A few days ago, I lit the fuse on a working prototype of a government service. No team, no procurement cycle, no waiting for approval. Just me, a few prompts, and a handful of AI tools. And honestly? Fireworks.
Vibe coding (or vibecoding) is an approach to producing software by using artificial intelligence (AI), where a person describes a problem in a few sentences as a prompt to a large language model (LLM) tuned for coding. The LLM generates software based on the description, shifting the programmer’s role from manual coding to guiding, testing, and refining the AI-generated source code.
Vibe coding, Wikipedia
I’m not new to what’s now being called vibe coding. Over the last year ChatGPT has helped me to bring a few random ideas to life1. Last weekend I thought I’d see what Codex CLI could do and I was again blown away. I mentioned this at work and in the conversation that followed we mused on whether some of the frustrations we’d been feeling could be shifted by trying the same thing there.
So I sat down with a laptop, some product instinct, and a handful of different AI tools. I wanted to see whether we could finally conjure the ‘fireworks’ we’d been waiting weeks to set off. I started with ChatGPT and the scale of the task was a bit intimidating. But then I remembered about Firebase and in minutes had something to show off. As I did, another colleague responded by asking if I’d seen Stitch, and another colleague said I should check out Jules.
And once I discovered Jules, that was when things got really interesting. Very quickly I had something live. Not a sketch or simulation, but something real. It’s up and running on Render (and I’d love to give you the link but I probably shouldn’t let it escape into the wild; at least not yet).
Obviously it’s just a prototype. But that also seems to do it a disservice. What is true is that it absolutely appears to do the job we had in mind. No engineers. No designers. Just me, some prompts and decisions, and it works, and it works in a way that will absolutely elicit the right sort of oohs and aahs.
I suppose I ought to make one small confession. I really shouldn’t have done any of this. Inside the department, everything except Copilot is blocked (and even then you only get Copilot on a Windows machine, not a Mac). Which means this burst of delivery joy has happened off network, off platform, and probably against better judgment. But that, too, is part of the problem. When the path of least resistance leads outside the system, it’s the system that needs fixing, not the people finding their way around it. Well, I would say that wouldn’t I?
Now, for our purposes as a team this exercise might be the perfect fireworks but more broadly for government, what are the repercussions?
I’m going to call it: Jules and Codex earn their hype.
One person can now create the fast, dazzling demos that make people believe again in what digital can do. When a non-technical person can build something tangible that users can interact with, the idea of specialisation for those who “do digital” starts to crumble.
But I sort of hope you read that and go, well that’s not news, that’s just underlining what we’ve always known. That digital isn’t a brick in the policy wall of government; it’s the mortar that binds the whole thing together.
What it means to build digital government maturity
Digital government maturity doesn’t come naturally, and it doesn’t come easily. In 2025 it might be taken as a given that countries would have a single government domain or a central coordinating unit but in the UK, and elsewhere in the world, those things have marked a decisive break with earlier ICT-era orthodoxy.
And achieving that is usually something of a battle. The entrenched interests are everywhere. You’re challenging procurement practices, funding models, organisational structures and culture, legacy technologies, etc. It’s a hard slog. You’re a small outpost of capability and talent; a pocket of resistance looking to achieve the wholesale transformation of government. A lonely, difficult place to be.
So steps are taken to scale that. One of those steps is establishing a professional identity: a shared name and a shared frame of reference. In the UK it used to have the catchy moniker of DDaT (Digital, Data and Technology)2. Capital letters to mean business. Pay frameworks to (slightly) address the shortfall to the private sector. Career paths to help people grow.
None of that was done because digital is niche. It was done because we want to recognise its worth and because we want it to spread.
But somewhere along the way, the thing we were trying to embed became the very thing we isolated. Instead of making digital universal, we build a perimeter. The hoped for saturation in the system became fenced off as a specialism.
And into that contradiction walks this new reality of vibe coding. Jules and Codex and whatever else are not the stuff of the future, they are the very real present. The barrier to building real things fast, really fast, has just been obliterated. And in doing so they challenge the idea of a digital profession.
Which is actually right in line with the original intent.
The point of digital transformation in government has never been about filling more digital roles for the sake of creating a stronger bloc of digital. It was always about transforming government itself. And you only do that if you transform every role.
During my time at GDS part of our theory of change centred on the rooms where decisions about technology were taking place. The world we were pushing back against was one where technology decisions were made by gatherings of policy, technology, commercial and suppliers. One person, maybe two, might have a grasp of digital government. The balance was off, and inevitably led to poor outcomes.
But while efforts were made to recruit a new layer of senior leadership, the long-term sustainable solution was never about replacing the people in the room. It was about shifting the balance of understanding. Replacing vendor dependency. Embedding new models for procurement. Bringing policy, digital and ops together. Training everyone.
