At the VB Transform 2026 event, SAP shared in-depth insights on transitioning from conventional chatbots to autonomous AI agents capable of executing real business processes. According to Max McPhee, senior solution advisor at SAP, the key lies in grounding these agents in actual enterprise context rather than general knowledge. He likened this onboarding process to preparing a new employee but optimized for AI.
Background & Causes
The biggest difference between a standard AI assistant and a true AI "coworker" is the capacity to comprehend the internal context of an enterprise. Most current chatbots still stumble over corporate acronyms or organization-specific workflows, creating a major barrier to deploying AI for complex automated tasks. To resolve this, organizations must provide structured data resources that AI can accurately retrieve and process. According to SAP, modernizing legacy infrastructure is a prerequisite before adopting advanced AI models.
Technical Analysis & Technology
Technically, SAP proposes combining knowledge graphs with vector-embedded data. This combination allows AI agents to easily locate, retrieve information, and comprehend "tribal knowledge" — implicit understanding that is often not officially documented. Additionally, SAP integrates its Joule assistant across its cloud applications and Business Technology Platform with a dual-permission mechanism. Specifically, both the user and Joule must be granted access rights individually, preventing risks of bypassing security controls. The company has also invested in n8n to natively embed it into Joule Studio for building intent-based agents.
Expert Opinions & Insights
Max McPhee emphasized that process governance has been SAP's core strength for 50 years, which now must be modernized to accommodate the flexibility of autonomous agents. He noted that this trend is driving "a bit of a revival of machine learning," as customers utilize ML for anomaly detection and validation as guardrails. At the same time, the SAP representative warned that running modern AI agents on legacy on-premises systems is like "trying to drive a Ferrari around a dirt track," inevitably leading to throughput bottlenecks and poor performance.
Impact & Future
The shift toward autonomous AI agents requires enterprises to re-evaluate their entire system architecture mapping. SAP's acquisitions of companies like LeanIX (described as "Google Maps for your architecture") and Signavio reflect its effort to bridge non-SAP systems, which often dominate customer landscapes. For Vietnamese businesses undergoing digital transformation, SAP's insights demonstrate that investing in data cleansing and constructing internal knowledge graphs are prerequisite steps before attempting to deploy genuine "AI coworkers."