OPINION - Kill the Prefix: Notes from a Post-Digital Civil Service

I hear the quiet frustration in the corridors of Whitehall and the tired sighs on crossdepartmental Teams calls. You sit in another "Digital Board" debating an "AI Strategy" as if it were separable from the public services your department exists to provide. You feel like a mediator in a marriage that should have been a single organism a decade ago.

Your discomfort is professional lucidity.

We are witnessing the death of a category. The transition from "doing digital" to being a modern state is profound. The current obsession with AI is a distraction from the fact that our institutional foundations remain unresolved.

The Taxonomy Trap

Across the Service, stakeholders still talk about "the Business" and "IT" like rival factions in a 1990s sitcom. This is the Taxonomy Trap.

We have entered the post-digital condition where technology is now inseparable from administrative and legislative function. Discussing it in isolation is an ontological error. I recently watched a Policy Lead present a 50-slide strategy that did not mention data architecture once. When asked how the systems would interoperate, the answer was a blank stare and a reassurance that "the digital team will handle implementation."

This is how failure is incubated.

When we prefix a project with "Digital," we provide a linguistic escape hatch. Policy colleagues are subtly licensed to disengage from delivery logic. Technical architecture becomes someone else’s problem. Core policy decisions are deferred to specialists or, increasingly, to external vendors.

The mandate is simple: kill the prefix.

If we are iterating the state pension, it is a pension service. If the technology fails, the policy fails. There is no distinction.

AI Hype and Institutional Debt

We are sprinting toward an "AI-First" Civil Service while still shackled to late-90s databases. There is a seductive narrative circulating in briefing rooms suggesting that tools like Consult or Extract will unlock productivity and modernise the state. AI is a layer, not a foundation. You cannot build sophisticated automation on fractured data architecture and expect coherence.

More dangerously, generative AI has made it effortless to produce 40-page strategies that say nothing. The tip-toeing and risk-avoidance of yesterday can now be automated at scale. "AI slop" has become a new form of bureaucratic fluency. Meanwhile, the fundamentals remain neglected:

• Fragmented data estates.
• Legacy systems that cannot interoperate.
• Service design capability stretched thin.
• Staff expected to acquire "AI skills" before mastering digital basics.

In one department, seven figures were spent on an AI-enabled triage dashboard to "optimise casework flows." The annual report praised the innovation. What it did not mention was that frontline staff were still manually re-keying data between two noninteroperable legacy systems. The automation layer was celebrated while the structural inefficiency was preserved.

This is decoration. It is easier to buy a chatbot than to fix a database.

The Political Economy of "The Project"

Departments continue to chase the Transformation mirage. This is the belief that an eighteen-month sprint will deliver a "transformed state" after which normality resumes.

Go-Live is celebrated with cake. Six months later, 400 unresolved defects remain. The Project has closed, the budget line has ended, and the team has dispersed. This is rational behaviour under current funding models.

Capital expenditure (CapEx) spikes are attractive to the centre because they end. They can be announced, costed, and declared complete. Operational capability (OpEx) persists. It demands sustained funding and long-term accountability.

So we build projects. We disband teams at the moment of delivery. We mistake the ribbon-cutting for the reform.

In a post-digital state, change is the baseline. A "Digital Transformation Programme" is a contradiction in terms. There is no finish line, only services that must be owned, iterated, and maintained. We need permanent, multi-disciplinary teams accountable for outcomes across the full lifecycle of a service.

From Order-Taking to Orchestration

A persistent agony across digital functions is the sense of being a highly paid messenger, capturing requirements for pre-defined policies and passing them between silos. If your role begins after the policy is fixed, the battle is already lost.

Value in the modern state is co-created upstream. The difference between being told to "build an app for farmers" and being asked to "solve the subsidy application bottleneck" is the difference between order-taking and orchestration.

The seam between policy and delivery must disappear. That requires more than agile fluency. It requires the courage to tell a Director General that their "innovative AI dashboard" is a 21st-century paperweight. It requires reframing problems before procurement begins. It requires resisting vendor-led "black box" solutions that externalise capability rather than build it. This is uncomfortable work. It is the only work that matters.

The Paradox of Invisibility

The ultimate objective of our profession is invisibility. When contactless payments work, no one marvels at their digital sophistication. When GPS routes correctly, no one praises the infrastructure. It simply functions.

Public services should be the same. The louder we proclaim "Digital Government" or "AI-First Strategy," the more we signal how far we are from normality. Mature systems do not advertise their modernity; they embody it.

If you are leading in this space, you are often the only person in the room resisting AIwashing and vendor capture. It is lonely because structural critique is less exciting than technological theatre. Dashboards will not modernise the state. Declarations of being "AI-First" will not modernise the state.

The state will be modernised when we stop pretending that policy, technology, funding models, and delivery are separable domains. Until then, we are not transforming government.

We are decorating its interfaces.