Stephen Messer, co-founder of Collective[i] and Intelligence.com, argues that most firms treat artificial intelligence as a decorative layer rather than a catalyst for redefining what a company is. He contends that the real challenge is not to ask how AI can speed up existing tasks, but to ask which tasks should disappear altogether.
Why many AI projects fall short
Companies often boast about AI licences, pilot programmes, chief AI officers and responsible-innovation slide decks. Yet the underlying workflows, salespeople entering data into CRM systems, managers shuffling information between teams, and customers waiting for approval chains, remain unchanged. This superficial upgrade, which Messer calls the "AI Shuffle", creates the illusion of progress without delivering a competitive edge.
From addition to subtraction
The conventional response to new technology is to add more tools, dashboards and governance layers. Messer suggests the opposite: start by questioning every existing requirement, strip away unnecessary steps and only then automate the streamlined process. Automating a flawed process merely makes it faster and harder to detect inefficiencies.
Practical example: sales forecasting
Traditional sales forecasting relies on sellers inputting projections into a CRM, managers interpreting the data, and leadership negotiating a final number in a ritualised meeting. The data is often late, incomplete and distorted by incentives. An AI-first approach would replace the meeting with a system that analyses real buyer behaviour, market signals and relationship dynamics, removing the need for the outdated ritual entirely.
Implications for organisational structures
Beyond software, AI threatens the layers built around information gathering and decision-making. Leaders who add value will be those who understand business problems, can apply technology effectively and stay close to customers. Roles that exist mainly to preserve friction, control information flow or manage legacy processes risk becoming irrelevant.
The lasting competitive moat
While model performance and training size dominate headlines, Messer argues that sustainable advantage will come from the orchestration layer, the systems that decide which model to use, retain context, connect to proprietary data and learn from real outcomes. This learning system, not the model itself, will lock in value.
What this means for European firms
For companies across Europe, the choice is clear: use AI to reinforce existing hierarchies and data silos, or rebuild the organisation to act faster, learn continuously and serve customers directly. The former will generate headlines; the latter will widen the gap between firms merely adopting AI and those truly transformed by it. The window for decisive action is open now, but it will not stay open indefinitely.