The goal wasn’t to bring more digital chairs to the table to outnumber, it was to outgrow the idea that digital was someone else’s job.
A profession at odds with its purpose
That’s why, alongside the radical creation of the DDaT profession, came the equally radical ambition of the Digital Leaders network. Not a network of people with job descriptions that said “do tech” but a network of leaders, from up and down the hierarchy and across the profession spectrum, who understood what digital makes possible. 3
Those twin acts, profession and leadership, marked a shift from government as a consumer of systems integrators to government as a builder of its own services. It gave civil servants the space to learn, deliver, iterate. It seeded capability that departments had long since lost.
But a profession, once formalised, tends to consolidate. It builds barriers to entry, it defines its own language, it can grow defensive. Has what was intended as the source of transformation become the limit of it? Has it become easier for non-digital colleagues to defer than to engage? Is it easier to see digital as the job of someone else?
Setting DDaT headcount targets only reinforces the misconception: that digital is a profession to grow, rather than a mindset to embed. As if the key to digital maturity lies in one particular tribe. The truth is that the ambitions of digital government happen when 100% of public servants operate in digital ways, not when a small percentage carry the burden for the rest.
A broader baseline for everyone
I’m proud of the work we did at the OECD on the Framework for Digital Talent and Skills in the Public Sector. We made what I think is a clear and well-evidenced case for investment in three essential layers:
- Creating the environment that encourages digital transformation
- Equipping (and understanding) the skills that support digital government maturity
- Establishing and maintaining a ‘digital’ workforce
All three matter but we need to start by raising the floor.
A digitally enabled state only works if the society it serves is also digitally capable. We have to build a baseline where everyone, not just some, has the skills to thrive in the 21st century. That means confidence in using digital tools. Understanding the value of data. Having the ability to navigate services, raise concerns, and act online as well as off. Without that, no amount of GovTech polish will amount to meaningful transformation.
But the argument we made at the OECD goes further than that. Because just as we need to raise the digital baseline across society, we also need to raise it across the civil service.
Digital skills in government are not some exotic toolkit. They’re basic tools of modern administration: working in the open, using data well and handling it with care, designing with users, learning by iteration. These are not specialist superpowers. They’re things every public servant should be able to do. Here’s what we mean.
Skill | It means public servants can… |
|---|---|
| 1. Recognising the potential of digital for transformation | Spot where technology and data could re-shape their work, know the enabling guidance and platforms, and challenge the status quo. |
| 2. Understanding users and their needs | Identify real users, map end-to-end journeys, champion research, and design for everyone including people on the wrong side of the digital divide. |
| 3. Collaborating openly for iterative delivery | Work in the open, involve the public early, join multi-disciplinary teams and release small changes often. |
| 4. Trustworthy use of data and technology | Handle information securely, apply privacy law and data ethics, demand reliable uptime and build contracts that support those aims. |
| 5. Data-driven government | Find, share and reuse data, choose open standards, measure outcomes empirically and use evidence to improve policy and services. |
These five skills don’t just sit with product managers and engineers. They belong in every policy team, ops team, and leadership conversation.
Because a truly digital civil service isn’t one where we’ve filled all the DDaT vacancies. It’s one where public servants everywhere work with these tools and these mindsets because they’re part of the way government gets things done.
That idea, that digital competence is foundational and not optional, is echoed in the OECD’s data-driven public sector work, where raising the digital floor across civil service and society is the prerequisite for meaningful transformation.
How we got distracted
At the start of this government Labour made a bold move. They transferred GDS into DSIT. And with it came a break from several years of meandering and stagnation. Here was the prospect of rekindling the centre as a place of strong leadership and excellence. And here perhaps was the recognition that digital was infrastructure, no longer a special case, but something embedded enough to move from the centre to the fabric of everyday government. My greatest hope was that this reflected a more digitally native cabinet and that we were moving beyond “digital as special” to “digital as standard.”
But, unfortunately, I’m not feeling it.
The state of digital government review says sensible things. The blueprint isn’t wrong. But this month GOV.UK will undergo a horrible and unnecessary rebranding4 and there will be a GOV.UK app, that seems to exist because of a focus on the needs of a generation whose oldest members haven’t finished their GCSEs5. Are those the priorities when 7.3 million UK adults lack the ‘Essential Digital Skills for Work’?6
It’s hard not to feel that the leadership we’re getting, isn’t the leadership we need. It’s digital as new, shiny, performative. It’s not fixing the plumbing and doing the difficult things that have been put off for too long.
And while some of that plumbing is hardware and software, the actual infrastructure that underpins transformation isn’t rare earth metals and code. It’s people, and a culture that enables them to use their skills most effectively. It always has been.
