Scaling requires coordinated organisational capability
The World Economic Forum’s 2026 AI Playbook for Financial Services, based on input from more than 150 senior leaders across 100 institutions, focuses on strategy, governance, workforce transformation, data foundations and agentic AI as connected elements of scaled adoption (WEF, 2026).
This is significant because it shifts the centre of gravity from isolated technical teams toward an enterprise operating model. The question becomes who owns the capability, who decides risk, how shared assets are funded and how domain teams consume them.
The best operating model reduces duplicate work
A bank can build a separate retrieval pipeline, evaluation process and governance workflow for every business unit, but that reproduces cost and risk. A stronger model provides reusable components—identity, model gateways, document services, knowledge access, evaluation, observability and controls—that domain teams can combine for specific outcomes.
Centralisation alone is not the answer. Domain accountability still matters because risk, evidence and value are context-dependent. The operating model has to combine shared technical capability with business ownership.
- 01Business domain
- 02AI product ownership
- 03Shared architecture
- 04Risk & governance
- 05Data / knowledge
- 06Measurement
The operating model connects reusable enterprise capability to domain outcomes.
Regulation reinforces the need for joined-up ownership
The EBA’s June 2026 risk assessment links AI use to operational resilience, data security, cyber risk, third-party dependence and DORA obligations. Those concerns cross traditional organisational boundaries, making purely technology-owned AI difficult to sustain (EBA, 2026).
An effective operating model therefore connects product, architecture, security, risk, legal/compliance, data and business operations early in the lifecycle rather than using risk functions only as final approval gates.
Arvec view: organise around capability and decisions
The useful organisational unit is not 'the AI team'. It is the set of capabilities and decision rights required to move an objective from idea to controlled production.
AIF7™ makes those capability dependencies explicit. The target operating model can then allocate ownership to each capability, define interfaces between teams and identify which assets should be enterprise-wide versus domain-specific.
Key implications
- 01
Fund reusable AI capabilities separately from individual use cases.
- 02
Keep business ownership close to the domain decision and evidence.
- 03
Bring risk, security and architecture into delivery before production approval.
- 04
Use the operating model to make ownership and interfaces explicit.
Related AIF7™ capabilities
AI Organisation & Operating Model
Defines accountability, funding and decision rights.
AI Engineering & Delivery
Provides reusable technical services across domains.
AI Governance & Trust
Embeds control functions into the delivery lifecycle.
AI Value Realisation
Keeps capability investment tied to measurable domain outcomes.
Sources & further reading
01 · WEF 2026
The AI Playbook for Financial ServicesWorld Economic Forum · 24 June 2026
02 · EBA 2026
Risk Assessment Report – June 2026European Banking Authority · June 2026