Enterprise AI Agents Need Context, Not Just Models
At VB Transform 2026, SAP's Max McPhee outlined how enterprises can move beyond chatbots to autonomous AI agents by grounding them in company-specific context through knowledge graphs and governance controls. The key difference between assistants and true agents lies in providing enterprise context rather than relying on general knowledge, combined with identity and permission controls that prevent agents from circumventing access restrictions. SAP's recent acquisitions of LeanIX and Signavio, plus investment in n8n, are designed to help agents navigate complex, multi-system enterprise landscapes where SAP represents only a portion of the technology stack.
TL;DR
- Enterprise AI agents require knowledge graphs and vector-embedded data to understand company-specific context and internal terminology, not just general knowledge
- Governance must evolve to handle agent flexibility, including machine learning-based anomaly detection and dual identity/permission requirements for both user and agent
- SAP's strategy addresses the reality that customers tell the company it represents only 10% of their IT landscape, requiring integration with non-SAP systems
- Older on-premises systems may create throughput limitations as enterprises scale autonomous agent deployments
Why It Matters
The gap between chatbots and true autonomous agents is widening as enterprises demand systems that can execute real business processes. Grounding agents in enterprise context through knowledge graphs and governance controls is becoming table stakes for vendors competing in this space. This shift signals that enterprise AI maturity depends less on model capability and more on integration, governance, and understanding of customer-specific operations.
Business Impact
Enterprises investing in AI agents need to understand that off-the-shelf models alone will not deliver business value without proper grounding in company context and governance frameworks. Organizations with fragmented IT landscapes, common in large enterprises, face additional complexity in enabling agents to operate across multiple systems safely. The emphasis on dual identity controls and anomaly detection suggests that security and compliance concerns are central to agent deployment, not afterthoughts.
Key Implications
- Knowledge graph and vector database infrastructure is becoming a prerequisite for enterprise AI agent deployment, not an optional enhancement
- Governance and security controls must be designed into agent systems from the start, with machine learning-based validation becoming a standard guardrail alongside traditional access controls
- Enterprises with significant non-SAP or legacy on-premises infrastructure will need additional integration work and potentially system modernization to support agent scalability
- Vendors are consolidating capabilities through acquisition and embedding, with SAP using LeanIX, Signavio, and n8n to address the multi-system enterprise reality
What to Watch
Monitor how enterprises implement knowledge graphs and governance frameworks in their agent deployments, and whether throughput or integration issues emerge as adoption scales. Watch for competitive responses from other enterprise software vendors to SAP's acquisition strategy and native embedding of integration tools. Track whether machine learning-based anomaly detection becomes a standard practice or remains a differentiator in agent governance.
Subscribe to the newsletter
The latest stories and analysis, delivered to your inbox.
Free. No spam. Unsubscribe any time.
