Matthew Jeffery is one of Britain’s most experienced global talent and recruitment leaders, with more than 25 years advising boards and C-suite executives on workforce strategy, skills, and productivity.
‘Modernisation, not Austerity’ is Matthew’s second mini series on the AI revolution. You can read the first two-parter here, and here and part 2 of this series will be published at the same time tomorrow
Britain does not have a technology problem. It has a courage problem.
Artificial intelligence is already reshaping how work gets done across the economy, yet the state continues to behave as if it is waiting for permission to act. While the UK runs pilots and incremental programmes, other countries are moving with far greater ambition. Estonia’s newly launched Eesti.ai programme (January 2026) aims to double the value of work by 2035 through systematic AI integration across the entire public sector, supported by an international advisory board and a clear national strategy. A country of Britain’s scale and resources has no excuse for aiming lower.
From healthcare to welfare, courts to transport, the tools already exist to run public services faster, cheaper, and more effectively than they are today. This is no longer a question of whether the technology works, whether it is coming, or whether it can be trusted. It is already here, already proven, and already in use at scale.
The real question is simpler and far more uncomfortable. Is government willing to use it to fundamentally change how the state operates, or will it continue to protect systems that are visibly no longer fit for purpose? This is not a debate about innovation. It is a test of political will.
The Fiscal Reality Britain Cannot Avoid
Debt is rising and public spending is drifting out of control, yet Britain continues to avoid the conversation it needs to have about its public finances. Public sector net debt is currently around 93.1 per cent of GDP, based on the latest Office for National Statistics data for February 2026, and is forecast to approach or exceed 95 per cent in coming years. This places it among the highest sustained levels since the early 1960s. Total public spending is close to 45 per cent of GDP, significantly above pre-pandemic norms, while the tax burden is already at a post-war high and projected to rise further.¹ These are not abstract macroeconomic signals. They are hard constraints on what government can promise, what it can sustain, and what it can deliver.
At the same time, demand continues to rise across health, pensions, welfare, and social care, driven by demographic change, longer life expectancy, and structural pressures that are unlikely to reverse. This is not cyclical pressure that will ease with economic recovery. It is persistent, cumulative, and compounding. Each year adds new demand to systems that are already stretched, and each year makes the fiscal arithmetic more difficult to sustain. Without structural reform, the trajectory is clear, with higher spending chasing rising demand without ever resolving it.
For decades, governments have attempted to manage this through incremental efficiencies and periodic restraint, but without fundamentally changing how the state operates. Costs removed in one part of the system reappear elsewhere through delay, complexity, or failure demand.² The system has not become leaner. It has become more fragile, more reactive, and more expensive to run. This is the fiscal trap Britain now faces, and it is one that cannot be escaped through marginal improvements alone.
The Opportunity in Plain Sight
We talk endlessly about how artificial intelligence will shape the economy of the future, but far less about how it can reduce the cost of government today. That imbalance matters, because the most immediate opportunity for AI is not future growth but present-day productivity inside the state itself. Artificial intelligence is already doing things inside public services that would have sounded implausible even a few years ago. Systems can review tens of thousands of documents in minutes rather than days, detect fraud patterns across millions of transactions in real time rather than retrospectively, optimise infrastructure continuously rather than periodically, and return hours of time to frontline workers every single day.
These capabilities are not theoretical. They are already operational, already delivering measurable improvements in speed, accuracy, and capacity, and already changing how work gets done. The real question is whether they are deployed at scale, whether they are embedded into core systems, and whether their gains are captured in a way that changes the cost trajectory of the state. Without that discipline, productivity gains will be absorbed into rising demand, service expansion, or institutional inertia rather than translating into fiscal savings.
For the Shadow Chancellor, this is an opportunity hiding in plain sight. It offers a way to rebuild economic credibility and demonstrate to both business and the markets that there is a serious and deliverable plan for restoring discipline to the public finances. Because the argument is no longer simply about cuts. It is about capability, and about whether the state can operate more effectively with the resources it already has. It is also one of the few credible ways to show how you would govern, not just how you would criticise.
