Core and Flex: The Workforce Model Taking Shape in AI-Led GCCs
Workforce planning in GCCs used to revolve around a single number: how many people in the center would need. In AI-led centers, the more consequential question is a different one, which kinds of roles the center needs, and on what terms. Across new and transforming centers alike, the answer taking shape is a two-layer model: a durable core, and a flexible layer around it.
The core: roles built to last
The core is the talent a center expects to hold regardless of how far AI advances. It has two parts. The first is the new job families AI itself is creating: orchestration roles, context engineering, and the work of directing agents toward business problems. The second is the set of roles where human judgment remains structural to the work, whether that means a human in the lead, a human in the loop, or people supervising and being supported by agents.
What unites these roles is context. Core talent carries business knowledge, institutional memory and judgment that make AI effective inside a specific enterprise. This is also why the core deserves the center’s deepest investment in AI fluency: these are the people who will determine how much value the technology produces.
The flex: capacity without permanence
Around the core sits a flexible layer built for continuous evolution.Nobody can yet say precisely which work AI will absorb, or when. The flex layer answers that honestly: it delivers the work that may be automated over time through contingent talent and services partnerships rather than permanent hiring.
Logic is as humane as it is commercial. When a piece of work is absorbed by AI, the center scales its partner’s capacity down instead of letting its own people go. Flexibility is built into the structure of the workforce rather than extracted from it later, painfully, one difficult decision at a time.
The ratio is shifting
What makes this model striking is how far from the ratio has moved. In the newest AI-native setups, the core can be as small as twenty to thirty percent of the workforce, with the flex layer forming the majority. These centers are deliberately starting smaller than their business cases might justify, on a simple asymmetry: expanding a team is easy; contracting one is not. If the experiments underperform, more hiring is always available. The reverse offers no such comfort.
Partnerships make the model work
A flex layer is only as good as the partners behind it. This is drawing GCCs and service providers into a far closer working relationship than the industry has been used to, with providers supplying quality talent at short notice, absorbing demand peaks, and supporting delivery alongside the center’s own teams. The old separation between the two sides is dissolving, and the centers getting the most from the model are the ones choosing partners on the quality of talent they can field, not on commercial terms alone.
Designing the workforce in layers
For leaders, practical work is defining the boundary. Which roles genuinely belong in the core because they carry context and judgment the enterprise cannot rent? Which work belongs in the flex layer because its future is uncertain? Defaulting everything into permanent roles recreates the rigidity this model exists to avoid; pushing too much into flex hollows out the context that makes AI effective. The boundary is a leadership decision, and it deserves to be made deliberately rather than inherited from an old org chart.
The centers built this way hold a quiet advantage: they can move with the technology instead of bracing against it. In a period where only certainty is further changed, a workforce designed in layers is not a compromise. It is a strategy.
