Skip to content
Services/AI Operating Model Design

AI Operating Model Design

Strategy without an operating model is a slide deck.

McKinsey's Global AI Survey (2024) identifies operating model and organisational design as the #1 barrier to AI scaling, cited by 57% of executives. Enterprises have the technology. They have the use cases. They lack the organisational architecture to deliver and sustain value.

BCG research (2024) finds enterprises with mature AI operating models are 6× more likely to achieve enterprise-scale AI deployment. The operating model is where strategy meets execution.

57%

cite operating model as the #1 AI scaling barrier

McKinsey, 2024

more likely to scale AI with a mature operating model

BCG, 2024

faster use case deployment with AI translator capability

McKinsey, 2024

70%

of successful AI scalers use a hub-and-spoke model

McKinsey, 2024

Three operating model archetypes

Not every operating model is right for every enterprise at every stage. We design the model appropriate to your current maturity, regulatory environment, and scale ambition

Centralised (AI CoE)

Advantages

Consistent standards and governance

Efficient platform and tooling investment

Clear accountability

Considerations

Creates bottlenecks as demand scales

Business unit adoption is slower

Risk of being perceived as a cost centre

Enterprises building foundational capability. Stage 1–2 of the Gartner AI Maturity Model

Hub-and-Spoke (Hybrid)

Recommended

Advantages

Governance and standards from the centre

Execution speed at business unit level

Scales efficiently as use case volume grows

Considerations

Requires careful governance design to avoid fragmentation

Demands strong AI translator capability in business units

The target-state model for most large enterprises. Used by 70% of AI-mature organisations (McKinsey, 2024)

Federated (Embedded)

Advantages

Maximum business unit autonomy and speed

AI closely aligned to specific domain needs

Considerations

Governance fragmentation

Duplication of platforms and tooling

Inconsistent standards across the enterprise

Enterprises at Stage 4–5 of the Gartner AI Maturity Model with strong existing governance infrastructure

How we design your AI operating model

Five stages from diagnostic to transition-ready operating model design. Every stage produces documented deliverables your organisation owns

Stage 01

Operating Model Diagnostic

We assess your current AI operating model against the Deloitte four-stage maturity curve (Ad Hoc → Structured → Integrated → Optimised). We map your existing roles, governance structures, data ownership, and decision rights — identifying where your model is creating friction and where it is leaving value on the table

Operating model maturity assessmentRole and accountability gap analysisDecision rights mapping

Stage 02

Target State Architecture Design

We design your target-state AI operating model. For most large enterprises, this is a hub-and-spoke model: a central AI Centre of Excellence governs standards, platforms, and enterprise risk, while business units own use case execution. McKinsey data confirms 70% of enterprises that successfully scaled AI operate a hub-and-spoke model by the time they reach maturity

Target state operating model designCoE mandate and charterBusiness unit AI roles framework

Stage 03

Talent Architecture

We define the roles your AI operating model requires — from data engineers and ML engineers to the 'AI translators' McKinsey identifies as the most critical and scarce capability in enterprise AI. Enterprises with strong translator capacity (people who bridge business problems and AI solutions) deploy use cases 3× faster. We design the hiring framework, capability development pathways, and partnership model your organisation needs

AI roles frameworkCapability development pathwaysBuild-buy-partner analysis

Stage 04

Process & Decision Rights Design

We design the processes, policies, and decision rights that allow AI to be deployed and governed at speed — without creating new compliance or operational risk. This includes the intake process for new AI use cases, the model risk management process, the escalation and incident management protocol, and the ongoing performance monitoring regime

AI intake and prioritisation processModel risk management policyPerformance monitoring framework

Stage 05

Transition Planning & Enablement

We design the transition path from your current state to your target operating model — with realistic sequencing, change management integration, and the internal capability-building programme required to sustain the model without ongoing external dependency

Transition roadmapChange management planInternal capability programme

Build the organisational architecture that makes strategy executable

Most enterprises design their AI operating model as an afterthought — after the technology is already deployed. We design it before deployment, so your AI programme scales rather than stalls.

or

See also: AI Strategy · Change Management · AI Governance