. Language models analyze documents, compare quotes, search for information, and formulate recommendations. But does more AI automatically mean better procurement decisions? In this interview, Hartmut Schwadtke, CEO and founder of Futura Solutions, explains why he prefers to use the term “procurement intelligence,” why business context is crucial for AI, and what specific roles SAP, procurement expertise, skills, and capabilities play in this context.

Mr. Schwadtke, everyone is talking about AI in purchasing. Why do you prefer to talk about procurement intelligence?

Because AI is, first and foremost, just a technology. Companies should instead ask themselves how they can make better purchasing decisions. After all, there’s usually no shortage of information. It’s already there—in SAP, requests for proposals, quotes, contracts, or supplier data. The key is to understand it within the right business context.

That is exactly why we are talking about procurement intelligence: By that, I mean AI that understands a company’s business context and applies procurement expertise to a specific decision.

What is AI currently lacking that would enable it to truly help people make better purchasing decisions?

Above all, it lacks the business context. An AI model can understand text, extract information, and compare offers. But that doesn’t mean it knows what this information means for a specific purchasing decision.

Is the price right? Is the supplier a good fit? What has been our experience so far? What are the requirements and alternatives? What do we already know from SAP? It’s only when data is viewed in context that it becomes a basis for decision-making. What AI can do is one question. Whether it understands the context of the decision is another.

What exactly does "business context" mean?

Business context means that information is not viewed in isolation. Let’s take an offer price of 100 euros. On its own, this number doesn’t mean much. It only becomes relevant when considered in the context of quantity, specifications, past prices, market trends, terms and conditions, the supplier, and possible alternatives.

An experienced purchaser draws such connections based on their knowledge and experience. If AI is to support them in making decisions, we must also provide it with this context. This isn’t about giving the AI as much data as possible. It simply needs the right, approved context for the specific task.

What role does SAP play in this?

A central one. In many companies, SAP contains a significant portion of the reliable business context. This is because it includes, for example, requirements, purchase orders, materials, suppliers, terms and conditions, and historical transactions. Above all, however, SAP maps real business processes.
For Procurement Intelligence, SAP is therefore far more than just an additional data source. This is where we differ from most others: Our applications run within the SAP system itself, based on the SAP data model.
This means we don’t have to create a new connection to this business context specifically for AI; instead, we can build on an integration that has evolved over many years.

But surely the entire context of a purchasing decision isn't contained in SAP?

That's right. Other relevant information can be found, for example, in requests for proposals, bids, contracts, technical documentation, and evaluations. In addition, there is external information such as market data, price trends, and creditworthiness information.

The challenge, therefore, is to make this diverse information usable for making concrete decisions—specifically, right where it is generated and managed. A new, centralized AI data repository would create a “second truth” that no one would maintain.

“In the past, the key skill was building the analysis. In the future, it will be knowing how to handle the results with confidence.”

How do you incorporate this shopping knowledge into the AI?

We break it down. We translate specialized procurement expertise into clearly defined skills, which we call “capabilities”: mapping bid line items to a request for proposal, evaluating pricing structures, and assessing bidder eligibility against award criteria.

The key is reuse. We build “Evaluate Condition Structures” once—after that, this capability is used in the “ award,” in the “Supplementary Review,” and in “Price Plausibility Check.” A utility analysis is therefore not a separate AI function, but rather a combination of existing capabilities.

For purchasing, this means: the same logic everywhere, making changes in one place instead of five. And the next use case doesn't start from scratch.

As a result, AI is gaining ever deeper access to corporate data and systems. How do you handle this?

This is where governance becomes crucial. Which AI model accesses which data, which capabilities it uses, and what actions it is permitted to perform—all of this should be in the customer’s hands.

The skill provides the framework; approval comes from the company. We call this principle “Controlled Openness”: open to technological development, but controlled when it comes to data, processes, and access rights.

Who will we entrust our company data to in the future?

The issue is not just whether AI is used, but also which AI model is used.

Today, companies should not commit to a single AI model indefinitely. Large corporations in particular, such as E.ON and RWE, are increasingly specifying which AI models are approved for processing their corporate data and expect that only those models be used.

That is precisely what our architecture is designed for: FUTURA remains model-agnostic. The customer decides which approved AI model to use. The business context, processes, and data access remain separate from this and under the company’s control.

What does procurement intelligence mean for the role of the buyer?

It’s shifting. Today, purchaser spend a lot of time searching for information, compiling data, comparing offers, and preparing analyses. Procurement Intelligence can take on an increasing share of this work. As a result, other skills are becoming more important: critically evaluating results, weighing conflicting objectives, negotiating, thinking strategically, and taking responsibility.

In the past, the key skill was building the analysis. In the future, it will be handling the results with confidence. AI can support decision-making, but the responsibility for the decision remains with humans.

If you had to sum up Procurement Intelligence in a single thought: What sets it apart from “AI in procurement”?

AI in procurement primarily describes the use of a technology. For me, Procurement Intelligence begins one level above that.
It combines AI with a company’s business context, with procurement expertise, and with the skills needed to make a specific procurement decision.

And that is precisely why, in my view, even the largest or most powerful AI model will not provide the decisive competitive advantage in the future. Models will evolve and become increasingly interchangeable.

The value that is difficult to replicate lies in the business context, in the processes, in procurement expertise, and in the ability to intelligently integrate all of these elements.
That is exactly where our FUTURA sourcing solution comes in. We don’t want to build the smartest AI in procurement. We want to ensure that AI can understand the right decision.

Thank you for the interview.