bmwelby's blog

The infographic presents statistics related to trust issues in Britain. It states that 20% of people felt they had a say, which has increased to 34%. Meanwhile, administrative service satisfaction decreased from 74% to 65%. The headline highlights the complexity of Britain's trust problem, and there is a question prompting curiosity about what has changed and for whom. Design elements include upward and downward arrows indicating changes in statistics.

Britain’s trust problem isn’t what it once was

September 11, 2026 / Benjamin Welby / 0 Comments

Earlier this week I was having a conversation about data centres, AI, and the fairly febrile state of mind surrounding both of them at the moment.

There are sensible arguments to have about where data centres go, what they require and who benefits from them. There are even bigger arguments about what AI might do to work, to public services and to the institutions we rely on. But both also seem to have become lightning rods for a wider uncertainty about where all this is heading, who is in control of it, and whether the people making the decisions can be trusted to make them well.

And in the course of the conversation, my colleague Barbara asked whether I’d had a look at the most recent OECD work on public trust.

I hadn’t. Which was a surprise given that public trust is something I come back to quite often.

It’s central to the thinking I’ve done about data in government and whether people feel they have any meaningful say in policymaking. It turns up in apparently narrower questions, like rolling out CivWifi for all, and in much bigger ones about what sort of institutions are trusted enough to carry us through whatever AI does to the world of work.

And yet, despite that, I’d completely missed the fact that the OECD’s latest edition had come out. I hadn’t seen anyone give it much coverage either. Maybe it simply passed me by over the summer.

But I think back to the 2024 edition, which also seemed to pass without much comment, and I think public trust is hugely underweighted in the way we talk and think about the challenges facing the UK and elsewhere. If I’d been in any of the conversations about the missions (RIP?), I would have called rebuilding public trust one of them. Not because trust is an end in itself, or because everybody ought to agree with government, but because as we’ve seen, it’s hard to do very much that is difficult, disruptive or long term if large numbers of people start from the assumption that government is incompetent, dishonest, unfair, uninterested in them, or some combination of the four.

AI is already a trust question

Anyway, back to the trust survey. For the first time, the OECD asked people what they expect if government agencies use AI. The British results do not say that people are simply against it.

They are more uncertain than that. 39% are confident AI would produce more tailored services and 36% that it would reduce government costs. When the questions turn to protecting personal information, keeping humans in control of important decisions, transparency and fair treatment, confidence sits between 31% and 35%.

But on all six measures Britain is less positive than the OECD average.¹

A man in a suit is speaking animatedly on the left, while a figure portrayed as Jesus looks on thoughtfully on the right. In the background, the iconic Big Ben clock tower and other structures from London can be seen. The text at the top reads "SUICIDAL EMPATHY?" in bold letters.

What if Robert Jenrick is talking about Jesus?

August 26, 2026 / Benjamin Welby / 0 Comments

A note before you go further: I’m conscious that this piece plays deliberately with a word that describes something devastatingly real.

Suicide is not a metaphor: around 125 people die by suicide each week in the UK, and around three quarters of them are men. Suicide is the single biggest killer of men under the age of 50 in the UK.

There are amazing organisations doing amazing work to try and change those stats including Movember every year, and CALM, which offers free, confidential support.

Robert Jenrick thinks Britain has a problem with “suicidal empathy”

He means something more than empathy being partial, or governments having to think about incentives and finite resources. His accusation is that we become so preoccupied with somebody else’s circumstances, so willing to prioritise their needs, that eventually we imperil ourselves.

And the trouble is, looked at from a certain angle, that sounds rather a lot like Jesus.

I’ve wondered before what would happen if Robert Jenrick hung out with Jesus for a bit¹. Not because I imagine Jesus quietly handing him an upgraded policy platform, but because spending time with Jesus has an awkward habit of giving us perspective when we’re adamant that we’ve got all the answers.

And now he’s made a claim that brings me back to that question.

As Reform launched proposals it claims would cut around £50 billion a year from welfare spending, including large reductions in disability benefits², Jenrick described current disability spending as “not generosity … it is suicidal empathy”.