A Commitment to Redesign, Not Just Efficiency
There is a deeper question that sits beneath the opportunity outlined so far. Not how much artificial intelligence can save, but how far government is willing to go in redesigning itself around it.
A serious government should commit to a clear, measurable objective. Not incremental efficiency gains, not a collection of pilots, but a structural shift in how the state operates.
A credible ambition would be to halve the administrative overhead of the state by 2030 while improving outcomes. This is not a target rooted in headcount reduction alone, but in redesign. It implies removing entire layers of low-value activity, eliminating duplication across departments, and rebuilding services around systems that can act, learn, and improve continuously.
This level of ambition is not theoretical. Estonia’s Eesti.ai programme, launched in 2026, is explicitly designed to double the value of work by 2035 through systematic integration of artificial intelligence across government. It treats AI not as a digital upgrade, but as a national operating model. For the UK, with significantly greater scale, resources, and institutional capability, the benchmark should not be whether this is possible, but why it has not already been pursued with similar clarity.
This is not without precedent. Smaller digital states have already demonstrated what is possible when systems are designed around shared data, automation, and continuous optimisation. The question is not whether the model exists. It is whether a country of the UK’s scale is prepared to pursue it with the same level of intent.
Framed in this way, artificial intelligence is not a tool for marginal efficiency. It becomes the foundation of a fundamentally different kind of state. One that is leaner not because it has been reduced, but because it has been redesigned.
Not Austerity. Modernisation.
Handled properly, this is not austerity. It is modernisation, and the distinction is fundamental. Austerity reduces inputs and hopes outputs survive. Modernisation changes how the system works so that the same or better outputs can be delivered with fewer inputs. This is a fundamentally different proposition, both economically and politically, because it reframes the debate away from reduction and towards redesign.
The message is simple but powerful. Not smaller government, but smarter government. Not cuts for their own sake, but a state that delivers more for less. Not reduction, but redesign. However, this requires confronting an uncomfortable truth that governments have historically avoided. Artificial intelligence does not cut spending. Governments do. Without deliberate decisions to remove cost, productivity gains will expand capacity, improve services, or be absorbed into complexity rather than translating into savings.
Without workforce reform and system redesign, AI risks becoming a cost multiplier rather than a cost reducer. The barrier is not technology. It is the willingness to redesign roles, remove work, and take cost out of the system. That is the real test of political seriousness, and it is where most reform efforts have historically failed.
The Structural Problem Inside the State
The issue is not simply how much the state spends, but how effectively it converts spending into outcomes. Efficiency has too often meant deferral rather than redesign, pushing cost forward rather than removing it entirely. Artificial intelligence offers one of the first credible opportunities in decades to change that equation. It enables systems to be rebuilt around capability rather than constrained by legacy processes, allowing government to operate in fundamentally different ways.
It is not costless. Adoption requires investment in infrastructure, data, cybersecurity, and workforce capability.³ This creates a fiscal paradox, because savings require upfront investment at precisely the moment when fiscal headroom is limited. Delivering this at scale could require tens of billions over a parliament.⁴ However, early evidence suggests that returns can be both substantial and relatively fast, with many deployments showing payback periods of 12 to 24 months.⁵ This reframes investment as a route to compounding efficiency gains rather than a simple cost pressure.
What This Actually Looks Like in Practice
The case for artificial intelligence becomes far more compelling when translated into real systems, real roles, and real costs. Across the state, AI is compressing timelines, removing labour intensity, and unlocking capacity. What appears as a marginal improvement at the level of an individual task becomes transformative when applied across millions of interactions. The effect is not linear. It is multiplicative, reshaping entire systems rather than improving isolated processes.
