The Plan Isn’tJust About AI. It's About People.
Why the hardest part of industrial AI adoption has nothing to do with algorithms - and everything to do with the workforce asked to trust them.
Chris Dungey's AI Adoption Plan for Advanced Manufacturing makes a point that deserves more attention than it has received. The UK, he writes, does not have an AI problem. We have world-class research, strong manufacturing sectors and established innovation infrastructure. What we lack is deployment — the disciplined, repeatable work of putting proven AI into live production and scaling it across the sector.
Read the plan closely and a quieter claim sits underneath that one. The hardest blockers to scaling industrial AI are not purely technical -legacy systems, safety-critical environments and validation all play their part - but two of the most stubborn are workforce confidence and fragmented operational data (GOV.UK). In other words: people, and the information environments they work in.
That is where the plan will be won or lost. And it is where this workstream is, so far, most often under-resourced.
The bit everyone skips
A factory pilot that works technically but is never adopted is not a success. It is an expensive demo.
This is the pattern the Scan-Pilot-Scale pathway is designed to break. Government and industry co-fund a pilot, generate evidence of return, and stage-gate the scale-up on productivity gains and workforce impact (GOV.UK). The logic is sound. But between "pilot proven" and "scaled across the enterprise" sits a gap that even strong Catapult, Made Smarter and BridgeAI technical support will not close on its own: funding alone will not fill it. The human transition is the missing piece.
Engineers need to trust outputs enough to act on them. Operators need new skills and the confidence that AI makes their work safer, not redundant. Leaders need to realign teams, governance and daily routines around systems that did not exist eighteen months ago. And the fragmented data the plan flags? That is rarely a storage problem. It is a trust and ownership problem — no one is sure whose job it is to keep it clean, who is allowed to use it, or whether it will still be there next quarter.
None of this is solved by better technology. All of it is solved, or stalled, by people.
What adoption actually requires
At Brave & Heart, we have spent years on exactly this transition - not in the machine learning, but in the moment an organisation decides to actually use something new. Three things consistently determine whether a deployment scales or stalls.
Readiness, honestly assessed. Before a pilot, someone has to ask: where is the genuine, high value opportunity, who will be affected, and is the organisation ready to act on what the pilot shows? That is a strategy and business-case question, not a technical one. Get it wrong and the pilot proves the wrong thing.
Confidence, deliberately built. Adoption is a leadership and culture task. Teams adopt what they trust, and trust is built through involvement, transparency and early wins - not slide decks. The workforce-capability workstream in the plan is not a soft adjacency to the technical work; it is the work that decides whether the technical work lands.
Data people can actually use. Trusted data environments are not just secure repositories. They are workforce-facing layers - reporting, collaboration, decision support - that give people confidence the information in front of them is reliable. This is the bridge between fragmented operational data and everyday decisions on the shop floor.
What this looks like in practice
For Brave & Heart, adoption support across the plan's pathways means four concrete things:
Readiness and business case. Honest AI-readiness scans and opportunity framing behind the AI front door - identifying where value is real and whether the organisation is ready to act on it.
Workforce adoption. Leadership and team alignment, confidence-building and embedding new ways of working in daily operations — the People & Culture Evolution work that decides whether a deployment lands.
Evidence capture. Workforce-facing reporting, collaboration and decision-support layers that turn fragmented operational data into reliable, repeatable evidence of ROI and impact - the proof that stage-gates the scale-up.
Replication playbooks. Reusable models for lighthouse and SME fast-track deployments, so what works in one plant or supplier can be packaged for others.
Our role in the plan
We are not overtly an OT integrator, and we do not pretend to be. The HVM Catapult, Made Smarter and BridgeAI partners will build the factory floor AI. Our role is the layer that makes it stick.
We lead the adoption and people workstream for companies - the readiness scans behind the AI front door, the workforce-capability building, the change management around lighthouse deployments, and the workforce-facing data and reporting layers that turn fragmented operational data into decisions people trust. We have done this at enterprise scale for dispersed workforces, from DP World's ten-million-view intranet to P&O Ferries' 4,000-person digital workplace, and turned complex reporting into the kind of data-driven system that delivers insight in seconds for CPI.
The plan's ambition is national-scale industrial AI adoption. Ours is to make sure that when the technology arrives on the floor, the people are ready, willing and able to use it - and that the evidence of success is captured in a form others can follow.
The why behind it
Chris Dungey is right that this is a competitiveness challenge and an industrial transformation challenge. We would go one step further. Industrial transformation is, finally, a people transformation. The factories that scale AI fastest will not be the ones with the best models. They will be the ones whose workforce was ready to trust them.
That is the work we do. With heart.
Brave & Heart is a B Corp certified strategy, people and digital consultancy, and a G-Cloud 14 and Digital Outcomes & Specialists 7 partner. We help organisations turn proven technology into adopted practice. braveandheart.com