SAP is betting that agentic AI, specifically its Joule copilot and Autonomous Suite, can make multi-year ERP migrations faster and cheaper, claiming a 35% reduction in migration effort through what it calls an agent-led transformation approach. At SAP Connect Day 2026 in Johannesburg, the company also projected AI will handle 48% of average business tasks within two years, up from 30% today. Standard Bank’s group CIO Vanessa Padiachee offered the sharpest ground-level testimony, confirming that large-scale ERP modernisation<\/a> hinges less on the technology than on data quality, change management, and starting foundational work earlier than feels necessary.<\/p>\n If your organisation is still running on-premise SAP or managing end-of-life ERP infrastructure, this story is directly about you. The 35% effort-reduction claim is vendor-supplied and deserves scrutiny, but the direction is credible: agentic tools that handle custom code migration, test management, and configuration work do compress the manual labour that makes these programmes so expensive and slow. The organisations most exposed are those who’ve deferred migration, because the gap between their current platform and where SAP’s AI capabilities live is widening every quarter.<\/p>\n The Standard Bank case is the most instructive thing in this article, and not for the reasons SAP would highlight. Padiachee’s warning about Unicode conversion, a deeply technical data-encoding standardisation step required before modern SAP systems can run cleanly, is a quiet alarm for any CIO who’s underestimated foundational readiness. The recurring failure mode in large ERP programmes isn’t the technology selection or even the change management; it’s the discovery, mid-programme, of a dependency that should have been a prerequisite. Her framing, that foundational technical readiness reduces downstream project risk, is the closest thing to a hard lesson this article contains.<\/p>\n SAP’s 30-to-48% task-automation projection deserves a skeptical read. The company, pitching its own platform to customers already in its ecosystem, has every incentive to set an optimistic ceiling that makes the upgrade decision feel urgent. What the number actually signals is the architectural direction: SAP is positioning Joule and the Autonomous Suite as the orchestration layer sitting above every business process, which means the strategic question for any CIO on a renewal or expansion decision isn’t whether agentic AI is real, it’s whether SAP is the right place to consolidate that layer, or whether a more composable architecture keeps more options open.<\/p>\n Agentic AI refers to AI systems that don’t just answer a question but pursue a goal over multiple steps, checking results and adjusting as they go, the way a junior analyst might work through a task rather than just hand you a single answer. In an ERP context, that means an agent can run a migration script, review what broke, and try a correction without a human approving each step. The business implication is fewer human touchpoints per task, but also fewer human checkpoints per mistake.<\/p>\nWhat this means for your business<\/h2>\n
Concept deep-dive: Agentic AI<\/h2>\n