How we organise when building is cheap
Estimated read time: 6 minutes
If building becomes cheaper and faster, the question isn’t how to spend less on people – it’s how to spend differently on capability. Badly-built software is a drain on government: brittle services, repeated rebuilds, solving the same problems multiple times. The answer remains: fund teams, not projects – durable, multidisciplinary teams who own a problem, learn continuously, and keep improving the service they steward over time.
What changes in a vibe-coded world is the shape and rhythm of that organisation. Teams no longer need to be huge to have impact. With the right mix of skills, and good platforms beneath them, a smaller cross-functional team can deliver at a pace that used to demand a programme. The goal isn’t to reduce the size for the sake of it but to make every pound and every hour count towards a better outcome.
Two models that work together (standing + mission)
When we can have impact in short bursts, how do we avoid defaulting to short-lived project teams? I suggest we need two models that work together:
- Standing service teams that own and evolve a product over time,
- Mission teams that surge around specific outcomes or policy moments.
The standing team holds the continuity – reliability, feedback, iteration – while the mission team brings some extra focus to seize an opportunity or unblock something, then folds its learning back into the core. Neither replaces the other. Together they form an ecosystem of continuous service stewardship and adaptive delivery.
This isn’t the return of “keep-the-lights-on” skeleton crews. Quite the opposite: the standing team has to be set up to iterate, not merely maintain. It’s where institutional memory lives, where user feedback lands, and where change accumulates safely. Its health determines whether a service matures gracefully or fossilises. That means giving those teams stable funding and permission to keep improving, not just to keep running.
At higher speeds, this could be how we get speed without chaos: surge capacity without organisational amnesia. Missions give you a way to concentrate effort on what matters, without constantly rebuilding the scaffolding of a programme. And they stop specialist disciplines being treated as intermittent obstacles. Accessibility, data, legal, security – these are not people you contact at the end. They’re perspectives you need alongside you while you’re still deciding what the thing is. Again: not a new idea. Just multidisciplinary work, practised properly.
Funding that follows evidence
And when building becomes cheap, do we also get the freedom to match our funding models to the same logic? Instead of having to secure huge upfront budget for a multi-year plan (with all the expectations and rigidity that comes with it), can we move to greater tranche-based funding: funding released in slices, tied to evidence and outcomes. Fund a discovery and alpha mission; if it yields a working prototype and promising data – improved takeup, reduced processing times, fewer avoidable contacts – then release the next chunk of funding to scale and harden it.
Funding becomes iterative, just like development.
This reduces waste – we stop pouring money into ideas that don’t survive contact with reality – but it also changes the discipline of decision-making. Spending can follow proof of value: a working thing, credible measures, clear (and openly published) stories about what we are learning. That is easier to govern well. Because funding is released against evidence and a model of assurance that happens continuously, in proportion, and without waiting for a single all-or-nothing business case or a spending review cycle.
Platform as tarmac (not just tooling)
All of this depends on the health of our shared digital infrastructure – but we should be careful what we mean by ‘infrastructure’. In a vibe-coded world, the limiting factor is not compute or code. It’s whether teams can move quickly on a smoothly tarmaced road: clear standards, shared ways of working, and common building blocks.
Yes, some of that is technical: environments that are quick to spin up; routine deployment; security, monitoring and accessibility baked in; common components like identity, payments and notifications; approaches to data that unlock its value across three tenses. But the platform play is bigger than tooling. It also includes the shared guidance and best practice that stops every team relearning the same lessons; spending, governance and assurance that work at pace; and a serious commitment to digital inclusion so faster doesn’t just mean faster-for-the-already-confident.
That’s the job of long-lived enabling teams and functions: not to build the service for everyone else, but to keep that common machinery healthy – technical and institutional – in good order, and to remove friction for delivery teams. When those foundations are weak, teams compensate by rebuilding them locally, in different ways, at different quality levels. At higher pace, that failure mode accelerates. If we want mission teams to surge without chaos, we need to strengthen this Government as a Platform enabling layer – and give it both the responsibility and the authority to make other teams faster, safer, and more focused on the needs of their users.
Talent is part of the platform
Finally, there’s the cultural side – though in truth it’s not really finally at all. Talent and capability are part of the platform. If AI assistance makes routine tasks easier, then that changes how we think about early-career roles. The challenge is not to hollow out the junior ranks, although there may be fewer of them, it’s to reimagine what it means to learn the craft.
If all we think juniors do is low-level, repetitive work, we misunderstand the point of apprenticeship. Learning to design public services has never just been about roadmaps or code; it’s about judgement, ethics, teamwork and care. The new generation should still have places to practise those – through reviewing AI output, working with users, understanding policy context, and shaping the human side of systems.
Our organisations may become flatter in terms of who contributes ideas – because if anyone can test an idea cheaply, more voices can lead to more experiments. But flat doesn’t mean leaderless. Leadership’s job in this environment is to set direction, create the environment for success, and curate learning: to channel energy towards outcomes, not let a thousand prototypes bloom into noise.
When building becomes cheaper in time and money, organising well means being light on our feet without losing coherence. Small teams that form, deliver, and reform elsewhere. Funding that follows the evidence of value. An enabling platform – technical and institutional – that liberates teams rather than constrains them. And an ethos that good ideas can come from anywhere, because the cost of trying them is low. But the discipline of learning still matters. This is a vision of a more organic approach to public sector organisations: one that can continuously reconfigure itself in response to new challenges, while staying accountable, inclusive, and trustworthy.