AI is moving beyond analysis and into the systems that operate the physical world.
Across manufacturing, energy, utilities, transport, logistics, robotics and critical infrastructure, intelligent systems are beginning to influence what gets inspected, prioritised, escalated and acted upon.
Once AI starts shaping operational decisions, accuracy alone is not enough. Organisations must be able to demonstrate that systems remain reliable when conditions change, that decisions can be traced and challenged, and that human control remains meaningful.
This UKAI panel will examine what must be true before AI can be trusted to influence live operations.
Hosted by Tom Greenlees
The session will be hosted by Tom Greenlees of Intelligent Core, whose work sits at the point where real-time operational data becomes predictive insight and controlled action.
His experience across telemetry, edge and on-premise deployment, predictive systems, operational oversight and controlled autonomy gives him a strong perspective on the practical questions organisations face when AI moves from the dashboard into complex physical environments.
Tom will guide a discussion focused on the conditions required for dependable deployment, rather than technology in isolation.
About the discussion
Tom will be joined by senior leaders from AI, engineering, operations, infrastructure, cyber security, safety and assurance.
Drawing on live examples from different sectors, the panel will explore how organisations are introducing AI into environments where decisions affect assets, people, service continuity, compliance and operational performance.
The discussion will examine where deployment is already creating measurable value, where projects are becoming stuck, and what separates a promising pilot from a system that can be trusted at scale.
The session will explore
-
The point at which decision support becomes operational control
-
How organisations distinguish between AI that reports, recommends, prioritises, orchestrates and acts
-
Live examples of AI influencing maintenance, asset performance, network operations, logistics and physical workflows
-
The evidence required before a system moves from pilot into operational deployment
-
How edge, on-premise and cloud architectures affect resilience, latency, security and control
-
How human override and escalation should work when conditions change
-
How model drift, changing data and supplier updates affect assurance after deployment
-
How responsibility should be allocated across developers, integrators, operators and asset owners
-
What boards, regulators, insurers and procurement teams should require before approving deployment at scale
