No Priors: Artificial Intelligence | Technology | Startups SAP: Bringing the ‘Operating System’ of a Company into the AI Era with CTO Philipp Herzig
191 snips
Apr 23, 2026 Philipp Herzig, SAP CTO leading its enterprise AI strategy, talks about turning business software into an AI-ready system of record. He digs into why enterprise AI breaks at scale, why security and fragmented data slow adoption, where agents can deliver value first, how pricing may shift in the AI era, and why quantum optimization has his attention.
AI Snips
Chapters
Transcript
Episode notes
Agent Mining Turns Human Decisions Into Training Data
- Verifiable enterprise agents improve when companies capture decision traces, tribal knowledge, and user clarifications that never entered systems of record.
- Philipp Herzig calls this agent mining, extending process mining into a flywheel of observability, evals, anomalies, and better standard procedures.
Why Tool Calling Will Beat Computer Use Most Often
- Philipp Herzig expects most enterprise agents to use tool calling and background workflows, with computer use filling gaps where APIs or modern integrations are missing.
- He argues UI automation is improving fast, but structured integrations remain preferable for reliability and scale.
Why LLMs Alone Cannot Run Enterprise Forecasting
- LLMs excel in unstructured work, but enterprise planning still needs specialized predictive models for demand, cash flow, classification, regression, and time series.
- Philipp Herzig says SAP built RPT-1 to bring transformer-style democratization to relational tables without requiring armies of country-by-country data scientists.