And unfortunately, the infrastructure that sat behind a whole of government embrace of what digital makes possible has been dismantled, and not resuscitated. Communities of Practice have withered without support. The Digital Academy is gone. GOV.UK PaaS has been shuttered. What remains are pockets of brilliance with no shared scaffolding to stand on.
We are, once again, in danger of making transformation look like a job for experts. And I say that as someone currently employed as a “Transformation Lead”, a role that only exists, ironically, because we haven’t yet normalised the idea that everyone should be doing this work.
And in doing so, we forget what made the early years work: that digital wasn’t the job of a few. It was the environment we built together.
From vibe to velocity: what’s stopping us from shipping?
So, how quickly is this going to change everything? Probably not that soon.
Because even when you’ve got something live on your laptop, getting it into the world is another matter entirely. Deploying it in a way that’s secure, sustainable and governed still depends on decisions outside your control. The infrastructure for ‘just putting something online’ in tangled and often inaccessible. And in many cases that’s perfectly legitimate.
But last week’s experience speaks to a profound change. It is now possible for someone like me, who thinks in product terms, who’s built up a practice of articulating user needs, defining outcomes, and prompting well, to go from zero to a working service prototype in a matter of hours.
Not a wireframe. Not a sandbox. A thing that works. That can be hosted. That solves a real problem.
Of course what Jules built with my encouragement isn’t integrated to the morass of internal government messiness. But maybe that’s something else it can help with. I wonder how Jules feels about COBOL.
But this didn’t happen because I hired a team. Or got permission. Or waited for procurement. It happened because I used public tools (including the amazing GOV.UK Design System), publicly available data, and (currently) some tools that cost little more per month than my weekly edit: daily commute (though no way does Jules stay free for long).
Vibe coding doesn’t just highlight what’s newly possible, it exposes what’s still in the way. You can have the instinct, the skills, and the tools, and still hit a wall. Because while the culture isn’t yet permissive enough, and the workforce not yet widely skilled enough, the third leg is also broken: the infrastructure that makes it easy to test, learn and ship safely. All three: culture, capability, and infrastructure, have to move together.
That same triad—people, process, platform—underpins how we build trust into public sector AI: not through compliance checklists or clever tooling, but by embedding governance, ethics, transparency and skills at the heart of how we deliver.
Infrastructure as gatekeeper: the PaaS-shaped hole
And that brings us to one of the most frustrating barriers: getting working ideas into the world. Even when you’ve got something live on your laptop that’s robust, usable and safe, deploying it in a secure, sustainable, governed way still depends on infrastructure. The kind GOV.UK PaaS used to offer, until it was shuttered.
I know from 2016, when we were incubating it and its fellow Government as a Platform common components, that it was never an easy sell – not like sending messages and taking payments. But we always recognised its value for the long tail. For serving that focus on how we could enable and equip teams to focus on meeting the needs of their users more quickly. It was helping to give teams, particularly those without large platform engineering budgets, a way to test, learn, and deploy early. It supported innovation not by offering shiny tools, but by removing unnecessary blockers. It recognised that strategy is delivery, and that delivery depends on making it easier to ship working software. Fast. Safely. Publicly.
It’s a decision that was made a while ago now and I get that I must sound like I’m re-litigating the past but shuttering PaaS didn’t just cut off access to hosting. It revealed the absence of a wider vision, often present in other countries around the world where data-driven public sector leaders are embracing interoperable registers, APIs, standards, and federated platforms that empower teams to focus on user needs, not internal plumbing.
Losing PaaS embodies the strategic and visionary failure to see what it would take for digital teams across government to move at pace. To test new ideas. To show real things. To be trusted. And it was that unleashing of potential, not bits of commodity tech, which was (and is) the point of anyone ever thinking about Government as a Platform. It’s not the same stuff, more quickly, it’s about fundamentally creating the momentum and opportunity to do government differently.
And I bring infrastructure up in a piece about the DDaT profession because I wonder if this is the ultimate form of gatekeeping. Not just who gets to build, but who gets to deploy. If we’re serious about transformation, the role of the digital centre isn’t to oversee everything. It’s to enable and empower the capability already woven throughout the civil service by building infrastructure that trusts people to act, and removes the friction that slows them down
When I decided to see where AI could take me last week it was as much out of frustration as curiosity. We’ve done plenty of good research. We’ve fleshed out a solidly ambitious roadmap. Our thinking is sound and the service vision is compelling. But we have really struggled with recruitment and building out the team to have something to show for it. And understandably that meant confidence was ebbing – in us, in the vision we cast, in agile and user centred design.
