Enterprise AI's Real Value Lies Beyond Chatbots: SAP CFO
The company's CFO says business value will come from AI embedded in finance, supply chains, and operations, where accuracy matters more than novelty.
News
- Enterprise AI's Real Value Lies Beyond Chatbots: SAP CFO
- ServiceNow's AI Deployments Grow Ninefold as Safety Takes Priority
- Jeff Bezos Pushes for AI Overhaul of Amazon Prime Video
- AI Demand Fuels Intel's Fastest Revenue Growth in 15 Years
- AI Dispatch: As Global Burns Cash, Local Builds Trust
- Jordan Nears Full Digital Government as 85.5% of Public Services Go Online
[Image: Nomita Samaiyar/MITSMR Middle East]
Enterprise AI’s next competitive advantage will come from reliability, not model size, according to SAP Chief Financial Officer Dominik Asam, who used the company’s second-quarter earnings to argue that businesses are entering a new phase of AI adoption.
After two years of heavy investment in generative AI, many organizations are still questioning whether those deployments have delivered meaningful productivity gains. SAP’s view is that the biggest returns will not come from rolling out general-purpose large language models across the enterprise, but from embedding AI into specific business workflows and grounding it in high-quality, well-governed enterprise data.
Asam said most AI token consumption today is concentrated in the “low-hanging fruit” of enterprise AI, particularly coding assistants and chatbots. Those use cases have gained traction because occasional hallucinations or factual errors typically carry limited consequences. A chatbot can produce an imperfect answer, or a coding assistant can suggest flawed code that is later corrected by a developer, without creating significant business risk.
The equation changes when AI is deployed in core business functions such as finance, supply chain management, or compliance. In these environments, a single error can ripple through multiple stages of a workflow, increasing operational and regulatory risk. In finance, for example, an inaccurate output can affect reporting, reconciliation, and compliance, making reliability far more important than in consumer-facing AI applications.
That, Asam argued, makes enterprise AI less about deploying a plug-and-play language model and more about designing systems around the needs of individual businesses. Effective enterprise AI requires company-specific knowledge, governance frameworks, and structured data to produce outputs that are accurate, auditable, and compliant with corporate and regulatory requirements.
Data quality remains one of the biggest barriers. Organizations operating with fragmented legacy systems and disconnected data repositories should not expect AI to solve longstanding information management problems on its own, Asam said. Poor-quality data not only reduces the reliability of AI systems but also increases computational demands, driving up token costs and weakening the business case for deployment.
He also pushed back against the assumption that the most advanced frontier models are always the best fit for enterprise use. Instead, he said, companies will increasingly choose the least expensive technology that consistently delivers reliable, safe outcomes. Depending on the application, that could mean conventional software, open-source AI models, or premium commercial models where higher performance justifies the added cost.
