The headlines say UK organisations are in the middle of an AI revolution. In reality, when you sit in the rooms, it feels different. It feels tired, a bit anxious, and quietly unsure.
Most of my week is spent with councils, charities, and SMEs across the North East and beyond. The technical challenges are real, but they are rarely the thing stopping progress. What holds people back is how this all feels from the inside. Nobody puts that in a board paper, so it goes unspoken, and then everyone is surprised when the programme stalls.
Here is what I hear.
“Everyone else seems to have worked this out”
This is the most common feeling in the room, and it is almost never true.
LinkedIn is a highlight reel. Conferences are a highlight reel. A competitor announcing an AI partnership is a press release, not a working system. Meanwhile, the team I am with has one Copilot pilot that half the team forgot about, and they are convinced they are years behind.
The national picture tells a different story. The ONS found that the average UK business using AI has gone from about 1.4 AI technologies in late 2023 to 1.6 by June 2026. That is not a revolution. It means most organisations have one or two things running, and not much else. The gap between the noise and the reality is enormous, and the noise is does real damage. When organisations think they are falling behind, they make rushed decisions. They buy licences to feel like they are moving. They start three things at once. They skip the unglamorous work because it looks too slow next to everyone else’s announcements.
You are not as far behind as you think. Almost nobody is where the marketing says they are.
Change fatigue is the thing nobody accounts for
Most teams I meet are not resistant to AI. They are just worn out.
They have just finished a finance system migration, or a restructure, or two years of doing more with fewer people. Then someone arrives and says the way you work is about to change again. The response is not hostility. It is a polite, very British lack of enthusiasm, which is much harder to shift.
This often gets misread as a communications problem, so the answer becomes more communications. It is not. It is a trust problem. People have been promised that the last three systems would make their lives easier, and their lives did not get easier. Until something actually gives them time back, they will treat the next promise the same way, and they will be right to.
Nobody says the quiet part in the workshop
In the session, people ask sensible questions about data and integration. In the corridor afterwards, someone quietly asks if their job will still exist in three years.
That question is present in every room, whether it is spoken or not, and it shapes everything. It is why people do not share the shortcuts they have worked out. It is why nobody mentions the spreadsheet they built that the whole team depends on. If you are worried that describing your work in detail is how you automate yourself out of it, you describe your work vaguely.
So, the organisation ends up with a vague picture of its own processes and builds the wrong thing.
You cannot fix this with reassurance in a slide. You fix it by being clear about what is and is not on the table, and by involving the people who do the work in shaping what changes. When someone helps design the thing, they stop being its subject and start being its owner. The feeling in the room changes, and so does the quality of what you learn.
The person leading it has no time to lead it
In many organisations, AI has been handed to someone as an extra. They are enthusiastic, capable, and already stretched. It sits on top of a day job that was already full.
The Local Government Association’s research on councils found the top barriers to deploying AI were funding, workforce skills, and staffing capacity, with 52% naming capacity. That matches what I see well beyond local government. It is not that people lack ideas. It is that the AI work is the thing that gets dropped when something urgent comes up, and something urgent always comes up.
The honest result is that any initiative needing sustained officer or staff time upfront, with a payoff eighteen months out, will quietly die. Not because it was a bad idea. Because nobody had a spare afternoon. That is why we push hard for a first piece of work that gives time back quickly and visibly. Reclaimed time is the only budget most of these teams have.
Senior people are nodding along and hoping nobody asks
There is a real discomfort at leadership level that rarely gets talked about.
Executives are expected to have a view on AI. Many are not confident they understand it well enough to have one, and the environment does not make it safe to say so. So they nod along, approve things they cannot really evaluate, and defer to whoever in the room sounds most certain. Sounding certain is not the same as being right.
The DSIT AI Labour Market Survey found the most cited skills gap across UK organisations was not programming or engineering. It was understanding AI concepts, with over 57% of businesses reporting a technical skills gap, while 30% reported a non-technical skills gap. You do not close that gap by hiring a data scientist. You close it by giving your existing leaders enough grounding to ask good questions and say, “I don’t follow, explain that again,” without it costing them anything.
The best sessions I run are the ones where a chief executive says exactly that. Everything gets better afterwards.
Everyone is already using it, and nobody has said so
The last feeling is a strange one: a quiet, collective sense that the rules have not caught up.
Microsoft’s research with Censuswide, covering just over 2,000 UK employees, found 71% had used unapproved AI tools at work, and around half were still doing so weekly. In most organisations, the leadership team has no real idea, or suspects and would rather not confirm it.
I do not see that as recklessness. I see it as people trying to get their work done in a vacuum where nobody has told them what is allowed. In the absence of permission, capable people improvise. If you haven’t given them a sanctioned option and a clear line about what is and isn’t acceptable, that is the organisation’s gap, not theirs.
The upside is that this improvisation is the best insight you will get for free. Where people are quietly using AI is exactly where the work is painful. That is your roadmap, already written.
Where this actually starts
It does not start with a platform decision. It starts by naming the feeling in the room, because the mood in an organisation is a delivery risk like any other, and it is the one nobody puts on the risk register.
Say clearly what is and is not on the table for people’s roles. Make it safe not to know things, starting at the top. Pick one thing that gives someone their Friday afternoon back, and let that do the persuading. Recognise the tiredness rather than talking over it.
Do that, and the technology conversation gets much easier. Skip it, and the best-designed solution in the world will sit unused, and nobody will quite be able to tell you why.
Written by Ben Dawson, Business Development Executive
If you want help shaping your own AI or digital strategy, our team at Marra is always happy to talk.