Try it on Jesus

So what has that got to do with Jesus?

Jesus has an almost reckless tendency to come too close.

People interrupt him and he stops. He touches people everyone else has decided not to touch. He eats with the wrong people. He lets another person’s grief become his grief. He weeps. He tells stories in which the respectable people walk past and the foreigner is the one who comes near. He tells us to love enemies, to give without always expecting something back, to forgive again and then again and then again after that.

He leaves the ninety-nine to go looking for one.

This is not obviously the behaviour of somebody committed above everything else to prudent self-preservation.

And eventually it gets much worse. He goes to Jerusalem. He knows what is waiting, and he refuses the obvious routes of self-protection.

And he says this utterly upside-down thing: whoever tries to save their life loses it, and whoever loses their life finds it.

If “suicidal empathy” means allowing someone else’s predicament to become so thoroughly your concern that you willingly accept a cost you didn’t have to bear, then the accusation against Jesus is a strong one.

A smartphone is centered in the image, displaying a screen with a list of app features. Colorful icons representing different functionalities are connected to the phone with pipes, set against a blue background. Text on the left reads, "Apps are pointless unless you actually app with them."

Apps are pointless unless you actually app with them

August 17, 2026 / Benjamin Welby / 0 Comments

Last week A-level results dropped.

And a video popped up in my YouTube subscriptions from GOV.UK suggesting that students download the GOV.UK app to search GOV.UK for what to do next.

It’s a bit odd when we already have an app for searching the web. A browser.

OK, I accept that’s a cheap shot from someone who is GOV.UK-app-sceptical at the best of times. So let me acknowledge that the published ambition is bigger than search: personalisation, proactive support, services inside the app and agents that help people do things.

But the video got under my skin because it’s just showing off search behind authentication inside an app.

Search is useful. Guidance is useful. The app needs to offer search. But if the selling point on results day is that the app does something any browser already does perfectly well, without the hassle of a download and a login, then let’s be honest: that’s a weak answer to the question of what the app is actually for.

Authentication is useful if what you get in return can actually know something about you, remember what you were doing and help you get on with it. If all you get is public search, it is mostly friction.

But the thing bothering me isn’t really that GOV.UK has put search in an app.

It’s that the app gives us a good view of the much more interesting problem underneath. Government increasingly has bits of information, infrastructure and services that could make somebody’s journey through a moment like results day genuinely easier. Yet the person still too often experiences those bits as separate institutions, separate forms and another search box.

So this is really a post about the plumbing, and what becomes possible when we finally join enough of it up.

Obviously it wouldn’t have helped anyone to watch a fake video of a future that doesn’t yet exist. But there is a good future that could exist.

And results day is a good test because results aren’t just information; they’re an event.

These results trigger things. Lives change on results day. A university offer became a confirmed place, or it didn’t. Clearing suddenly mattered. That might change student finance, or make a different apprenticeship or job (im)possible. The interesting thing about the grade is what it unlocks.

That’s the bit I’m interested in. Not putting the result somewhere digital for the sake of it, but allowing something newly true about somebody to change what they need to see and do next.

So, am I saying the GOV.UK video should have shown results being revealed in the app?

On that I’m not so sure.

Five things I think about GDS, data and AI being split up again

July 22, 2026 / Benjamin Welby / 2 Comments

When the first briefings appeared over the weekend about abolishing the Department for Science, Innovation and Technology (DSIT), I decided to wait.

Machinery of Government changes have an awkward period in which the headline is certain but most of the consequential detail is not. Which teams are going where? Who owns the budgets, controls and responsibilities? Which minister is accountable? What has actually been abolished rather than renamed or moved? And what is this all supposed to make better?

The dust has not entirely settled and there is still detail to follow. But we know that the Government Digital Service (GDS) is moving into the Department for Digital, Culture, Media and Sport (DDCMS). We also know that responsibility for AI is being pulled towards the new Office of the Prime Minister and Cabinet, with the Minister for AI attending Cabinet.

The reason there’s so much dust to settle is that this is more than blowing the bloody doors off.

