Scaling Agentic AI: Frameworks, Governance and Real-World Impact
Gavin Goodland, Chief Data Officer, National Grid; Miryem Salah, Director Digital Data & Transformation, VodafoneThree; Perry Philipp, Chief Data Officer, Entain; Shreenivasa Rajanala, VP, Global Head of Data Science and Advanced Analytics – Markets & Customer Powerhouse, Bayer; Viveca Pavon, Data & AI Lead for Public Sector, Accenture
For some, Agentic AI feels out of reach, but those who are moving from experimentation to deployment are feeling the benefits of first-mover advantage. In this panel session, leaders already leveraging Agentic AI spoke about how they navigated complexity, differing levels of readiness, and balanced ethical responsibility in deploying programmes.
From experimentation to direction
Many organisations begin with a broad wave of pilots; with one panellist calling it letting “a thousand flowers bloom.” Over time, that approach gives way to more curated programmes tied to clear business drivers. The emphasis should be on identifying where Agentic systems genuinely improve efficiency, decision-making, or customer experience, rather than pursuing AI activity for its own sake.
Designing for a hybrid world of human and AI actors
The leaders noted that the design challenge has fundamentally changed. Customer journeys and internal processes increasingly assume that both the user and the service provider may be human or AI. This uncertainty makes experience design more complex, forcing organisations to rethink who (or what) the “customer” might be in the future.
Governance is evolving from restriction to structured enablement
Early responses to generative and Agentic AI were often defensive, with some organisations temporarily banning tools outright. That stance has quickly proved impractical. In its place, more nuanced governance models are emerging with rigorous risk-based guardrails, traceable agent identities, and clearer control layers that enable experimentation while maintaining accountability.
Workforce change is as significant as the technology itself
Leaders stress the real work lies in organisational adaptation. Many map which roles will be augmented, created, or phased out, with HR and technology teams working together on reskilling plans as an example. Alongside formal training programmes and internal AI academies, there is recognition that fear and uncertainty among employees must be addressed openly.
Real value comes only when pilots translate into operational impact
The gap between proof-of-concept success and scalable enterprise deployment is a common experience. Organisations are putting structures in place, such as dedicated value realisation functions, to track whether AI initiatives genuinely affect revenue, cost efficiency, or operational performance. The consensus from panellists is clear: the technology is advancing quickly, but extracting measurable business value remains the harder task.
What Next for the Role of the CDO?
Wade Munsie, Chief Data Officer, Heathrow Airport; Laia Collazos, Chief Data and Analytics Officer, Coca-Cola Europacific Partners; Ryan Den Rooijen, MD AI and Monetisation, Currys.
As data and AI move from enablement to execution, the role of the Chief Data Officer is rapidly evolving. The three leaders in this fireside chat explored what they expect to be next for CDOs.
The CDO role is shifting from builder to orchestrator
CDOs are increasingly coordinating complex ecosystems that span internal teams, external vendors, and governance structures in addition to balancing value creation with rising expectations around AI safety, compliance, and oversight.
Influence at board level depends on translating technology into business language
CDOs are expected to shape the AI narrative for senior leadership. The most effective leaders rely on clear stories and analogies that connect data and AI initiatives to business priorities such as revenue, resilience, and risk management.
Ownership of AI is becoming more distributed
AI responsibility is increasingly shared across organisations. Many are introducing models such as designated AI owners for specific applications and cross-functional decision forums, ensuring the value delivered and the risks managed are clearly accountable.
The role increasingly demands adaptability and hands-on credibility
Rapid advances in AI are reshaping expectations of data leadership, stressing the importance of continuous learning, experimentation with new tools, and demonstrating tangible outcomes. Credibility now comes from strategic direction and showing working examples of value in production environments.
Removing the Human from the Loop without Compromising Customer Experiences
The roundtable “Can We Remove Humans from the Loop Without Compromising Customer Experience?” generated strong conversations across two separate sessions. Leaders made clear that the goal is not removing humans from processes but designing service models where automation and human support each play the right role.
