Skip to content
UK Economy

Britain's Productivity Problem Has One Realistic Answer Left, And It Is Sitting In Everyone's Back Office

The OBR downgraded UK productivity growth, and that single revision did more damage to the public finances than most policy decisions of the last decade. Growth is running near 1.1% for 2026, inflation is forecast to peak just below 4% at year end before falling back, and the Bank is expected to keep cutting toward 3%. Britain has spent fifteen years trying to fix productivity with infrastructure, skills policy and planning reform - all correct, all slow. There is one lever that works on a two-year horizon rather than a twenty-year one, and the uncomfortable truth is that it is not a government programme. It is thousands of firms automating the administrative sludge that consumes a third of the working week.

AlchmAI Editorial13 min read

~1.1%

Expected UK GDP growth for 2026, with consensus forecasts spanning roughly 1.0% to 1.8%

~4%

Where the IMF expects headline inflation to peak temporarily at the end of 2026 before falling back

1%

Medium-term productivity growth assumption following the OBR's downgrade - the revision that reshaped the fiscal picture

~4.25%

Estimated 10-year gilt yield at end-2026, with the Bank's policy rate expected to move toward 3%

There is a number in British economic policy that does more work than any other, and almost nobody outside the Treasury thinks about it. It is the Office for Budget Responsibility's assumption about medium-term productivity growth. When the OBR revised it down, the effect on the public finances was larger than most individual tax and spending decisions of the past decade, because productivity growth compounds through every projection: tax receipts, wage growth, debt servicing, the lot. A downgrade of a fraction of a percentage point over five years is worth tens of billions, and it arrives without a single ministerial announcement.

The surrounding picture is one of an economy that is stable rather than dynamic. UK GDP growth is expected to run around 1.1% in 2026, with consensus forecasts spanning roughly 1.0% to 1.8%. The IMF's Article IV mission expects headline inflation to rise temporarily, peaking just below 4% at the end of 2026, before returning toward target. Ten-year gilt yields are estimated around 4.25% at end-2026, with the Bank of England expected to continue cutting toward a policy rate near 3%. None of this is crisis. All of it is a country running at a pace that makes the debt arithmetic uncomfortable and the political arithmetic worse.

Why The Usual Answers Are Correct And Insufficient

The standard prescriptions for British productivity are, in our view, genuinely right. They are also almost all slow, and slowness is the binding constraint in a country with a five-year electoral cycle and an urgent fiscal position.

  • Infrastructure. Transport links that let labour markets function properly are transformative and take a decade or more from decision to benefit, assuming the decision is not revisited - which, in Britain, it usually is.
  • Planning reform. Genuinely fundamental, since Britain's inability to build near where the productive jobs are may be its single largest self-inflicted economic wound. The government's AI Growth Zone work, routing data centres through the Nationally Significant Infrastructure regime, is real progress on exactly this - and the target is still cutting average consenting time from eighteen months to twelve. That is an improvement measured in months on a project measured in years.
  • Skills. Correct, generational, and no help whatsoever to a firm trying to improve output per hour in the current financial year.
  • Capital investment incentives. Effective, well-evidenced, and dependent on firms choosing to invest, which in an economy growing at 1.1% many understandably do not.
  • R&D funding. Britain does research exceptionally well and commercialises it poorly. The gap is real and decades old.

Every one of these deserves to be pursued. None of them changes output per hour in 2027. And that is the gap into which AI-driven process automation falls, because it is the only intervention we know of that is available to an individual firm, funded from the firm's own budget, with a payback measured in months.

The Unglamorous Case For Automation As Industrial Policy

Here is where we should declare our interest plainly: we build AI and workflow automation for financial firms in London, so we are hardly disinterested. But the argument does not depend on our opinion, and the mechanism is more prosaic than the phrase 'AI' suggests.

Walk into almost any British professional services, financial or insurance firm and you will find well-paid, well-qualified people spending a large fraction of their week on work that creates no value: rekeying data between systems that do not talk to each other, chasing documents, reconciling reports that should already agree, reformatting the same information for three different audiences, and checking things that a rule could check. This is not a failure of effort or management. It is the accumulated sediment of thirty years of buying software that solved one problem without connecting to the software that solved the last one.

“Britain's productivity problem is not mainly that its workers are unskilled or its capital stock is thin. It is that an enormous share of skilled time is spent being the integration layer between systems that were never joined up.”

