AI Hub-and-Spoke Operating Models

ING's gap vs peers: The hub exists (Analytics Office + CAO) but the spokes are still forming. There is no firmwide CDAO with CEO-line authority, no unified self-service data platform, and spoke teams in business units don't yet have dedicated AI headcount on par with DBS or JPM. Transforming to "Data and AI Org" means elevating the hub to CDAO level and maturing spoke teams into business-owned squads with clear shared KPIs.

Key differentiator: The "2-in-a-box" model forces business and data to share ownership — no spoke can say "that's the analytics team's problem." Every BU has embedded data scientists reporting to both the Data Chapter hub and their business line. ADA is the single data platform all spokes draw from, creating reuse and consistent governance at scale.

Key differentiator: JPM has replicated the hub structure at the LOB level — each major business line has its own CDAO feeding into the firmwide hub. Business data ownership is mandated: "teams closest to the data should have responsibility for it." The firmwide CDAO sets strategy and standards; LOB CDAOs execute.

Key differentiator: ENBD chose a "data mesh" seed approach — a tiny central team built infrastructure and governance, then pushed data ownership to business domains. Federated learning in the wealth app demonstrates advanced privacy-preserving AI. Still growing spoke headcount and AI maturity vs DBS/JPM.

Key differentiator: BofA bet on one great NLU engine (Erica) and reused it everywhere. The same patent-protected model powers consumer, wealth, private banking, and internal ops. This maximises ROI per AI dollar but can reduce spoke agility — the flip side of the DBS/JPM distributed model.

Key differentiator: Radical structural transformation. While competitors plug AI into existing systems, Klarna tore out Salesforce and Workday entirely, channeling operational data into a unified Neo4j Knowledge Graph. The 2025 pivot back to a "hybrid" support model represents an important lesson: ultimate efficiency requires human empathy as a safety guardrail.

Key differentiator: Symbiotic peer partnership. Instead of a detached CoE, the Enterprise AI and CDO organizations work hand-in-hand with lines of business. Combined with a "data-as-a-product" framework and strict compliance safeguards, Capital One's "AI Factory" shows how a highly regulated major bank can achieve digital agility safely.

Key differentiator: ABN AMRO's compliance burden became an AI accelerant — the regulatory requirement to fix AML at scale forced rapid ML adoption in financial crime, building institutional AI capability that now extends into commercial and retail banking. Conservative Dutch governance culture means slower agentic deployment vs neobanks, but higher trust in regulated domains.

Key differentiator: Rabobank's mission as a cooperative agricultural bank created a unique AI dataset that is impossible to commoditise — farm-level climate, crop yield, and financing data going back decades across 40+ countries. AI built on this dataset (farm advisory agents, climate risk scoring) is a genuine competitive moat. The cooperative governance model means AI adoption must be earned, not mandated — creating slower but stickier deployment across the network.