Context

The organisation had established a central AI function to explore use cases, build early capability, and prove value through pilots. Over time, several of these initiatives demonstrated enough credibility to move beyond experimentation. Business teams wanted to scale and embed AI into their own operations, while the central team remained involved out of necessity rather than design. AI systems were no longer new, but the operating model had not yet adapted to this new stage of maturity.

‍

The Challenge

Success created its own tension. Central teams were not sized or structured to own an expanding portfolio of live AI systems, yet business teams were hesitant to take responsibility for systems they had not initially built. Keeping AI centrally owned made it easier to maintain consistency, but it also slowed decision making and limited adoption. Moving ownership too quickly risked fragmenting standards and recreating the very issues centralisation had been meant to address. The challenge was to move execution closer to the business without dissolving accountability or coherence.

‍

The Decision

The organisation decided that once AI initiatives moved beyond pilots, ownership would shift explicitly to the business functions using them. The central AI team’s role changed from builder and owner to enabler and steward. Rather than continuing to run systems by default, they focused on defining expectations, supporting transitions, and maintaining shared principles. The alternative-retaining ongoing ownership with central teams, or abandoning central involvement altogether-was consciously avoided in favour of a clearer handover model.

‍

What Changed

Business teams became accountable for AI systems that directly affected their operations, including decisions about how those systems were maintained or evolved. Central teams were able to step back from day to day execution and focus on enabling reuse, consistency, and learning across the organisation. Some transitions slowed delivery as ownership was clarified, but fewer systems remained stuck in a semi central, semi business state. AI became part of normal business execution, rather than an extension of a central programme.

‍

Why This Matters

Many enterprises struggle to evolve their AI operating model after early success. Keeping ownership central for too long limits scale, while pushing it out too quickly creates fragmentation. Explicitly deciding when and how ownership shifts allows AI to mature as a business capability rather than remaining dependent on specialist teams. The operating model, not the technology, determines whether early momentum can be sustained.