Over the last decade or so, I’ve seen how rapidly the CIO role has expanded. What was once primarily about maintaining systems is now deeply tied to growth, transformation, and proving value.
AI has accelerated that shift. Last year, 80% of CIOs were responsible for researching and evaluating AI products, placing them at the center of enterprise AI strategy rather than just IT operations.
From my experience building enterprise software at Celonis and throughout my career, I’ve seen a consistent pattern: AI success doesn’t come from running more pilots. It comes from building the right foundation so AI can operate at scale.
As advice to CIOs today, I would focus on three priorities to ensure AI delivers ROI.
1. Understand the current state of the enterprise
Many enterprises are experiencing a “great disconnect.” Departments speak different languages, and core systems don’t communicate as they should. This disconnect threatens the promise of enterprise AI – and understanding the disconnect is the first step to fixing it.
Too many organizations try to bolt AI onto fragmented tech stacks and murky processes and expect immediate returns. That just causes frustration and disappointment. Out-of-the-box AI models lack the business context of how the enterprise truly operates — they don’t have a way to understand how your unique business actually does things.
Without a deep view into your underlying processes, AI can’t reliably answer your queries, much less operate agentically. To fix the great disconnect, understanding your business’s processes is critical.
2. Scale agentic AI safely
The gap between current AI experiments and sustained RoAI (return on AI investment) is scalability. To close it, CIOs should focus on agentic AI systems that don’t just observe or suggest, but safely execute business tasks. And that requires your business’s full context.
Building a foundation for agentic AI requires a shift in how CIOs view data. The traditional, static data lake won’t be enough. Increasingly, enterprises are building a living digital twin of their business operations.
Recent industry research highlights the urgency: 93% of process and operations leaders state that a business-wide digital twin would be a “game-changer” for their optimization efforts.
Digital twins have become foundational for scaling AI because they enable CIOs to get a real-time, unified model of how work flows across the enterprise. By mapping processes, systems, and interactions to a single, dynamic model, a digital twin gives AI the context it needs to act safely and effectively.
3. Prioritize composable architecture
Given the speed at which AI is revolutionizing tech stacks, composable architectures should be the gold standard for CIOs. Modularity and flexibility help not only with current enterprise AI ambitions, but also let CIOs help the business respond more easily to any future shifts and shocks.
With this focus on composability, it’s also crucial for enterprises to have more say in how they work with their technologies. The philosophy is straightforward: your data and processes belong to you, and should not be held captive or dictated by any rigid systems of record you operate on. We call this movement “Free the Process.“
True modularity, however, requires a shared language for enterprise data. That’s why Celonis designed the Process Intelligence Graph. It provides a common language for both people and systems — a holistic view of your company’s processes.
The 2026 year-end review
If we project ourselves to December 2026, the divide between successful and struggling enterprises will be clear.
The CIO who considers the three points above to shape their roadmap – starting by making their processes work to build an AI-driven, composable enterprise – will end the year with a fundamentally different company. They will have moved the needle from merely maintaining infrastructure to actively orchestrating growth and contributing to RoAI. Those who miss this shift may look back on 2026 as a year of mounting technical debt and missed AI potential.
I’ve seen firsthand how enterprise AI is industrialized by process intelligence. My most earnest piece of advice for any CIO in 2026 is straightforward: use your processes to fuel your AI ambitions.