I can’t say that bodes well.

The other reason not to dive in was that wise people were quick to say much of what needed saying.

Gavin Freeguard’s original thread and subsequent update have kept track of what is being announced. Tom Loosemore has been characteristically clear that rearranging departments is an unserious use of a government’s limited time, and about the dangers of separating AI from the platforms and data it relies on. Paola Roccuzzo’s roads and buses analogy gives the infrastructure argument to a prime minister who understands devolution. James Plunkett found a good idea within the bad one: turning digital government outwards towards places, local innovators and frontline services. Dave Rogers reflected on what happens when moving the machinery of government. Theo Blackwell asked whether failing the place test was part of DSIT’s undoing, while Sarah Munby has pointed to the hard graft involved in building the thing now being dismantled.

Sam Freedman has offered an important counterargument to the idea that this is simply random rearrangement. His case is that Burnham has a theory of power: a stronger prime-ministerial centre, a clearer set of priorities and a deliberate attempt to put more strategic and delivery capacity under his own control.

His piece is a useful warning for me. When digital government is your hammer, a lot of things will look like nails. It would be wrong to assess the whole Burnham project through what is happening to digital, data and AI, or to treat either continuity or change as inherently good or bad.

There is continuity I welcome. I’m pleased that the work Alan Milburn has led on young people and employment is continuing. I wrote about the importance of its diagnosis in Strong branches and good shade, and I’m pleased too that Pat McFadden remains at DWP to oversee the emerging delivery of the reimagined jobs and careers service (and I assume skills brief is staying there too and is not on the move again).

Mind you, there is continuity I am less enthusiastic about, including an effective endorsement of the Home Office’s thinking behind “earned settlement”, which left me pondering what it tells us about belonging, character, and the country we are becoming.

So, as I now join everyone else who has committed key to board, I’m trying to ask a narrower question: does an apparently serious theory of political power include a sufficiently serious theory of how an AI era state will operate?

A cozy living space features a wooden armchair with a small device resting on its seat. To the left, a dark bench contrasts the light-coloured chair. A lush indoor plant stands nearby, and a warm lamp casts soft lighting in the room. Text in the image reads: "Whatever is nearest. Whatever is open. Whatever replies."

Whatever is nearest. Whatever is open. Whatever replies.

July 19, 2026 / Benjamin Welby / 0 Comments

On Thursday 9th July 2026 I sat in the audience for the launch of Mind’s Mental Health in the Age of AI Commission. Here are some thoughts provoked by that evening.

On the face of it, “AI and mental health” sounds like it fits neatly around a panel and a commission. In practice, it spills out of health care into a messy overlap of consumer protection, online safety, public-service design, and the raw capacity of health services. The tidy categories of policy do not hold up particularly well against the lives people actually live.

After the panel I joined a breakout group focused on regulation. The conversation flowed well and we discussed the Online Safety Act and the boundary of responsibility between users and systems; where the MHRA’s role starts and stops; how to restrain actors who have no intention of behaving well, and what to do about those who say the right things while taking unacceptable risks.

And one conclusion seems hard to avoid: AI is already part of our mental health ecosystem, whether or not the tools being used describe themselves that way.

It fills the gaps left by waiting lists. Someone might use it to prepare for a GP appointment. A conversation about employment might be rooted in the ways their mental health is a work-limiting condition. Someone who doesn’t think they’re ill enough to ask for help, is too ashamed to ask, or has asked before and had a bad experience, might well be typing instead.

So arguments about whether people should use AI for mental health are largely academic. The reality has outpaced them: people are using it.

The harder question is what happens when that use takes place through tools and services that do not think of themselves as mental health services at all. Because “AI” isn’t one thing. It ranges from regulated clinical technologies and purpose built tools to general purpose chatbots and AI embedded in services designed for something else. Their risks are not identical, but people move between them far more easily and quickly than regulation does.

Mind’s Commission begins in that blurred reality, asking how AI is changing help-seeking and support pathways, how people can be kept safe, and what it means for inequality in access and experience.