Fireworks help to paint a vision of a future. Vibe coding seems to mean that those displays are more like repeatable, reprogrammable drone shows than single shot rockets.
This isn’t a marginal shift. It is a productivity shock to the assumptions that underlie the DDaT profession. Not because digital roles stop being useful. But because the notion that digital must always be channelled through them becomes hard to sustain.
The shape of what’s next
Vibe coding is not magic. For my own sense of worth I don’t think just anyone could have done what I did. I’d like to think I achieved something good because I brought my years of experience to the task. But that’s kind of the point. The new scarce skills are not engineering, but product sense. Not design craft, but ethical clarity. Not deep stacks, but wide understanding. The sort of thing generalist civil servants can be quite good at.
If a digitally minded policy official can get from idea to prototype in a couple of days, the job of the digital specialist shifts. Less gatekeeping, more coaching. Less holding the pen, more shaping the room. Digital transformation becomes less about delivery teams showing others how it’s done, and more about public servants everywhere being equipped to do it themselves.
But do you know what? That has always been the work. The culture of digital in government seeded by UKGovcamps and made mainstream by OG GDS was always about an open invitation, not a closed specialism. To show, not do. To teach, not tell. To deliver, not dream.
And that’s still the opportunity. But let’s be honest about what’s missing. Without safe, scalable platforms to host and share these things, even the most promising prototypes remain trapped. Without leadership that models digital behaviour, not just sponsors it, permission shrinks. Without universal skills, trust in agile, and visible delivery, we revert to old habits.
So yes, I’m excited by what AI tooling enables. But I’m also clear-eyed. If we keep treating digital as a brick to be stacked rather than the mortar that holds government together, we will miss the moment. And if we let the profession become the product, rather than the people it was meant to serve, we will waste the potential these new tools unlock.
What I reckon we need to do
We need to re-anchor the idea that digital is for everyone. That it is the condition in which good government operates. That vibe coding isn’t a trick, it’s a glimpse of what a digitally capable civil service could be if we gave it the tools, and the trust.
We need to teach those five digital government user skills. Not to some, but to all. We need to rebuild infrastructure (of people, culture and technology) not out of nostalgia, but because it worked. We need to stop asking how many digital roles we’ve filled, and start asking how many policy and ops leaders are delivering in digital ways.
And we need to bridge the chasm.
That deep and persistent disconnect between digital, operations, and policy. The real risk isn’t just that we underinvest in digital it’s that we let its logic remain separate. Service designers over here, policy drafters over there, operations downstream. That logic is broken. And tools like Jules, like vibe coding, make that fracture visible. Fixing it is now urgent.
But the solution isn’t just skills or tooling. It’s culture. Because the real risk of everyone being able to build is that no-one is incentivised to build well. If we build prototypes to satisfy the demand for announceables, or see velocity as virtue in itself, we hollow out the very promise of transformation. Good digital practice doesn’t just mean building any old thing, it means building the right things, the right way, with users at the centre and sustainability in mind. That’s a cultural choice, not a technical one.
So this isn’t just about what happens inside government.
If we want the public to understand and trust digital services, we need public servants who live and breathe digital understanding. If we want society to have the skills to adapt to this new wave of tools, we need government to model what digital confidence, and digital integrity, looks like. The gap between government capability and societal digital inclusion is shrinking. Both must rise together.
We need to build a culture where people can try. Where alpha doesn’t mean fragile, but purposeful. Where people don’t just ship fast, but ship responsibly. Where fireworks aren’t locked away, but opened up for everyone to show off with, yes, but also to learn from, to test with, to improve together.
The future is being vibe-coded already. And the best way to meet it? Light the fuse, and let people build.
- Most notably what is becoming PrayReps, but also the Alt Tagging plugin for this website, and a couple of other things that I should blog about but haven’t. ↩︎
- The idea that digital in government needed to have its own ‘brand’ and so DDaT should become Government Digital and Data, as though that was something worth spending time on, wound me up to a surprising degree at the time, and I obviously can’t say I’m on board with it today. ↩︎
- As an aside, I’ve just finished some work with the Kosovan government, and it was so encouraging to see a small country setting a new expectation: that all leaders will engage with digital transformation, not just the ones with IT in their titles. They get it. ↩︎
- With a highly curious choice of colours, compared to the iconic black and white, of something that puts me in mind of a couple of political parties ↩︎
- And which will require you to authenticate via One Login before you can use it to browse GOV.UK. Because nothing says public service like putting a padlock on the homepage of government. ↩︎
- The 2024 Lloyds Consumer Digital Index reported 12.1 million UK adults (23%) with low or very low digital capability, and 7.3 million adults in the labour force lacking the Essential Digital Skills for Work. ↩︎