Enterprise AI is a Context and Governance Problem
Enterprise AI governance describes how organisations control, audit, and secure AI systems so that automated decisions and actions stay aligned with business rules, data policies, and accountability standards across complex processes. SAP’s latest messaging puts this view at the centre of its AI strategy. At SAP Sapphire in Orlando, the company argued that the enterprise race will not be won by whoever builds the smartest chatbot, but by whoever can give AI agents rich business context, reliable data access, and enforceable controls. This “business context AI” approach builds on ERP foundations: process models, authorisations, compliance rules, and customer-specific extensions. SAP is packaging these assets into the SAP Business AI Platform, which unifies Business Technology Platform, Business Data Cloud, the Autonomous Suite, and tools such as Joule Work. The goal is for agents to operate inside systems of record with traceable decisions, not as isolated assistants.
Models Are Commodity; Context Is the Differentiator
SAP leaders are blunt that large language models are no longer the main source of differentiation. CTO Philipp Herzig said, “What’s not differentiating is the LLMs… Use OpenAI models, use Anthropic models, whatever you like.” The company’s view is that enterprise value comes from everything wrapped around the model: shared data schemas, domain models, security, and governance. SAP partners with Anthropic, Mistral AI, and Cohere for general-purpose models while building SAP Domain Models and SAP-RPT-1.5 where its own domain expertise and data can add an edge. This reflects how business context AI changes buying criteria. Instead of asking which model is smartest, enterprise buyers ask whether an AI platform understands their finance, supply chain, or HR entities, respects role-based access, and can be tested on real workloads rather than a small set of “vibe-check” prompts.
The Autonomous Enterprise and Agentic AI Control
SAP’s Autonomous Suite pushes beyond copilots toward AI that can act across core processes. SAP plans more than 50 domain-specific Joule Assistants orchestrating over 200 specialised agents across finance, supply chain, procurement, HR, and customer experience. Joule Work is designed as the single entry point into these workflows, spanning SAP and non-SAP systems. This raises hard questions about agentic AI control. Business leaders must understand who approves actions, how agent behaviour is audited, and where accountability sits when an agent executes across multiple applications. Enterprise AI governance here means defining guardrails before agents reach systems of record: scoped permissions, change logs, evaluation pipelines, and rollback paths. Without such controls, a colourful demo of autonomous workflows can hide the risk that agents might trigger changes that are difficult to trace, reverse, or explain to auditors and regulators later.
Execution: Where Autonomous Enterprise Challenges Get Real
While SAP solidifies the intelligence and orchestration layers, partners like Redwood Software focus on the execution layer, where autonomous enterprise challenges come into sharp relief. Chief Product Officer Charles Crouchman notes that customer questions have shifted from “Can AI understand my business?” to “Can AI actually execute inside my business?” Redwood’s background in workload automation for financial close, MRP runs, billing cycles, and supply chain orchestration shows why. These processes involve thousands of interdependent steps, strict timing, approvals, and audit trails. Probabilistic agents can introduce subtle inconsistencies that do not fail loudly; they propagate downstream until remediation work cancels out earlier productivity gains. Redwood is extending its platform toward full agentic orchestration, with MCP server support and multi-agent orchestration, to keep deterministic control over how AI-triggered actions run. This is where enterprise AI governance must prove it can handle real production risk.