Transport demonstrates that automation is not theoretical, but its true impact extends far beyond labour cost. Autonomous transport reduces accidents, which lowers demand on the NHS, policing, courts, and insurance systems. Human error contributes to the vast majority of road accidents, and even partial reduction produces cascading savings across emergency services, healthcare, and legal systems.⁷ It optimises traffic flow, reducing congestion, improving economic productivity, and lowering energy consumption. It transforms public sector fleets, including NHS logistics, local authority services, and wider government operations, into continuously utilised systems that reduce idle time and improve efficiency.
By increasing utilisation, autonomy reduces the total number of vehicles required, lowering both capital expenditure and maintenance costs. Vehicles no longer sit idle for most of the day but operate as shared, continuously optimised assets. This also generates real-time data that improves infrastructure planning, reduces congestion bottlenecks, and informs long-term investment decisions. Across these combined effects, autonomous transport represents a multi-billion-pound opportunity that reshapes both public spending and economic performance.
In healthcare, the impact is immediate and human. A doctor who once spent 43 minutes a day typing now spends that time with patients, with Microsoft 365 Copilot trials demonstrating this at scale.⁵ Across 1.5 million NHS staff, this translates into hundreds of thousands of hours returned every month. This is not simply efficiency. It is capacity, fundamentally changing how care is delivered without increasing headcount.
The largest opportunity sits across the broader public sector workforce. AI has the potential to save around one-fifth of public-sector workers’ time, or conservatively one-sixth, equating to tens of millions of hours annually and a potential £34–41 billion wage-bill impact over time.¹⁰ Even partial capture of this productivity translates into £5–12 billion in realistic savings over a parliament. A civil servant, NHS worker, or DWP work coach gaining back two full working weeks per year fundamentally changes how the state operates.
In welfare and tax, AI is already transforming fraud detection, with £7.5 billion recovered through enhanced systems.⁸ This is continuous enforcement at machine scale, identifying patterns and anomalies in real time rather than after loss has occurred. It represents a shift from reactive correction to proactive prevention, reducing both cost and administrative burden simultaneously.
In courts and justice, where the Crown Court backlog exceeds 80,000 cases, AI tools such as document review, transcription, and intelligent scheduling are already demonstrating measurable impact. Ministry of Justice reform programmes show that digital case handling and AI-assisted workflows can significantly reduce administrative burden and accelerate resolution.¹⁴ This reduces both cost and delay, improving outcomes across the justice system.
Procurement remains one of the largest and most immediate opportunities, with £300–350 billion in annual spend. Even modest efficiency gains of 3 to 5 per cent translate into £9–15 billion in savings.⁹ AI enables continuous optimisation of supplier selection, contract management, and pricing, shifting procurement from periodic review to real-time discipline.
Across the government estate, energy costs of £14–15 billion can be reduced by 10 to 20 per cent through AI-driven optimisation.¹¹ This reduces waste while maintaining performance, delivering consistent savings at scale across buildings and infrastructure.
Local government and planning represent another significant opportunity. AI tools such as Extract are digitising planning systems at scale, reducing processes that once took weeks to minutes.¹² This accelerates housing delivery, infrastructure development, and local economic growth while reducing administrative cost. Across local government, AI can unlock substantial savings through improved case management, scheduling, and service delivery.
Demand Reduction: The Real Step Change
The most powerful savings do not come from doing work more efficiently, but from eliminating the need for work altogether. Artificial intelligence enables systems to reduce demand before it enters the system, fundamentally reshaping cost structures across public services.
In healthcare, predictive modelling reduces hospital admissions by identifying risk earlier and enabling intervention before conditions escalate. In welfare, systems prevent overpayments before they occur rather than recovering them afterwards. In justice, disputes can be resolved before reaching courts, reducing backlog and cost. In transport, fewer accidents remove entire chains of cost across emergency services, healthcare, and legal systems. In planning, improved decision accuracy reduces appeals and rework, eliminating layers of administrative burden.
This is how the cost curve bends. Not through efficiency alone, but through prevention, redesign, and the removal of avoidable demand. It is a fundamentally different model of the state, one that intervenes earlier, operates smarter, and spends less because it needs to do less.
