AI AND DIGITAL TWINS TRANSFORMING DOWNSTREAM OPERATIONS, EXPERTS SAY AT FUTURE DOWNSTREAM 2025

Industry leaders examine the realities of scaling AI, managing IT/OT convergence and putting people at the centre of digital transformation.

Senior leaders from refineries, chemical manufacturers and technology providers came together for Future Downstream 2025, a virtual conference focused on the practical realities of deploying artificial intelligence, digital twins and data-driven technologies across downstream operations. The event brought together a strong international speaker line-up to explore how organisations can move beyond pilots to deliver real operational, commercial and cultural impact.

Centred on the theme of AI and digital transformation, discussions throughout the day acknowledged both the significant opportunity and the challenges facing downstream organisations. Speakers repeatedly stressed that while technology has matured rapidly, success ultimately depends on strong data foundations, organisational readiness and a workforce equipped to adapt. Foundational capabilities in data literacy, governance, transparency and accountability were highlighted as essential enablers for long-term resilience and value creation.

Adam Soroka, CEO of Cavendish Group, emphasised the human dimension of this transformation. “We should always remember that behind any artificial intelligence or technology integration there is a human driving this process. Digital transformation ultimately is not about technology - it’s about people, and the need to transition organisations to think differently while aligning strategy,” he said.

Opening the conference, Shehzad Rehman, Global Head of IT & Technology at INEOS, joined a keynote Q&A exploring change management and leadership alignment. He emphasised that effective digital transformation must start with clear, unified communication from leadership and a deliberate effort to address employee concerns around automation and AI. “We deliberately avoid a one-size-fits-all approach. At INEOS, countries are given the freedom to innovate locally, but governance, security and ethical guardrails are mandated centrally to ensure everyone operates in a safe and trusted environment,” he said.

The first panel, “Accelerating AI for Sustainable Plant Operations and Asset Performance Management,” examined how AI is enabling more autonomous operations across complex assets. Speakers from Linde, MOL, Borouge and Galp discussed how AI is being deployed to manage market volatility, address workforce scarcity and improve commercial performance. While business drivers such as cost, efficiency and margins remain unchanged, panellists agreed that today’s computing power, tools and infrastructure now make it possible to solve problems that were previously theoretical.

A recurring theme was the importance of managing AI models once deployed. Preventing model drift, validating training data and continuously monitoring performance were identified as critical practices, particularly in closed-loop systems supporting critical infrastructure. Speakers also stressed the need to define clear metrics for adoption and value, and to share these transparently across organisations to build trust and reduce resistance.

Human and organisational factors featured prominently throughout the discussion. Panellists warned that technology alone is insufficient without strong governance, clear ownership of AI-driven decisions and investment in upskilling. Several speakers noted resistance from skilled professionals concerned about job displacement, arguing instead that future competitiveness will favour those who learn to work effectively alongside AI.

Shane McArdle, CEO of Kongsberg Digital, reinforced the message that mindset must come before technology. He urged leaders to model data-driven behaviour in everyday decision-making, replacing slide-based reporting with dashboards and evidence-based discussions. “It’s not about the technology. We’re building a culture around AI, data and digital twins. It’s about mindset first, and capability comes afterwards.”

McArdle also advocated a two-speed operating model: rapid prototyping to demonstrate possibility, combined with slower, deliberate change management to ensure safe and sustainable adoption at scale. “Going back to the basics, making sure people are part of the process and allowing them to control the speed - prototype fast and change at a slower pace is how you eventually see it take off.”

Later panels explored IT/OT convergence and change management at scale, highlighting the cultural, technical and security tensions that continue to slow progress. Speakers noted that unifying IT and OT require more than new architectures; it demands new hybrid skills, shared accountability and top-down clarity around purpose and value. Zero-trust security models, edge computing and data-layer integration were identified as practical enablers for unlocking AI while protecting critical assets.

The final panel focused on overcoming cultural resistance and process inflexibility, with contributors from SABIC, Chevron, Dow, Motiva, BASF and Kongsberg Digital. Discussions examined the role of mid-level management, incentives and co-creation with frontline users in building trust. Panellists agreed that successful transformation depends on clearly articulating the “why” behind change, reinvesting productivity gains into reskilling, and designing solutions that genuinely improve day-to-day work.

Across the conference, one message was consistent: digital transformation in downstream operations is not a technology challenge alone, but a human one. Organisations that combine strong data foundations, clear leadership, cultural empathy and realistic timelines will be best positioned to scale AI responsibly and competitively in the years ahead.