Back in 2001, Ray Kurzweil, the American computer scientist, author, and inventor wrote: “the future will be far more surprising than most observers realise: few have truly internalised the implications of the fact that the rate of change itself is accelerating.”
Boy was he right. It seems like only yesterday that the first, rudimentary AI chatbots hit the internet. Today, we find ourselves on a cusp of a new age of autonomous, agentic AI systems capable of making decisions and taking action without human interaction. According to Gartner, 40% of enterprise apps will feature task-specific AI agents this year, up from less than 5% in 2025. Change is accelerating fast.
The visible and hidden faces of AI
The implications for businesses are profound. AI is becoming visible at the top of the stack. Systems like SAP Joule are acting like virtual concierges for enterprise applications, enabling user access to the hundreds of agents and assistants that comprise modern enterprise software stacks. For users, things have never been easier. They simply describe an outcome and let agents orchestrate behind the scenes.
So much for the visible layer of agentic AI. What’s less well known is that there’s also an invisible layer that plays a vital role in ensuring optimum outcomes.
Agents are only as strong as the foundation on which they are built. For agents to act safely, enterprise systems need to be healthy, observable, and consistent. Agents also require clean process telemetry, automated remediation, and governance that extends across all infrastructure. Without this foundation, agentic AI succeeds only in introducing risk.
Future-proofing the enterprise stack therefore comes down to building a robust operational layer that makes AI agents a source of measurable business value. Over the next few years, we’re likely to see two different types of business emerge: those that put in place the operational platform for AI agents to thrive, and those that waste months of effort failing in the attempt.
A practical approach to ERP transformation
Enterprise leaders are primed to make the shift to agentic AI. According to the SAPinsider 2026 research study, 43% of organisations cite AI readiness as the primary driver of their transformation investment.
For many large enterprises, the path to AI is a marathon, not a sprint. There’s a marked preference for incremental change over wholesale reinvention.
In many cases, brownfield migration is seen as the ideal starting point. This approach enables businesses to shift existing systems onto modern platforms while preserving established processes and minimising disruption. The stability and continuity this approach provides is especially useful in complex landscapes with extensive integrations and dependencies, where even a small disruption can escalate rapidly.
Taking a brownfield approach, enterprises can move forward in a structured way, stabilising their core systems before introducing agentic AI. The migration to cloud ERP is a key element of this move. Managed, scalable ERP systems provide a platform that supports both current operations and future capabilities, delivering continuous updates and making it easier to integrate new services.
Putting in place the right cloud ERP platform is particularly important for AI. With intelligence becoming embedded in enterprise applications, cloud platforms provide the infrastructure needed to support them at scale. From advanced analytics to autonomous execution, AI capabilities are being delivered as part of the platform rather than as separate tools.
Of course, as organisations shift to the cloud there will be a transition period where they operate both on-premise and cloud-based systems in tandem. These hybrid environments add complexity in areas such as governance, monitoring, and integration. To overcome these challenges, organisations should clearly define roles and responsibilities, and introduce an operational layer that’s observable, automatable, and consistent across the hybrid environment.
Readiness at two levels
Agentic AI operates at two levels – an interaction layer and an execution layer – and both have to work. At the visible, interactive layer, AI is becoming the front door to the enterprise stack, enabling people to interact with enterprise systems naturally, using conversation to describe outcomes.
Meanwhile, hidden from view, AI is also becoming integral to the underlying infrastructure. We can expect to see AI agents find their way into areas including observability, automated remediation, capacity and performance management, and in operational disciplines. This invisible layer will directly influence how the visible layer performs.
Clean and timely intelligence is essential. Agents will need access to the highest quality operational data and process telemetry to understand the live context of the IT stack at any given time and to act appropriately. Agents fueled by incomplete or stale data will provide confident answers that will likely be completely wrong. As always with IT, garbage in means garbage out. Conversely, agents that act on a well-instrumented, automated estate will deliver optimal outcomes.
Operational readiness is a competitive advantage
Agentic AI will be one of the most disruptive technologies the world has ever seen. Businesses should therefore take their time planning for its arrival. That means building the strongest foundations so that continuous improvement can take place safely. Core systems need to be modernised, but change should be incremental. Emerging opportunities absolutely should be integrated into operations, but this should happen without introducing additional risk.
Success means preparing across both the visible and hidden layers of AI; across interaction and execution. The North Star is an enterprise consisting of intelligent operations that deliver measurable business value with AI enabling outcomes at every level of the stack. Any business leader contemplating what will define long-term competitiveness now has an answer: resilience, adaptability, and operational discipline.
