SAP's finance chief said on Thursday that artificial intelligence in enterprise software must move beyond chatbots and coding tools into more complex business processes, where clean data, reliability and cost control matter more than access to the most powerful model.
Companies have poured money into generative AI but are still seeking evidence of broad productivity gains, and SAP is arguing that the returns will come less from general-purpose models than from governed systems embedded in specific business processes.
CFO Dominik Asam told reporters after SAP's second-quarter results that the "lion's share" of AI token consumption today was spent in "low-hanging fruits" coding assistant and chatbots, where AI's hallucinations matter less because the output carries limited risk if it fails.
But applying AI to finance, supply chain or other core business processes is harder because errors carry over multiple steps, increasing risk against compliance standards, he said.
"If you have some hallucinations in the process, the errors will actually compound statistically over many steps," Asam said, referring to finance workflows. "It requires much more excruciating assurance levels."
The "high-hanging fruit" of AI, Asam said, is less about applying a generic plug-and-play large language model across a company than about building systems around specific businesses.
That requires companies to make their own data usable and governed, so AI can operate with the knowledge of the company. "The idea that AI will solve all these problems if they are messy, legacy data silos is not true," Asam said, adding that such an approach came with "extremely high token costs."
The most advanced model is not always the right one, he said. In practice, he said, companies will use the cheapest reliable tool that can deliver the required outcome safely, whether that is simple software, an open-source model or an expensive frontier model.
