This is a shorter version of the full article. The full learnings can be found on the members only DataIQ Hub.
At the 2026 DataIQ 100 Discussion, a panel formed of Chris Gullick, Chief Data & Artificial Intelligence, Ofgem; Murtz Daud, Director of Data & AI, British Gas; Tom Spencer, Director of Customer Data Sciences, Aviva; Sarah Barr Miller, Director of Data & AI, British Airways moved the conversation of transformation beyond technology. The real story was about knowing where AI genuinely matters, where it does not, and how to move organisations forward without being carried away by the hype cycle.
The AI mandate
Across sectors, data teams are now expected to deliver visible progress on AI. Boards want updates, investors ask questions, and competitors are making announcements. But the reality is that adoption remains cautious.
For many, the focus sits inside the organisation rather than directly in customer interactions. The priority has been productivity improvements and operational use cases, partly because customer-facing AI introduces a different level of risk.
That pattern is common in regulated industries. Leaders are keen to explore the potential, but the consequences of deploying immature systems in customer journeys are significant.
Yet the pressure to act is still useful. The AI agenda has unlocked funding for long-standing data challenges such as governance, tooling, and architecture that had previously struggled to secure investment. AI is acting as a catalyst rather than the transformation itself.
AI is not new, but the organisational challenge is
For some sectors, the idea that AI is new feels misleading. Insurance companies, for example, have been running machine learning models for years. Pricing risk and managing claims have always relied on advanced analytics.
What has changed is not the existence of AI, but the organisational context around it. Historically, data science could operate comfortably within business silos. Today, the most valuable opportunities require organisations to connect data across those silos, particularly around the customer.
For companies with millions of customers and multiple product lines, that is a significant undertaking. It demands enterprise-level thinking about data, architecture, and governance.
From experimentation to value
One of the most persistent frustrations in recent years has been the gap between proof of concept and real deployment. The panel’s experience echoed what many organisations have discovered in that the projects that scale successfully tend to begin with a clear business problem.
When initiatives are driven by curiosity about the technology, they rarely survive beyond experimentation. But, when they originate in a real operational challenge, such as improving a process, removing friction, or reducing cost, the pathway to production becomes much clearer.
In those situations, AI may or may not be the answer. Sometimes a simpler analytical model or process change delivers the result just as effectively. The key is discipline as not every problem needs an AI solution, however tempting the label may be.
AI works best when treated as a tool
One analogy offered during the discussion described AI like a power drill. You do not wander around drilling holes simply because you have one. You only use it when the task at hand calls for it.
The same principle applies to AI in organisations. When leaders start with the technology rather than the task, they tend to generate scattered experimentation with limited impact.
When they start with the outcome, for example, improving customer experience, increasing efficiency, or enabling better decisions, AI becomes one of several possible approaches. The leadership challenge is maintaining that clarity when the external narrative continues to focus on the technology itself.
The workforce shift is in action
If customer transformation remains gradual, the workforce impact is already visible. Tools like Copilot are spreading rapidly through organisations, particularly for administrative tasks and routine analysis. Employees are generally enthusiastic about using AI to remove repetitive work. But deeper questions are starting to surface.
If AI automates entry-level tasks, how do people develop expertise? How do early-career employees build the foundational knowledge that previous generations acquired through repetition? These are questions that have not been fully answered. What is clear is that AI is already reshaping how work gets done, and leadership teams are beginning to rethink workflows, roles, and training.
Culture remains the hardest challenge
The panel returned repeatedly to the conclusion that the biggest barriers to AI transformation are rarely technical but are cultural.
Budget approvals and technology choices are visible obstacles. The harder challenge lies in organisational friction with processes designed for stability rather than experimentation, teams reluctant to change established workflows, or expectations that AI systems should behave like traditional software.
Interestingly, resistance often does not come from employees themselves and often appears in the middle of the organisation.
Some managers remain uncomfortable with AI-assisted work, interpreting it as cutting corners rather than improving productivity. Yet those same managers may use the tools privately themselves. This tension reflects a deeper cultural transition. Work is shifting away from manual effort and towards judgement, interpretation and oversight.
This is a shorter version of the full article. The full learnings can be found on the members only DataIQ Hub.
