The government wants the UK to be the fastest AI-adopting country in the G7. Last month, it published its plan for how manufacturing can play its part, setting out a clear ambition to move proven AI applications from successful trials into wider industrial use.
For manufacturers that have already explored AI, the challenge now is less about discovering what is possible and more about creating the conditions needed to scale it effectively.
The AI Adoption Plan for Advanced Manufacturing, written by government AI Champion Professor Chris Dungey, sets out the opportunity clearly. Wider AI deployment could add £5-6bn in gross value added to the economy every year and lift productivity by 2.5%. Manufacturing already contributes around £234bn annually, supports 2.5 million jobs, and drives almost half of all private sector R&D investment in the UK. The plan is part of a wider £200m+ package announced at the government’s AI Adoption Summit, aimed at moving AI “out of the lab” and into everyday use across eight priority sectors.
At the advanced end of manufacturing, the direction of travel is already visible. Rolls-Royce, BMW and Siemens are running predictive maintenance, digital twins and AI-driven quality inspection at scale. The next challenge is enabling more manufacturers to achieve the same level of adoption.
Three-quarters of UK manufacturers have already run an AI pilot, but many are still working through the steps required to move beyond experimentation. The plan highlights the barriers that continue to slow progress: legacy production systems, fragmented and inaccessible data, uncertainty around return on investment, and workforce capability gaps that prevent promising pilots becoming embedded production systems.
This is the gap bespoke software firm Propel Tech works in. Working across manufacturing and logistics, the company helps businesses tackle the practical challenges involved in scaling digital transformation: modernising legacy applications, connecting fragmented data, and building bespoke systems that allow AI tools to operate within live production environments rather than isolated test environments.
Wil Jones, Technology & Solutions Director at Propel Tech, says: “The ambition in this plan is right, but scaling AI in businesses isn’t really a model problem. It’s about mapping the workflows where AI can take busy work off people, then making sure the model has the right data and context to make good decisions. That means the data can’t be sitting in disconnected systems, and production environments have to actually be built to support it. That’s software engineering work as much as anything else.”
Make UK chief executive Stephen Phipson called the plan “the missing piece of the jigsaw” for manufacturers looking to scale up and grow. Siemens UK and Ireland CEO Brian Holliday described it as “a welcome signal of intent” for tackling the UK’s productivity challenge. Both points to the same underlying issue: the focus now needs to shift from proving AI’s potential to building the infrastructure that allows businesses to use it effectively. As the plan itself acknowledges, this is not about inventing entirely new AI capabilities; it is about scaling technologies that are already delivering value.
Propel Tech’s structured approach to AI-assisted delivery, covering architecture, governance and validation before acceleration, mirrors the phased “scan, pilot, scale” pathway the government itself has proposed. Its experience working with manufacturers reflects a wider industry challenge: successful AI adoption depends as much on the systems, data and foundations underneath the technology as it does on the AI itself.
Government has set the direction, but what separates the businesses that benefit from those that don’t will come down to basics like whether machine data is structured properly, accessible in real time, and whether the systems actually talk to each other.
