At VB Transform 2026, Max McPhee, senior solution advisor at SAP, discussed with Rob Stretchay, lead analyst at VentureBeat Research, what enterprises must do to progress from chatbots to autonomous AI agents that can execute real business processes. McPhee argued that the crucial difference is grounding agents in the company’s specific context instead of relying solely on general knowledge.
Building enterprise context with knowledge graphs and vector embeddings
McPhee said the principles used to onboard new employees should be adapted for onboarding software agents. He emphasized knowledge graphs and vector-embedded data as powerful formats because they are easy for agents to search and retrieve. That kind of enterprise grounding prevents agents from getting confused by internal shorthand or acronyms, a common issue in SAP environments.
Governance, identity and security for autonomous agents
According to McPhee, governance is an area where SAP’s 50-year history of process focus is an advantage; the company is modernizing governance and process controls to handle the greater flexibility agents introduce. Part of this approach is reusing machine learning to validate agent behavior — customers run agents within processes and add anomaly detection and ML-based validation as guardrails, similar to SAP’s earlier intelligent approval recommendations.
Identity and permissions extend governance into execution. McPhee explained that both the human user and SAP’s Joule — the generative AI assistant embedded across SAP’s cloud applications and Business Technology Platform — must each hold rights to access a given system. Even if a user has permission to an S/4 system, Joule must also be provisioned for that access to prevent an agent from bypassing access controls.
Reconciling standard SAP with customized enterprise landscapes
Much of McPhee’s role involves reconciling SAP’s own knowledge with decades of customer customizations and non-SAP systems. He noted that many customers tell SAP, “You’re only 10% of my landscape,” and that reality has shaped SAP’s strategy. Recent acquisitions such as LeanIX — which McPhee likened to "Google Maps for your architecture" — and process-mining company Signavio are meant to map the non-SAP majority so SAP’s agents can understand system interconnections.
SAP has also invested in Berlin-based automation company n8n and is embedding it natively into Joule Studio, its intent-based, low-code environment for building agents. These tools aim to help agents comprehend enterprise system relationships and operational context.
Legacy systems and scaling risks
McPhee warned that companies need to modernize older on-premises systems or risk running into throughput and performance limitations as they expand agent use. He used a metaphor: if you want to drive a Ferrari, you should upgrade the track first — otherwise you’ll encounter constraints.
Conclusion
The VB Transform 2026 conversation underscored that enterprise AI agents deliver real value when they are grounded in company-specific knowledge (knowledge graphs, vector-embedded data) and governed with robust identity, permissions and ML-based validation. SAP’s recent acquisitions and integrations reflect this approach, but modernizing legacy systems remains essential for scalable deployment.
Disclosure
This content was presented by SAP. For sales inquiries related to VentureBeat sponsored content, contact sales@venturebeat.com.