Matthew Jeffery is one of Britain’s most experienced global talent and recruitment leaders, with more than 25 years advising boards and C-suite executives on workforce strategy, skills, and productivity.
‘Modernisation, not Austerity’ is Matthew’s second mini series on the AI revolution. You can read the first two-parter here, and here and part 2 of this series will be published at the same time tomorrow
Britain does not have a technology problem. It has a courage problem.
Artificial intelligence is already reshaping how work gets done across the economy, yet the state continues to behave as if it is waiting for permission to act. While the UK runs pilots and incremental programmes, other countries are moving with far greater ambition. Estonia’s newly launched Eesti.ai programme (January 2026) aims to double the value of work by 2035 through systematic AI integration across the entire public sector, supported by an international advisory board and a clear national strategy. A country of Britain’s scale and resources has no excuse for aiming lower.
From healthcare to welfare, courts to transport, the tools already exist to run public services faster, cheaper, and more effectively than they are today. This is no longer a question of whether the technology works, whether it is coming, or whether it can be trusted. It is already here, already proven, and already in use at scale.
The real question is simpler and far more uncomfortable. Is government willing to use it to fundamentally change how the state operates, or will it continue to protect systems that are visibly no longer fit for purpose? This is not a debate about innovation. It is a test of political will.
The Fiscal Reality Britain Cannot Avoid
Debt is rising and public spending is drifting out of control, yet Britain continues to avoid the conversation it needs to have about its public finances. Public sector net debt is currently around 93.1 per cent of GDP, based on the latest Office for National Statistics data for February 2026, and is forecast to approach or exceed 95 per cent in coming years. This places it among the highest sustained levels since the early 1960s. Total public spending is close to 45 per cent of GDP, significantly above pre-pandemic norms, while the tax burden is already at a post-war high and projected to rise further.¹ These are not abstract macroeconomic signals. They are hard constraints on what government can promise, what it can sustain, and what it can deliver.
At the same time, demand continues to rise across health, pensions, welfare, and social care, driven by demographic change, longer life expectancy, and structural pressures that are unlikely to reverse. This is not cyclical pressure that will ease with economic recovery. It is persistent, cumulative, and compounding. Each year adds new demand to systems that are already stretched, and each year makes the fiscal arithmetic more difficult to sustain. Without structural reform, the trajectory is clear, with higher spending chasing rising demand without ever resolving it.
For decades, governments have attempted to manage this through incremental efficiencies and periodic restraint, but without fundamentally changing how the state operates. Costs removed in one part of the system reappear elsewhere through delay, complexity, or failure demand.² The system has not become leaner. It has become more fragile, more reactive, and more expensive to run. This is the fiscal trap Britain now faces, and it is one that cannot be escaped through marginal improvements alone.
The Opportunity in Plain Sight
We talk endlessly about how artificial intelligence will shape the economy of the future, but far less about how it can reduce the cost of government today. That imbalance matters, because the most immediate opportunity for AI is not future growth but present-day productivity inside the state itself. Artificial intelligence is already doing things inside public services that would have sounded implausible even a few years ago. Systems can review tens of thousands of documents in minutes rather than days, detect fraud patterns across millions of transactions in real time rather than retrospectively, optimise infrastructure continuously rather than periodically, and return hours of time to frontline workers every single day.
These capabilities are not theoretical. They are already operational, already delivering measurable improvements in speed, accuracy, and capacity, and already changing how work gets done. The real question is whether they are deployed at scale, whether they are embedded into core systems, and whether their gains are captured in a way that changes the cost trajectory of the state. Without that discipline, productivity gains will be absorbed into rising demand, service expansion, or institutional inertia rather than translating into fiscal savings.
For the Shadow Chancellor, this is an opportunity hiding in plain sight. It offers a way to rebuild economic credibility and demonstrate to both business and the markets that there is a serious and deliverable plan for restoring discipline to the public finances. Because the argument is no longer simply about cuts. It is about capability, and about whether the state can operate more effectively with the resources it already has. It is also one of the few credible ways to show how you would govern, not just how you would criticise.