A dimly lit restaurant kitchen features three elegantly plated dishes on a counter, each garnished and ready for service. In the background, two chefs in white uniforms are seen working, focused on their tasks amidst the bustling kitchen setup. The atmosphere is lively, with various kitchen tools and equipment visible.

When the first plates leave the pass

June 16, 2026 / Benjamin Welby / 1 Comment

There are some jobs you leave because the evening has ended.

The plates are cleared. The conversation has run its course. Someone is stacking chairs. It was good, perhaps even brilliant, but it’s clearly time to go home.

Leaving DWP has not felt like that.

The image I’ve had has been of walking out of a busy kitchen in the middle of service. Everything simmering away nicely on the stove. The ingredients good. The chefs knowing what they are doing. The smell promising. And yet, for reasons that are right and good and exciting, and of God, leaving before dinner was served.

Except that, last Monday, the metaphor shifted slightly.

Because some of the first plates are now leaving the pass.

The Prime Minister used his London Tech Week speech to announce a new AI jobs tool to help people out of work find the right jobs, create their CVs and get back into work. He was referring to the new experimental Work Hub, with its AI work assistant, CV builder, action plan, jobs board, and support for employers.

Some of this is new. Some of it is remaking things government has had around it for a long time. Some of it is very much at the start of its life. That’s why the experimental badge is important. It’s also why it’s brave. DWP has made something visible early enough for people to use it, test it, misunderstand it, criticise it and improve it.

So have a play, test it to the limits of your amusement, and maybe you’ll be pleasantly surprised, maybe you’ll learn something actively helpful.

Still, given what is being imagined for the eventual menu, this is very much the hors d’oeuvres.

But I’m no longer in the kitchen.

I’ve taken off the apron, stepped into a different kind of work, and now find myself watching the first plates appear with a strange mixture of joy, pride and longing.

A young man and an older woman look at a community noticeboard outside a local hall. The board suggests many forms of participation beyond paid work, including local groups, volunteering, learning, care, civic activity and community support.

If less work is the dream, why does it feel like a threat?

June 1, 2026 / Benjamin Welby / 0 Comments

About a year ago I wrote about Vibe Coding, Fireworks and the Mortar of Government, back when I was first getting excited by the new art of the possible. Since then I’ve had a lot of fun with these toys, at home and at work, while learning a lot along the way.

That fun left me thinking about what it all might mean for public sector product management and how we organise our work. I’m not going to spend this post re-running the argument that sets out why more members of the multidisciplinary team can now directly shape the product, or getting excitable again about what happens when the distance between intent and something runnable collapses. If you want to read that, you can do that in The Future of (Public Sector) Product Management in a Vibe Coded World.

This post finds its origins somewhere else: in a throwaway comment someone made that, by making those arguments, I seemed to be excited for the most pessimistic future of the labour market.

I’d rather think of it as being optimistic about being pessimistic. And hopefully by the end of this you’ll see what I mean.

Because this is a post that asks what follows if the argument is even half right.

I’m not here to make any kind of prediction that all office jobs will disappear by Christmas. I am not trying to join the queue of people announcing that everything has already changed.

But I think we should sit with a possibility.

What if AI doesn’t destroy work, but shrinks, thins or rearranges it enough to expose how much our social fabric relies on a particular framing of work in order for it to function?

A woman in a long, patterned skirt sits on a stone bench under a large tree, while a younger man in a light-colored outfit sits next to her. They are engaged in conversation, with the tree's lush green leaves providing shade. In the background, there is a brick built church and grassy area. The scene is peaceful and inviting.

Strong branches and good shade

May 28, 2026 / Benjamin Welby / 2 Comments

A report that should arrest us

This morning Alan Milburn launched the Young people and work interim report. I tuned in to hear his speech and the follow-up Q&A, and I’d recommend you find the time not only to read the report but to listen to him speak to it.

It is arresting stuff. Not only because of the scale of the crisis being described, though the scale is staggering, and not only because of the policy detail, though there is plenty there worth wrestling with. But what especially struck me today was the moral seriousness with which Alan spoke about young people who are quietly disappearing from participation in the life of the country.