The leaders agreed on starting with the customer outcome rather than the technology. Participants stressed that the question should not be whether humans can be removed from the loop, but whether automation improves the experience. Routine, high-volume tasks (ie, simple transactions, booking changes) are well-suited to automation, improving speed and convenience. However, complex, sensitive, or high-value interactions (ie, insurance, healthcare, banking) still require human involvement as customers expect reassurance and judgement.
The importance of escalation paths between automated systems and human support is essential. Customer service experiences can fail because users become trapped in rigid chatbot flows without a route to a human agent. When escalation is poorly designed, automation can actively damage trust in the brand. In contrast, organisations that allow automation to handle simple queries while quickly routing difficult issues to human staff achieve better outcomes.
Leaders noted the need to balance cost reduction with experience and brand identity. Automation is commonly justified by lowering the cost to serve, such as call centres. Yet excessive automation can undermine the qualities that differentiate a brand. Particularly for consumer-facing sectors, human interaction remains a core part of the value proposition. Automation is most effective when freeing employees to deliver better human service rather than replacing it.
It was stressed that successful automation depends on strong foundations and governance. High-quality data, clear ownership of AI-driven processes, and structured evaluation frameworks are essential. Organisations are prioritising AI projects only when they align with business outcomes and P&L ownership. In practice, the most sustainable approach is iterative by testing automation in controlled pilots, measuring customer impact, and scaling only where it demonstrably improves efficiency and experience.
What Does It Take to Lead Data, AI, and Cultural Transformation in 2026?
Sarah Barr Miller, Director of Data & AI, British Airways; Tom Spencer, Director of Customer Data Sciences, Aviva; Murtz Daud, Director of Data & AI, British Gas; Chris Gullick, Chief Data & Artificial Intelligence, Ofgem.
2026 looks set to bring unique opportunities and hurdles for data and AI leaders to navigate. This panel dove into how leaders need to deal with challenges such as impending AI disillusionment, appealing to a resistant workforce, and creating genuinely valuable AI use cases that can scale.
The focus is internal value, not AI spectacle
The immediate emphasis must be on operational efficiency and not AI products. Data and AI leaders are prioritising internal productivity, process automation, and decision-making before pushing AI deeper into customer-facing services. In regulated industries particularly, caution remains high. AI is viewed as a near-term tool to improve existing operations rather than a reinvention of the business.
The AI hype cycle has shaped investment for leaders to manage
Boards and markets are increasingly reactive to AI narratives, which creates tension for data and AI leaders. On one hand, the momentum around AI can unlock funding for long-overdue investments in data integration and governance. On the other, shareholder scepticism means organisations must be careful not to overstate their AI ambitions. The pragmatic approach is to use momentum to secure enterprise data investment but stay disciplined about where AI genuinely delivers value.
Scaling AI depends more on organisational alignment than technology
The biggest barrier to production-scale AI is the organisation. Successful initiatives start with clear business problems and involve domain experts, risk teams, and operational stakeholders from the outset. Projects driven by technological curiosity rarely survive proof of concept. Leaders repeatedly framed AI as a specific power tool: effective when applied deliberately to specific problems.
Cultural change is the defining leadership challenge
Technology is advancing quickly, but organisational adoption is uneven. Many employees are enthusiastic about using AI to remove repetitive work, yet the deeper implications for roles, career paths, and skills are still poorly understood. As a result, most organisations are leaning on gradual cultural change such as digital fluency programmes, AI champions, and voluntary experimentation. The shift is about reshaping how work gets done and bringing people with you.
Stay connected with the DataIQ website to uncover the in-depth analysis of these sessions, keynote presentations, and a candid conversation with Johanna Hutchinson, 2025’s top-ranked DataIQ 100 Europe leader, exploring the most pressing challenges facing data leaders today.
