Key Takeaways
- Consumer familiarity with AI tools raises expectations for enterprise‑grade AI deployments.
- Many organisations fall into “pilot purgatory,” launching numerous proofs of concept without clear direction or measurable outcomes.
- Allied Irish Banks avoided this trap by being highly intentional about the scope, purpose, and location of its AI initiatives.
- An AI Centre of Excellence (CoE) was created to coordinate testing, share best practices, and maintain governance while encouraging adoption.
- The CoE acts as a guide rather than a gatekeeper, balancing control with flexibility to accelerate value‑realisation.
- Early results show improved alignment between business needs and technology capabilities, reduced duplication of effort, and faster transition from experiment to production.
- Ongoing success depends on continuous learning, clear success metrics, and adapting the CoE model as AI technologies evolve.
Introduction: Consumerization Drives Enterprise AI Expectations
The rapid spread of artificial intelligence in everyday life—voice assistants, recommendation engines, generative chatbots on smartphones—has created a baseline familiarity among employees. When workers encounter AI at home, they naturally expect similar ease of use, speed, and relevance in the workplace. This “consumerisation” of AI shifts the conversation from mere technical feasibility to user experience and organisational readiness. Enterprises can no longer treat AI as a back‑office experiment; they must anticipate and manage the pre‑formed opinions and expectations that staff bring to any new AI‑powered tool. Ignoring this dimension risks low adoption, frustration, and missed opportunities to leverage AI’s productivity gains.
The Challenge of Pilot Purgatory
Many organisations fall into a few ways to revolution of been of AI is not a lack of ideas but a lack of focus: teams launch dozens of proofs of concept (PoCs) driven by enthusiasm rather than strategy. These pilots often linger in a state of “pilot purgatory,” consuming resources without delivering measurable business value or clear pathways to scale. The phenomenon is exacerbated when enthusiasm outpaces governance, leading to duplicated effort, conflicting data standards, and difficulty consolidating learnings into a coherent AI roadmap.
Intentionality in AI Initiatives
Allied Irish Banks recognised the pitfalls of unfettered experimentation and chose a different path. As Graham noted, the bank “hasn’t spun up loads of proofs of concept.” Instead, it adopted a deliberate approach, carefully selecting which problems to tackle, where to apply AI, and what success would look like. This intentionality meant defining clear business objectives upfront, aligning AI projects with strategic priorities, and allocating resources only to initiatives that had a vetted business case and measurable success criteria. By resisting the urge to chase every shiny AI use‑case, the bank conserved effort and increased the likelihood of meaningful outcomes.
Managing Expectations from Personal AI Use
Because employees already interact with AI in their personal lives, they bring a set of expectations about responsiveness, intuitiveness, and immediate utility to workplace tools. Graham highlighted that “any technology where there’s been consumerisation over the years means enterprises have to be mindful that colleagues will already have their own points of view about the tech.” To address this, Allied Irish Banks instituted regular feedback loops, user‑experience testing, and change‑management programmes that bridge the gap between consumer‑grade expectations and enterprise‑grade realities such as data security, regulatory compliance, and integration with legacy systems. Managing these expectations early reduces resistance and fosters a culture where AI is seen as an enabler rather than a disruptive imposition.
Establishing an AI Centre of Excellence
To translate intentionality into operational capability, Allied Irish Banks set up a dedicated Centre of Excellence (CoE) for AI. The CoE’s mandate is to coordinate AI testing, validation, and deployment across the organisation, ensuring that initiatives adhere to architectural standards, data governance policies, and risk‑management frameworks. Rather than acting as a restrictive gatekeeper, the CoE functions as a hub of expertise, providing guidance, reusable assets, and best‑practice templates that empower business units to move faster while staying within guardrails. This structure helps the bank avoid the fragmentation that often accompanies decentralized AI experiments.
Scope and Functions of the Centre of Excellence
The CoE’s activities span the full AI lifecycle: ideation, feasibility assessment, model development, testing, integration, monitoring, and retirement. It maintains a catalogue of approved AI tools, frameworks, and libraries, ensuring that teams do not reinvent the wheel or inadvertently introduce security vulnerabilities. Additionally, the CoE runs a sandbox environment where new techniques—such as code refactoring with large‑language models or predictive analytics for credit risk—can be evaluated safely before broader rollout. Knowledge‑sharing forums, internal workshops, and a community of practice further diffuse learnings, helping the organisation climb the learning curve collectively.
Balancing Guidance and Control
A critical design principle of the CoE is to act as a guide rather than a barrier. Graham emphasized that the centre “functions as a guide rather than a barrier to technology adoption.” This nuance is achieved by offering optional standards, consulting services, and pre‑built pipelines that teams can adopt if they wish, while still allowing them to pursue bespoke solutions when justified. Governance is embedded through lightweight review checkpoints rather than heavyweight approval gates, preserving agility. The result is a culture where innovation is encouraged, but decisions are informed by shared expertise and aligned with organisational risk tolerance.
Outcomes and Lessons Learned
Since its inception, the AI CoE has helped Allied Irish Banks move from scattered experiments to a more coherent portfolio of AI‑driven initiatives. Reported benefits include reduced duplication of effort, faster time‑to‑value for pilots that graduate to production, and clearer visibility into AI spend and performance. Teams have cited the CoE’s reusable components and expert advice as key accelerators, especially for complex use‑cases like anti‑money‑laundering detection and intelligent document processing. Lessons learned emphasize the importance of defining success metrics early, maintaining a tight feedback loop with end‑users, and treating the CoE as a living entity that evolves alongside technological advances.
Future Directions for Enterprise AI Adoption
Looking ahead, Allied Irish Banks plans to expand the CoE’s remit to include emerging AI paradigms such as generative AI for customer service and AI‑augmented decision‑making in treasury operations. The bank also intends to strengthen its data‑foundation work, recognising that model quality hinges on clean, well‑governed data. By continuing to balance intentionality with enablement—through a CoE that guides rather than constrains—the organisation aims to sustain momentum, avoid the pitfalls of pilot purgatory, and harness AI’s full potential to deliver superior customer experiences and operational efficiency.
Conclusion
The journey of AI adoption in the enterprise is profoundly shaped by the consumer technologies employees already use. Without deliberate strategy, enthusiasm can devolve into uncontrolled pilot proliferation and stagnation. Allied Irish Banks’ experience demonstrates that a focused, intentional approach—anchored by a well‑designed Centre of Excellence—can reconcile consumer‑grade expectations with enterprise‑grade rigour. The CoE’s role as a facilitator, not a gatekeeper, provides the scaffolding needed to scale AI responsibly, turning isolated experiments into lasting competitive advantage. As AI continues to evolve, maintaining this balance will be critical for any organisation seeking to reap real‑world value from its AI investments.