A Commitment to Redesign, Not Just Efficiency
There is a deeper question that sits beneath the opportunity outlined so far. Not how much artificial intelligence can save, but how far government is willing to go in redesigning itself around it.
A serious government should commit to a clear, measurable objective. Not incremental efficiency gains, not a collection of pilots, but a structural shift in how the state operates.
A credible ambition would be to halve the administrative overhead of the state by 2030 while improving outcomes. This is not a target rooted in headcount reduction alone, but in redesign. It implies removing entire layers of low-value activity, eliminating duplication across departments, and rebuilding services around systems that can act, learn, and improve continuously.
This level of ambition is not theoretical. Estonia’s Eesti.ai programme, launched in 2026, is explicitly designed to double the value of work by 2035 through systematic integration of artificial intelligence across government. It treats AI not as a digital upgrade, but as a national operating model. For the UK, with significantly greater scale, resources, and institutional capability, the benchmark should not be whether this is possible, but why it has not already been pursued with similar clarity.
This is not without precedent. Smaller digital states have already demonstrated what is possible when systems are designed around shared data, automation, and continuous optimisation. The question is not whether the model exists. It is whether a country of the UK’s scale is prepared to pursue it with the same level of intent.
Framed in this way, artificial intelligence is not a tool for marginal efficiency. It becomes the foundation of a fundamentally different kind of state. One that is leaner not because it has been reduced, but because it has been redesigned.
Not Austerity. Modernisation.
Handled properly, this is not austerity. It is modernisation, and the distinction is fundamental. Austerity reduces inputs and hopes outputs survive. Modernisation changes how the system works so that the same or better outputs can be delivered with fewer inputs. This is a fundamentally different proposition, both economically and politically, because it reframes the debate away from reduction and towards redesign.
The message is simple but powerful. Not smaller government, but smarter government. Not cuts for their own sake, but a state that delivers more for less. Not reduction, but redesign. However, this requires confronting an uncomfortable truth that governments have historically avoided. Artificial intelligence does not cut spending. Governments do. Without deliberate decisions to remove cost, productivity gains will expand capacity, improve services, or be absorbed into complexity rather than translating into savings.
Without workforce reform and system redesign, AI risks becoming a cost multiplier rather than a cost reducer. The barrier is not technology. It is the willingness to redesign roles, remove work, and take cost out of the system. That is the real test of political seriousness, and it is where most reform efforts have historically failed.
The Structural Problem Inside the State
The issue is not simply how much the state spends, but how effectively it converts spending into outcomes. Efficiency has too often meant deferral rather than redesign, pushing cost forward rather than removing it entirely. Artificial intelligence offers one of the first credible opportunities in decades to change that equation. It enables systems to be rebuilt around capability rather than constrained by legacy processes, allowing government to operate in fundamentally different ways.
It is not costless. Adoption requires investment in infrastructure, data, cybersecurity, and workforce capability.³ This creates a fiscal paradox, because savings require upfront investment at precisely the moment when fiscal headroom is limited. Delivering this at scale could require tens of billions over a parliament.⁴ However, early evidence suggests that returns can be both substantial and relatively fast, with many deployments showing payback periods of 12 to 24 months.⁵ This reframes investment as a route to compounding efficiency gains rather than a simple cost pressure.
What This Actually Looks Like in Practice
The case for artificial intelligence becomes far more compelling when translated into real systems, real roles, and real costs. Across the state, AI is compressing timelines, removing labour intensity, and unlocking capacity. What appears as a marginal improvement at the level of an individual task becomes transformative when applied across millions of interactions. The effect is not linear. It is multiplicative, reshaping entire systems rather than improving isolated processes.