That is precisely the work current AI is good at, and it is worth being clear why: it is high-volume, rules-heavy, text-shaped, and has a definition of done that a human can check in seconds. Those four properties are the reliable predictor of whether an automation project succeeds. When that work is removed, the same people do the work they were hired for, with the same skills and the same capital, and output per hour rises. That is the entire mechanism. It requires no new technology, no legislation, and no waiting for a rail line.

Why Britain Is Unusually Well Placed - And Unusually At Risk

We will be openly partial here, because we think the case is strong. Britain's economy is heavily weighted toward services - financial services, professional services, insurance, legal, creative. That composition has been treated as a weakness for two decades, on the reasonable grounds that services productivity is hard to improve. The arrival of technology that automates document-heavy, judgement-adjacent knowledge work inverts that argument almost exactly. The sectors where this technology is most applicable are the sectors Britain is most concentrated in. On the specific question of who benefits most from AI-driven services automation, the UK's industrial structure is close to ideal.

The supporting assets are real too. London remains Europe's deepest fintech market, accounting for around 79% of UK fintech investment. Roughly 30% of UK venture capital now flows into AI, and UK startup funding reached about $17bn in the first half of 2026 with AI taking the large majority. The regulatory posture is a genuine advantage: the FCA supervises real systems in live conditions through its AI Live Testing programme rather than classifying hypothetical ones in advance, which is a considerably better environment for a firm trying to deploy something than a prescriptive rulebook would be.

What Would Actually Help, From The Policy Side

Taking the diagnosis seriously suggests policy that looks quite different from most current AI strategy, which is heavily weighted toward compute and frontier capability. Those matter. They are not what moves the productivity number.

  1. 01Target the diffusion, not the frontier. The productivity gain is in the ten thousandth firm adopting proven automation, not in the first firm training a better model. Diffusion policy is boring and is where the measurable return is.
  2. 02Make capital allowances explicitly cover software integration and process automation, not just hardware. A great deal of the necessary spend is engineering labour, which the current system treats less generously than buying a machine.
  3. 03Fund capability rather than licences. Subsidising software purchases produces shelfware. Subsidising the diagnostic and implementation work - the bit mid-sized firms genuinely cannot do themselves - produces working systems.
  4. 04Use public sector procurement as a demonstrator. Government is itself a vast, document-heavy, integration-poor organisation. Automating its own administrative sludge would improve public finances directly and produce a British reference case that the private sector could copy.
  5. 05Keep the regulatory posture. The FCA's live-testing approach is a genuine competitive advantage for UK financial services and should be protected against the pressure to write a prescriptive rulebook simply because other jurisdictions have.

And From The Firm's Side

The honest position is that most of this will not be solved by policy at all. It will be solved, or not, by several thousand individual firms deciding to do something specific. If you run one, the sequence that works is unromantic: measure where the hours actually go rather than where you assume they go, pick the single highest-volume process with a clear definition of done, automate all of it rather than part of several, keep a genuine human decision point, and measure the result honestly. That is an eight-to-twelve week exercise, not a transformation programme, and it produces a real number rather than a strategy document.

The Bottom Line

Britain's fiscal position, its growth rate near 1.1%, its gilt yields around 4.25% and its uncomfortable political arithmetic all trace back to one stubborn number: productivity growth that has been broadly flat since 2008 and which the OBR has now downgraded. Every conventional answer - infrastructure, planning, skills, R&D - is correct and slow. AI-driven process automation is the only lever available that a single firm can pull, fund itself, and see results from inside a year, and Britain's services-heavy economy is unusually well positioned to benefit from precisely this form of it. The risk is not that the technology fails; it is that it diffuses only to firms large enough to have an AI department, leaving the national figure unchanged and the gap wider. Closing that gap is a distribution problem, not a frontier one - which is why the most valuable thing for British productivity right now is not a bigger model, but a few thousand mid-sized firms automating the administrative sludge that is quietly consuming a third of their working week.

References & Further Reading

UK EconomyAI Agency UKAI Automation LondonWorkflow Automation LondonproductivityOBRFintech AI Agency London
Share Email
AI

AlchmAI Editorial

Research and analysis, London

The AlchmAI team writes about the markets, technology and regulation we work with every day. We build trading platforms, real-time charts and AI analysis tools for brokers, prop firms and fintech teams from our office in Mayfair, London. Every article lists its sources. Nothing we publish is investment advice.

This article is general information and commentary. It is not investment advice or a recommendation to buy or sell any investment. Important information

Keep reading

All articles