Over the last phase of my time at DWP, I shared my time between the team supporting this review and the jobs and careers service. Mine was a small contribution in the scheme of things, but enough to see at close quarters the seriousness, care and determination Alan and his team have brought to it.

It also means I am not a wholly impartial observer. I want this report, and the final report coming later in the year, to be read carefully, argued with seriously, and met with a level of response that matches the gravity of what it describes.

Because this is a report that asks important questions:

What happens to a society when more than one million young people are outside education, employment or training?

What happens when the pathways into adulthood no longer reliably function?

What happens when institutions each hold fragments of a person’s story, but nobody holds the whole?

And perhaps most importantly: what kind of country are we becoming if we learn to normalise this?

The review’s argument is not that this requires a temporary focus or a short-term policy tweak around the edges. We’ve been here before, oscillating between panic and amnesia. We briefly notice a crisis, announce something, fund something, brand something, and then move on before anything has changed.

No, the argument is that there has been a structural break in how young people move into work, confidence, belonging and adult life, and therefore structural change has to be the response. And that means harder questions to answer about institutions, relationships, participation, and the kind of social fabric that is required for young people to flourish.

This should arrest government.

But that is the minimum we should expect from a report like this.

Here’s one weird trick that could save Keir Starmer

May 12, 2026 / Benjamin Welby / 0 Comments

Editorial illustration of UK electoral reform: an ink-splattered constituency map of the UK, breaks into coloured pieces while ballot papers flow from a cracked ballot box towards citizens gathered around discussion tables. In the background there is the Palace of Westminster and in the foreground a multi-coloured proportional seat chart suggests a more representative voting system.

I am sorry about the title.

Not very sorry, obviously. But sorry enough to acknowledge that “one weird trick” belongs more naturally beside adverts for miraculous belly-fat solutions, obscure pension loopholes, and things dentists apparently hate. And that it is not usually the hook into a conversation about democratic renewal.

Still, I think there is this weird trick available to Keir Starmer.

It is this: govern.

Govern with the majority he has to begin fixing one of the things that is most obviously, repeatedly, structurally broken in British politics. Not by avoiding disagreement. Not by finding the correct grid of announceables. Not by setting up the right combination of technocratic delivery boards. Not by locating the mythical median voter hiding somewhere between a focus group in Nuneaton and a laminated pledge card.

Govern with the majority he has to fix British politics by making votes mean something closer to what voters intend them to mean.

In other words: electoral reform.

I know. I know. Nothing says “urgent political response to last week’s electoral earthquake and this week’s leadership crisis” like constitutional reform. I know you and everyone else is crying out: at last, regional deliberative democracy methodologies.

And yet. It kind of feels like precisely the moment when you swing for the fences and you say the bigger thing.

Because the noise now is all about Starmer. Whether he can survive. Whether he should survive. Whether the speech worked. Whether the cabinet is wobbling, marching, whispering, briefing, plotting, or merely standing near a door with an unusually thoughtful expression.

This is Westminster’s favourite kind of drama, because it can be reported almost entirely through human weather. Who is up. Who is down. Who is loyal. Who is “loyal”. Who is thinking of the party. Who is thinking of the country. Who is thinking of the country in a way that it just so happens involves them becoming prime minister instead.

And some of that matters. Of course it does. Leadership matters. Judgement matters. Political authority matters. If a prime minister cannot command confidence, that is not a trivial problem.

But the problem runs deeper.

The danger is that the Labour party convinces itself that this is fundamentally a personnel issue. Change the leader, change the mood, change the story. A new face, a new operation, a new grid, a new solemn promise that lessons have been learned and listening has occurred.

Perhaps that would help. Perhaps it would not. But I reckon the forces now swirling around Starmer would swirl around whoever succeeded him.

Because I don’t think last week’s elections were really simply a verdict on one man.

I think we’ve been getting closer to this sort of outcome for years.

The electoral map was wobbling. Now it has toppled.

