Créer le Centre d'Excellence IA

Ce qu'elle couvre

Ce programme dote les responsables de la transformation et les directeurs de la data des cadres stratégiques et opérationnels nécessaires pour concevoir, lancer et pérenniser un Centre d'Excellence IA. Les participants explorent les modèles hub-and-spoke et fédérés, construisent une feuille de route de recrutement, rédigent une charte de CoE et définissent un rythme opérationnel ancré dans les résultats business. Les sessions combinent analyse de cas, ateliers en groupe de travail et revue par les pairs pour produire des livrables présentables à la direction dès la fin du programme.

À l'issue, vous saurez

• Design and justify a CoE operating model (hub-and-spoke, federated, or hybrid) suited to your organisation's structure and AI maturity
• Draft a board-ready CoE charter defining mandate, scope, governance structure, and success metrics
• Build a phased hiring plan with role priorities, sourcing strategies, and retention levers for AI talent
• Define a funding model and internal chargeback mechanism that keeps the CoE financially sustainable beyond year one
• Establish an operating rhythm—including OKRs, steering committee cadence, and escalation paths—that embeds AI delivery into business-as-usual

Sujets abordés

• Hub-and-spoke vs federated vs centralised CoE operating models
• Writing and ratifying a CoE charter and mandate
• Hiring patterns: roles, sequencing, and build-vs-buy decisions
• Funding models: cost centre, profit centre, and internal chargeback
• Governance frameworks: AI ethics board, review gates, and policy layers
• Operating rhythm: OKRs, sprint cadences, and executive reporting
• Stakeholder alignment and change management across business units
• Measuring CoE ROI and demonstrating value to the board

Modalité

Typically delivered as a blended programme over 4-6 weeks: two in-person or virtual intensive days (kick-off and finale) bookending three to four facilitated online working sessions of 3-4 hours each. Participants work in cross-functional cohorts to build their own CoE blueprint in parallel with instruction. Materials include a CoE charter template, a RACI model for AI governance, a hiring-matrix workbook, and a funding-model calculator. Hands-on working time accounts for approximately 60% of total hours.

Ce qui fait que ça marche

• Securing a named C-suite sponsor with budget authority before the CoE charter is written
• Embedding CoE liaisons inside high-priority business units from day one to maintain trust and adoption
• Publishing a transparent intake and prioritisation process so business units see the CoE as a partner, not a gatekeeper
• Running quarterly value reviews tied to board-level OKRs to sustain funding and political capital

Erreurs fréquentes

• Centralising all AI work in the CoE and creating a bottleneck rather than enabling distributed capability
• Launching without a clear mandate or executive sponsor, leaving the CoE to fight for budget every quarter
• Hiring senior data scientists first instead of engineering and MLOps roles that unblock delivery
• Measuring the CoE on project output rather than on business value delivered to internal customers

Fournisseurs à considérer

• DataScientestdatascientest.com →
• Wavestonewww.wavestone.com →
• McKinsey & Company (QuantumBlack Academy)www.mckinsey.com/capabilities/quantumblack/how-we-help-clients/capability-building →
• MIT Sloan Executive Educationexecutive.mit.edu/course/artificial-intelligence-implications-for-business-strategy/ →

Sources

• Building an AI Center of Excellence — Gartner Research →
• How to Set Up an AI Center of Excellence — Harvard Business Review →