Transport demonstrates that automation is not theoretical, but its true impact extends far beyond labour cost. Autonomous transport reduces accidents, which lowers demand on the NHS, policing, courts, and insurance systems. Human error contributes to the vast majority of road accidents, and even partial reduction produces cascading savings across emergency services, healthcare, and legal systems.⁷ It optimises traffic flow, reducing congestion, improving economic productivity, and lowering energy consumption. It transforms public sector fleets, including NHS logistics, local authority services, and wider government operations, into continuously utilised systems that reduce idle time and improve efficiency.
By increasing utilisation, autonomy reduces the total number of vehicles required, lowering both capital expenditure and maintenance costs. Vehicles no longer sit idle for most of the day but operate as shared, continuously optimised assets. This also generates real-time data that improves infrastructure planning, reduces congestion bottlenecks, and informs long-term investment decisions. Across these combined effects, autonomous transport represents a multi-billion-pound opportunity that reshapes both public spending and economic performance.
In healthcare, the impact is immediate and human. A doctor who once spent 43 minutes a day typing now spends that time with patients, with Microsoft 365 Copilot trials demonstrating this at scale.⁵ Across 1.5 million NHS staff, this translates into hundreds of thousands of hours returned every month. This is not simply efficiency. It is capacity, fundamentally changing how care is delivered without increasing headcount.
The largest opportunity sits across the broader public sector workforce. AI has the potential to save around one-fifth of public-sector workers’ time, or conservatively one-sixth, equating to tens of millions of hours annually and a potential £34–41 billion wage-bill impact over time.¹⁰ Even partial capture of this productivity translates into £5–12 billion in realistic savings over a parliament. A civil servant, NHS worker, or DWP work coach gaining back two full working weeks per year fundamentally changes how the state operates.
In welfare and tax, AI is already transforming fraud detection, with £7.5 billion recovered through enhanced systems.⁸ This is continuous enforcement at machine scale, identifying patterns and anomalies in real time rather than after loss has occurred. It represents a shift from reactive correction to proactive prevention, reducing both cost and administrative burden simultaneously.
In courts and justice, where the Crown Court backlog exceeds 80,000 cases, AI tools such as document review, transcription, and intelligent scheduling are already demonstrating measurable impact. Ministry of Justice reform programmes show that digital case handling and AI-assisted workflows can significantly reduce administrative burden and accelerate resolution.¹⁴ This reduces both cost and delay, improving outcomes across the justice system.
Procurement remains one of the largest and most immediate opportunities, with £300–350 billion in annual spend. Even modest efficiency gains of 3 to 5 per cent translate into £9–15 billion in savings.⁹ AI enables continuous optimisation of supplier selection, contract management, and pricing, shifting procurement from periodic review to real-time discipline.
Across the government estate, energy costs of £14–15 billion can be reduced by 10 to 20 per cent through AI-driven optimisation.¹¹ This reduces waste while maintaining performance, delivering consistent savings at scale across buildings and infrastructure.
Local government and planning represent another significant opportunity. AI tools such as Extract are digitising planning systems at scale, reducing processes that once took weeks to minutes.¹² This accelerates housing delivery, infrastructure development, and local economic growth while reducing administrative cost. Across local government, AI can unlock substantial savings through improved case management, scheduling, and service delivery.
Demand Reduction: The Real Step Change
The most powerful savings do not come from doing work more efficiently, but from eliminating the need for work altogether. Artificial intelligence enables systems to reduce demand before it enters the system, fundamentally reshaping cost structures across public services.
In healthcare, predictive modelling reduces hospital admissions by identifying risk earlier and enabling intervention before conditions escalate. In welfare, systems prevent overpayments before they occur rather than recovering them afterwards. In justice, disputes can be resolved before reaching courts, reducing backlog and cost. In transport, fewer accidents remove entire chains of cost across emergency services, healthcare, and legal systems. In planning, improved decision accuracy reduces appeals and rework, eliminating layers of administrative burden.
This is how the cost curve bends. Not through efficiency alone, but through prevention, redesign, and the removal of avoidable demand. It is a fundamentally different model of the state, one that intervenes earlier, operates smarter, and spends less because it needs to do less.