From Better Models to Better Context
Enterprise AI governance is the set of policies, technical controls, and accountability processes that ensure AI systems act on enterprise data in a safe, explainable, and auditable way while still delivering automation and decision support at scale. At SAP Sapphire, SAP argued that the real enterprise AI race is not about building the smartest chatbot but about giving AI agents enough reliable business context, data access, and oversight to act inside mission‑critical systems. SAP’s new Business AI Platform consolidates Business Technology Platform, Business Data Cloud, Autonomous Suite, and Business AI into a context layer grounded in existing ERP processes, data models, authorizations, and compliance rules. CTO Philipp Herzig summed up the shift when he said, “What’s not differentiating is the LLMs… Use OpenAI models, use Anthropic models, whatever you like.” For CIOs, this signals a move away from model selection and toward AI context management as a strategic discipline.
SAP’s Business AI Platform: Context as the Moat
SAP’s Business AI Platform is framed as the infrastructure that turns decades of ERP footprint into a structured context layer for AI. It brings together SAP Business Technology Platform, SAP Business Data Cloud, SAP Autonomous Suite, SAP Business AI, and tools such as Joule Work into one business AI platform that can power autonomous enterprise systems. SAP’s Autonomous Suite will include more than 50 domain‑specific Joule Assistants orchestrating over 200 specialized agents across finance, supply chain, procurement, human capital, and customer experience. Rather than compete on foundation models, SAP partners with Anthropic, Mistral AI, and Cohere, while reserving its own domain models and SAP‑RPT‑1.5 for specialized tasks where its business data and process knowledge provide an advantage. For CIOs, the message is clear: differentiation comes from how tightly AI is wired into real processes, controls, and extensions, not from owning a proprietary LLM.
Why Data Strategy Matters More Than LLM Sophistication
SAP’s stance supports a broader shift in enterprise AI from model‑centric to context‑centric thinking. The question is no longer “Which LLM is best?” but “Does the agent understand my business entities and can it reach the right data under the right permissions?” Herzig criticized teams that rely on “vibe checking” instead of rigorous evaluations, testing only a few happy‑path prompts. In response, SAP aims to have its business AI platform automatically generate scaffolding such as product requirements, technical specifications, tests, data connections, security setups, observability, and governance. This approach only works if enterprises already treat data models, process definitions, and access control as strategic assets. CIOs who still see AI as a layer on top of messy processes will struggle, while those who invest in clean reference data, shared semantics, and governed access can treat models as interchangeable commodities behind a stable context layer.
The Execution Gap: When Agents Move From Advice to Action
A second challenge appears once context‑aware agents start executing work, not only recommending it. At Sapphire, SAP positioned its 200‑plus agents, Joule Work, and Company Memory as a path to more autonomous enterprise systems. That vision raises a new question: what guarantees determinism, auditability, and control when probabilistic AI touches financial close, MRP runs, or supply chain orchestration? Redwood Software’s Chief Product Officer Charles Crouchman describes this as the execution gap, where agentic AI meets long‑standing workload automation. According to Redwood’s Crouchman, “The question customers were asking was, ‘Can AI understand my business?’ By 2025 and into SAP Sapphire 2026, the question has moved to, ‘Can AI actually execute inside my business?’” His concern is that agentic failures often do not fail loudly; plausible but inconsistent actions can flow downstream, turning supposed productivity gains into cleanup work.
Designing Governing Rails for Autonomous Enterprise Systems
For CIOs, the lesson is that AI context management must be paired with governed execution before agent decisions reach systems of record. Redwood’s roadmap shift from classic workload automation toward an agentic orchestration platform—adding MCP server support, A2A multi‑agent orchestration, and upcoming Agentic Studio and Agentic Workflows—shows how vendors are encoding timing, dependencies, approvals, and exception handling around AI‑initiated actions. In this model, LLMs propose actions while automation layers decide how and when those actions run, log every step, and stop or rewind when inconsistencies appear. Enterprise AI governance therefore spans model choice, context and data policy, and execution controls in one framework. As SAP scales its Business AI Platform and Autonomous Suite, CIOs will need similar guardrails: clear ownership for agents, testable end‑to‑end workflows, and audit trails that match existing standards for finance, procurement, and supply chain operations.
