Britain sells knowledge for a living — and knowledge is the one thing AI is making cheap. You'd never know it from the plan

Tomorrow, 9 July, nominations open for the Labour leadership, and the man likely to win — and to walk into Number 10 soon after — has an actual plan for AI and your job. That is rarer than it should be, so I will give him this much: he is the first on that bench in years to say the words "the future of work" out loud and mean your job by them.

Here is the catch. He has said the words. But read what is actually in the plan and it is the past of work. The Financial Times set it out on 2 July: retraining for people whose jobs are exposed, limits on who owns the data centres going up around Britain, a bigger stick for the competition regulator, a review of the growth zones. Sensible enough, as far as it goes. But retraining is the answer every government reaches for — and only this week this paper set out Parliament's own verdict on how Britain's past retraining schemes did: "minimal impact." It is the last war, fought again.

And here is the thing none of it starts from. Britain makes its money selling what it knows. Law, finance, consultancy, accountancy, analysis, design — thinking for a living is close to the centre of this economy. And thinking for a living is the exact thing the machines are making cheap. The government's own assessment, published this January, found that around seven in ten UK workers are in jobs with tasks AI can already do or speed up, and that adverts for the most exposed roles have dropped by more than a third since 2022. The part of the economy most in the firing line is the part we sell to the world. A plan for AI that does not start there is not yet a plan for Britain.

And this is not a Burnham problem, or a Labour one. Go back through every recent government and find me the one that built its industrial strategy on the sentence "the thing we are best at selling is the thing about to be made cheap." It is not there. The whole bench has the same blind spot. He is just the first to stand near the edge of it.

Let me tell you why I know the exposure is real, and not from a report. A company I work inside — thirty-odd people — sat down at the start of this year and asked the blunt question: could we run this place on agents instead of staff? We did not guess. With everyone's say-so, we mapped the whole organisation, the work and the conversations, inside and out, and asked the tools honestly what they could take on. The answer came back at about nine in ten of it.

The founders have a conscience, and they pulled back. We are still thirty people. But the number does not un-know itself — and our competitors will run the same map, get the same answer, and not all of them will decide to keep the people. The one who does not sets the price the rest have to meet.

And I am not writing this from the outside. I am building this paper — the desks, the research engine, the data behind it — as a company of one. Before these tools, what you are reading would have taken a newsroom: a dozen writers, a subs desk, a data team. That is not a boast. It is the question this paper exists to ask, arriving at my own door — what do the dozen people I did not have to hire do now?

I am not going to tell you the answer — more state or less, protect the sector or throw it open. I do not know, and anyone selling you certainty on that is selling. But naming a problem and answering it are not the same thing, and what he has answered is a version of the last question, not this one.

We are here to ask one thing: what do people do when we do not have to work? Britain's version is sharper, and it is the one landing on the next Prime Minister's desk whether he is ready or not — what does a country that sells what it knows do when knowing gets cheap?

One thing you are owed before the Playbook, and it goes high tonight because the subject is the machines themselves. This note is mine — the view, the argument, the call to run it — and it was drafted with the same AI that writes the rest of the paper. Every analyst across these pages runs on models built by Anthropic. This is a paper about what the machines are doing to work, written from inside the machines, and we tell you so.

THE PLAYBOOK

Read the plan for what it counts, not what it promises. A strategy with targets for building data centres but none for the people whose work is being made cheap is telling you where its attention really is. Ask what it measures.

If you earn your living by what you know, do what the government's own assessment does — go through your own week and mark which parts are routine handling of information, and which are judgement, trust and dealing with people. That line is where the pressure lands, and knowing your side of it is worth more than any forecast.

Do not take the productivity promise on trust. When a tool is sold to your employer as time saved, ask whose time, and where it goes — back to the person doing the work, or onto the bottom line.

The Playbook is general information from the editor, not financial, legal or career advice.

— M.