Much of the summer’s commentary on AI policy has focused on who should be responsible for it. DSIT has been abolished, and an AI minister now attends Cabinet for the first time, with Kanishka Narayan’s brief split between the Cabinet Office and the new DBIST. The Institute for Government has warned that a change on this scale creates uncertainty, both for the civil servants delivering the work and for businesses that had built relationships with DSIT. That concern is fair, but for a fifteen-person business deciding whether to change how it works, the answer is unlikely to depend on which department holds the strategy.
The more useful evidence comes from businesses themselves. The British Chambers of Commerce reported in March that 54% of the firms it surveyed now use AI, up from 25% in 2024. Around 94% of those surveyed were SMEs. Yet among SMEs using AI, 86% say job roles have stayed the same and 95% say it has had no effect on headcount. Only about one in ten have moved beyond generic tools to applications built for their own business. These figures suggest that adoption is moving considerably faster than capability.
This is where we believe the real gap lies. It is less about people’s ability to use AI tools and more about management capability. Most employees can learn to use an AI tool in an afternoon. The harder task is looking at a business process and deciding which parts a machine should take on, which should stay with a person, who checks the output and what a good result looks like. These are questions of work design, and many managers have had little opportunity to practise them. Adding AI to a job without changing how the job is designed may produce faster emails and quicker admin. It will not necessarily produce a more productive business. The British Chambers of Commerce’s finding that smaller firms and manufacturers are adopting more slowly matters here, because these are often the businesses with the least capacity to spare someone to work out how AI fits into their operations.
The government has set an ambition to help ten million people develop practical AI skills. That is a significant ambition, but it needs to be matched with support for the owners and managers who ultimately decide how work is organised. The evidence shows why. Gardiner & Theobald’s survey for DSIT found that 97% of organisations in the AI sector report at least one skills gap, and that 88% rely on on-the-job training rather than structured programmes. Learning on the job has its value, but it rarely gives managers the structure or the time to redesign a process. Training should be practical, delivered close to the businesses it is intended to support, and focused on applying AI to real processes rather than demonstrating another set of tools. Whatever the final arrangement in Whitehall, businesses should not lose a year of support while the structure settles.
This is also the thinking behind the Skills Bootcamps we deliver at Primed Talent. Currently in the Applied AI for Business Innovation bootcamp in York and North Yorkshire, learners finish with an AI business case relevant to their own sector and in the AI for Work bootcamp in Buckinghamshire, they design and present an AI or automation solution for their own workplace. Both programmes are fully funded for eligible learners, but more importantly, both begin with the problem a business is trying to solve rather than the technology itself.
The machinery of government may continue to change, but businesses cannot afford to wait for every structural question to be resolved. The focus now should be on giving businesses, particularly smaller ones, the practical capability to turn AI adoption into meaningful changes in how work is done.
