Cloud has changed where technology lives, APIs changed how technology connects, and platform engineering changed how capabilities can be assembled. Software increasingly became a service rather than an installation, storage became abundant, connectivity became easier and the technical justification for placing every piece of data, process and functionality inside a single organisational application began to disappear.
But our behaviours did not change at the same pace. We continued to buy systems when what we increasingly needed were capabilities, continued to replace applications when what we often needed was better access to the data, and continued to write requirements as though every organisation needed its own bespoke version of eligibility checking, referral management, appointment booking, consent, messaging, evidence sharing, workflow and reporting. We carried procurement practices, governance models and assumptions about ownership from a world in which the application had to be the centre of everything into one in which it no longer does.
This is where the conversation about vendor lock-in becomes more complicated, because we cannot procure technology as though every organisation is unique, ask suppliers to reproduce years of local variation inside their products, embed those products deeply into the work and then be surprised when leaving is difficult. The market responded to the demand we created, building increasingly comprehensive products because we kept asking it to solve the whole problem, and the way out is not simply to demand better behaviour from suppliers while continuing to commission technology in exactly the same way.
Most suppliers are already engineering towards a different future because the wider technology market is moving from systems attempting to contain everything towards platforms, services, reusable components, and agentic applications. The public sector cannot ask the market to innovate in that direction while continuing to procure as though the best answer is a larger application with more features and another decade of accumulated customisation.
Buyers, suppliers, professions, commercial teams, policymakers and technology leaders are all part of the system that created the present position, which means none of us can change it alone or by waiting for somebody else to move first.
Rethinking how we buy technology
The next presumption should be different. Before procuring a new system, we should ask whether the need is genuinely for an application at all, whether the capability already exists elsewhere, whether it can be reused across organisations, whether the data can be separated from the product that happens to process it today and whether the investment will make the wider ecosystem more capable rather than only improving the organisation paying for it.
This does not mean removing local practice, professional judgement or the things that make services responsive to different populations and places. It means becoming clearer about the difference between valuable operational variation and accidental technical variation. People, relationships, priorities, professional decisions and service pathways may need to differ, but that does not mean identity, evidence, consent, referrals and information must be stored and exchanged differently in every place.
Good co-design can no longer mean taking every variation in an existing process and encoding it into another generation of permanent technology. It must help us understand where difference creates genuine value, where common capabilities can be reused and how people can work together across organisational boundaries without losing the local knowledge and professional judgement on which good services depend.
The application is no longer the only credible organising principle for technology, yet much of public sector transformation still behaves as though it is. The technology moved faster than the operating model, and the gap between the two is now showing up everywhere as technical debt, fragmented data, unaffordable replacement programmes and services that work efficiently within their own boundaries but struggle to work coherently with one another.
If you liked this content…
Moving from control to trust
Moving beyond that model is not simply a matter of issuing new architectural standards. The systems we built came to represent safety, accountability and control. Organisations learned to trust the applications they owned, the local administrators they knew, the suppliers with whom they had long relationships and the processes they could explain, even when the accumulated result was difficult to change.
Reuse asks organisations to depend on something designed elsewhere, federation asks them to rely on other organisations, and composability asks them to accept that no single supplier, profession or technology team will control the whole answer. Places need to trust that national standards will enable them rather than flatten local difference, the centre needs to trust places to compose services around their populations, professionals need to trust that shared components will support rather than replace their judgement, and suppliers need to trust that open ecosystems still provide room for investment, differentiation and long-term value.
Trust does not come from blaming one another for the current estate or attempting to design more control into the next one. It begins with acknowledging how each part of the system contributed to the outcome, remaining accountable for the decisions we now make and accepting that a different future will require all of us to think differently, buy differently, share more and trust one another with parts of the answer without expecting anyone to own the whole thing.
Problems don’t respect organisational boundaries
At exactly the point technology became more capable of operating across organisational boundaries, the problems facing public services became less willing to remain inside them. Economic inactivity is bound up with health, skills, transport, housing, confidence, caring responsibilities and the availability of work. Vulnerability can emerge through the interaction of debt, family circumstances, poor health, insecure housing, isolation and contact with multiple public services, none of which sees the entire picture alone. Children’s outcomes are shaped not only by education but by home, health, income, community, safety and opportunity.
These are not departmental problems waiting for a better departmental system. They are wicked problems, produced by the relationships between multiple factors, changing as we intervene in them and impossible for any organisation to solve alone. They sit between our institutions, while our funding, accountability, data and technology still sit largely within them.
From reacting to preventing
For much of the digital era we concentrated on helping individual organisations process demand more effectively once it arrived, and that improved services in many important ways, but the financial, demographic and social pressures facing the public sector mean we cannot simply become more efficient at responding to problems after they have become crises. We need to become better at recognising vulnerability earlier, understanding the connections between apparently separate events and coordinating support before someone reaches the point at which the most intrusive and expensive intervention becomes unavoidable.
That does not mean predicting individual lives or allowing an algorithm to make decisions about people it cannot understand. It means using data better so that professionals and communities can see patterns sooner, share relevant information safely, lawfully and purposefully, understand the wider context and organise support around a person rather than around the particular doorway through which they happened to enter the public sector. It requires clear stewardship, provenance, permissions and accountability, together with human judgement and the ability to understand and challenge how information has been used.
A future public service has to be proactive as well as reactive, finding opportunities to prevent harm rather than waiting for demand to declare itself through the language and thresholds of an individual service. It must offer a genuine tell us once experience in which information can be reused safely and appropriately, and it must coordinate around complex lives without requiring every organisation to treat the same person as a new case in a separate system.
The test is therefore not simply whether systems integrate. It is whether public services can identify and act on need earlier, coordinate around people and places, and make it easier for organisations to solve problems together than to manage their own fragment of the consequence alone.
