Oliver Fox, Director of Central Government and David Barber, Director of the UCL Centre for AI & Distinguished Scientist at UiPath, discuss how the UK Government can successfully implement its ambitions to scale AI across public services safely
Yet, despite ambition and investment, many government departments are struggling with execution. A recent report found that while most public sector executives believe AI can deliver cost savings and improve services, only 26% of departments have integrated it across their organisation. Legacy systems, siloed data and fragmented procurement continue to hold back progress.
The next phase of adoption requires unblocking the invisible infrastructure connecting departments, enabling smarter workflows and giving civil servants the tools and confidence to deliver AI-powered services safely and effectively.
AI in Government: Breaking down silos with interoperable infrastructure
The thing many government departments have in common is that they hold unprecedented volumes of data, but its value is often unrealised because it remains trapped in departmental silos. Organisational, technical and governance boundaries prevent data from being securely shared or reused across government departments. This has a knock-on effect of limiting the ability to tackle complex, cross-cutting challenges. The Cabinet Office, the National Audit Office (NAO), and the Government Digital Service have all highlighted persistent challenges in data quality, interoperability, and data sharing.
Underscoring these challenges is the invisible infrastructure that hinders the connection between government systems. Instead of data flowing securely and efficiently, it is often constrained by incompatible platforms and outdated processes. To make progress, the government must rethink how information moves through departments.
AI is most valuable when it understands tasks, knows what to ask and shares answers – not sensitive data – within strict, auditable boundaries. Agentic AI puts this model into practice by identifying the correct information across systems, taking action autonomously, and advancing processes more efficiently.
Combined with interoperable frameworks — common data standards, APIs, and trusted exchange protocols — agentic AI can reduce duplicate data requests, accelerate service delivery, and free civil servants to focus on higher-value work. Smarter use of linked data could boost public sector efficiency by up to 20%, illustrating the scale of what is possible when AI is deployed with intent. While AI is not a complete replacement for deeper structural reforms, it can significantly accelerate the benefits of improved connectivity.
Equipping civil servants for the digital era
For AI to deliver meaningful change, people play a pivotal, ongoing role in implementation. Civil servants today lack the tools, confidence and AI literacy to effectively exploit automation, which, even with the best technology, cannot deliver real outcomes. The Smarter Delivery of Public Services report highlights persistent gaps in digital capability, which directly contribute to service backlogs and poorer citizen outcomes. The AI can empower civil servants by reinforcing the expertise that underpins UK public services, rather than replacing it.
Rather than embracing modern, agile methods, many departments remain stuck with legacy procurement models defined by rigid contracts, inflexible supplier relationships, and reliance on external consultants. This slows deliverables, balloons costs and erodes institutional knowledge. Public spending on digital programmes now exceeds £14 billion per year, the NAO warns, driven in part by misaligned responsibilities and a shortage of skilled professionals.
Maintaining a strong human-in-the-loop model is essential for trust, security and quality, especially in high-stakes public services. Civil servants don’t need to be AI experts, but do require training, accessible tools and the confidence to oversee AI-enabled processes.
AI fluency programmes and new talent models can help the workforce adapt to evolving technologies. In a trial with 20,000 civil servants, AI tools for routine tasks saved two working weeks per person, per year, freeing time for higher-value work. Investing in training alongside technology ensures that automation empowers rather than replaces workers.
Scaling reusable AI assets
Once the workforce is equipped, the next challenge is scaling proven solutions. Too often, departments reinvent the wheel, building AI tools that already exist. This duplication wastes time, capacity and budget, slowing delivery and learning. Prioritising reusable AI components enables the government to build on what works, accelerating innovation across the public sector.
Despite over £26 billion in annual digital and data spending, a quarter of UK government services remain outdated, and 47% are not digitised. Departments have long invested in bespoke, one-off solutions rather than updating or scaling existing systems – NHS England, for example, operated around 50 different CRM platforms before recent restructuring.
A central repository of reusable AI components, including models, workflows, and APIs, would allow departments to build on existing capabilities rather than starting from scratch. The AI Opportunities Action Plan recognises that government purchasing power can further scale AI, provided procurement prioritises reuse, modularity and interoperability.
By shifting from bespoke builds to reuse-based scaling, the public sector can deliver more with fewer resources – moving from experimentation to enterprise-level delivery.
Delivering on AI ambition
The UK government has ambitious plans to achieve AI success. Delivering on that ambition hinges on execution; modernising underlying systems, empowering the workforce and adopting approaches that scale rather than stall. If the UK government aligns its intent with sustained, practical action, AI will not simply make services more efficient but help build a public sector equipped to tackle future challenges with confidence.