Reform is no longer merely an irritation on the right of the Conservative Party. The Greens are no longer merely a place for protest votes in seats where everyone already owns a bicycle. The Liberal Democrats continue to possess their strange and enduring ability to appear locally inevitable while nationally implausible. Independents, local parties, national parties and post-party moods are all tugging at the fabric.

The country is not behaving like a two-party country.

It has not done for some time. But last week made the point harder to avoid. Voters are moving in different directions at once. They are not simply swinging neatly from red to blue or blue to red, like a polite constitutional pendulum maintained by a retired civil servant in a cardigan. They are fragmenting, protesting, experimenting, hedging, punishing, searching.

And Westminster is still trying to cram that into a machine built for a different political world.

Title slide: "The Future of (Public Sector) Product Management in a Vibe-Coded World," authored by Ben Welby. The background features a subtle wave design, adding visual interest.

Beyond the vibes

February 26, 2026 / Benjamin Welby / 3 Comments

Last June I wrote a reflection on how vibe coding had made it possible to create some fireworks that helped bring our vision for the future of public employment support to life. In the months since, those fireworks have become a portfolio of provocatypes, and a lot of learning and a cascading number of new ideas.

And it all has the feeling of standing close to fast machinery. You’ll know what I mean if you’ve ever been to an old mill turned industrial museum where the looms are up and running. Powerful, amazing, machinery. That could absolutely rip your arm off.

It’s hard to imagine what it would have been like when Edmund Cartwright first showed off his power loom, let alone what it was like when mill after mill was filled with them and all you’d ever known was hand looms and spinning wheels.

AI-assisted delivery and the craft of confidence

But I think if you’ve spent any time with AI-assisted development tools over the last year and a bit then you’ll know what I’m talking about. Maybe you’ve only just started with the new models in the last few weeks and you’ve read Matt Shumer’s widely shared Something Big is Happening piece which says things I can absolutely identify with. Or maybe, like me, this has been a slower burn and for a while now you’ve had that intoxicating mixture of exhilaration and dizziness.

Either way, I do think we are past an inflexion point like the one that greeted the start of the 19th century: a kairos moment for digital delivery where the nature of what it is to work in digital government has changed shape.

There’s a popular term for this: vibe-coding. But as I explore in the chapter why vibe-coding matters (and what it isn’t), the phrase is really the hook, not the substance. This world where we can use natural language to describe what we want and have working software generated moments later. But there is a tension in that language. It immediately sounds like flimsy demo-ware, and like a rejection of discipline or craft.

Vibe coding: the hook, not the point

And when it comes to building services that respond to the needs of the public, vibes aren’t enough.

So some months ago now I started to try and write about what this might mean for our disciplines and our craft. It has taken me longer than I wanted, but I’m happy enough with them to publish my reflections on on what changes for product management in a vibe-coded world — and the broader argument collected at https://vc-product.wel.by/.

A small warning: it’s long. I think it deserves your time (which will be much less than the time I gave it), because it’s trying to take seriously what AI-assisted delivery changes, and what it doesn’t. So I’ve woven in links to take you into the dedicated chapters where I try and unpack each idea. You’ll see there’s breadth but that’s because the implications are broad: for how we work today, and how we’ll need to work tomorrow.

But while vibe-coding is the hook, I think we are really talking about something that needs its own name.

Andrej Karpathy, who coined the original phrase, has suggested ‘agentic engineering‘, and that’s useful because it insists that this is becoming a professional workflow: you orchestrate agents, you scrutinise output, and you keep the quality bar intact. That makes it serious rather than a good party trick.

But it’s not the frame we need when it comes to our public services.

‘Agentic engineering’ names how software gets written. But we’re grappling with how public value gets added. In government (and more than likely elsewhere too), the unit of delivery is not an individual with a clever workflow – the unit of delivery is the multidisciplinary team, and the enabling environment wrapped around it. It’s policy, ops, analysis, content, design, engineering, and their leaders getting closer to runnable reality sooner, together. That’s why in my writing I’m using AI-assisted delivery.

AI is already a trust question

Try it on Jesus

A report that should arrest us

AI-assisted delivery and the craft of confidence

Vibe coding: the hook, not the point

Never miss a new post

Tags