Q&A: Unlocking AI-driven transformation in government

What are the most persistent structural or cultural barriers preventing government departments from moving beyond siloed ways of working – and how can we start to dismantle them?

Some of the most persistent barriers stem from the very way the public sector is structured. Departments often operate as independent entities, each with its own systems, processes, and technology estates. This siloed structure makes it extremely difficult to standardise approaches, build interoperability, and reuse solutions across government. As a result, data remains fragmented and underused, which in turn severely limits the potential of AI, analytics, and other advanced digital technologies to deliver systemic value.

Another recurring obstacle is the inconsistent presence and empowerment of digital leadership at senior levels. Many organisations still lack leaders with sufficient digital fluency in executive decision-making spaces. This absence not only hinders strategic thinking about digital transformation but also means that data and AI are often treated as technical afterthoughts rather than central enablers of policy and service delivery. Outdated funding models further compound the problem, prioritising the launch of new programmes over the equally critical – but less politically visible – work of maintaining and modernising existing systems. This leads to risky legacy infrastructure consuming the majority of IT spend while starving innovation of resources.

Talent is another structural challenge. The public sector struggles to attract and retain skilled digital and data professionals, who are in high demand in the private sector. This challenge is amplified as experienced employees retire, widening the skills gap and limiting institutional capability.

Finally, cultural resistance to digital tools continues to impede progress. Many public servants remain understandably cautious about unfamiliar technologies, whether due to a preference for traditional methods, a lack of trust in data use, or anxieties about how AI might affect their roles. These cultural barriers cannot be underestimated – they are interwoven with those structural issues to form a complex, systemic challenge.

Dismantling these barriers requires a holistic approach rather than piecemeal reforms. At a policy level, initiatives such as a Digital Wallet, a ‘once-only’ rule for citizen data, and a National Data Library could serve as practical enablers that unlock cross-government data assets. Elevating digital leaders—by requiring digital expertise in senior appointments and embedding them in executive committees—will ensure data and AI governance is driven strategically, not tactically.

Equally important is reforming funding models so they support safe, sustainable services, with long-term investment in maintaining critical data infrastructure as well as in building new solutions. Cultivating a culture of continuous learning and adaptability is essential: every role should include digital skills development, anxieties should be openly addressed, and ‘AI champions’ should demonstrate the tangible value of AI to everyday work. Finally, collaboration is key. Stronger partnerships with industry and civil society can inject expertise in data management, analytics, and AI development, accelerating government’s ability to meet rising expectations.

We often hear terms like “mission-led” and “transformation-led”, but what does real transformation look like on the ground for frontline civil servants and public-facing operations?

Real transformation must move beyond strategy documents and rhetoric to deliver tangible improvements that frontline staff and the public can feel. On the ground, this often looks like data flowing faster and more seamlessly across government systems, enabling greater automation of repetitive, rules-driven processes.

For civil servants, this translates into the precious gift of time. Freed from administrative burdens such as manual data entry or duplicative paperwork, staff can focus on higher-value tasks requiring judgment, empathy, and expertise. Far from displacing human work, AI should be framed as a ‘time multiplier’ – a tool that augments human decision-making with richer data insights, enhances creativity, and improves morale by allowing staff to concentrate on what only people can do.

For the public, transformation is felt through faster, more reliable, and more intuitive services. Modernised systems can cut waiting times, simplify interactions, and deliver digital experiences that match the ease of consumer apps. Examples already exist: the NHS App saved 5.7 million staff hours and £622 million by streamlining patient data access, while the MoJ’s Family Law AI Chatbot has empowered families with self-help tools and eased pressure on court staff.

Other pilots show AI’s promise in high-impact areas: auto-redaction in policing to reduce the time spent processing digital evidence, Border Force’s use of AI on x-ray image databases to improve freight screening, and local councils leveraging AI to automate data capture in social care. These examples highlight that transformation is not an abstract ambition—it is already producing measurable savings, improved citizen experiences, and better use of staff capacity.

Spreadsheets and outdated systems are still core to many government functions. What are some practical first steps departments can take to move away from legacy tools without overwhelming staff or services?

Legacy IT remains one of the greatest inhibitors of progress. The UK government spends £2.3 billion a year – nearly half its technology budget – just to keep old systems running. That leaves barely 19 percent for innovation. Beyond the staggering cost, these systems are fragile, insecure, and prone to failure: NHS England alone reported 123 major outages in 2024, each forcing staff back to paper-based processes. They also deter talent, as employees are forced into lengthy training just to navigate unintuitive systems, and prospective hires are put off by outdated tools.

Modernisation is therefore not optional – it is an essential investment in security, resilience, and innovation. Yet moving away from legacy systems must be carefully managed to avoid disruption. Practical first steps include:

• Comprehensive systems assessment to map interdependencies, technical debt, and data flows. This creates a prioritised roadmap that tackles the riskiest and most costly systems first.
• Phased migration using approaches like the Strangler Pattern, which allow new and legacy systems to coexist, reducing risk and building momentum through small, visible wins.
• Cloud adoption through hybrid models that reduce maintenance costs, improve scalability, and strengthen security while still supporting essential legacy functions.
• User-centric design to ensure modern systems align with how people actually work, reducing training needs and boosting adoption.
• Data readiness initiatives to ensure fragmented and siloed data is cleaned, connected, and made accessible before layering on advanced AI.
• Strategic automation rollouts that demonstrate immediate value by eliminating repetitive tasks, freeing staff for higher-value work.

These phased, data-driven approaches avoid the pitfalls of ‘big bang’ replacements while steadily moving departments away from outdated, high-risk systems.

Capability building is often framed as training, but how do we shift from training alone to truly enabling civil servants to use data and AI confidently and intuitively in their daily work?

Traditional training programmes often fail to bridge the gap between theory and practice. To genuinely enable staff, capability building must be practical, applied, and continuous.

The government’s AI Accelerator Programme is an encouraging model. One-third of the programme is devoted to hackathons and real-world projects, where participants build AI applications with industry-standard tools. Combined with expert-led workshops on topics such as Responsible AI and cloud services, this hands-on approach builds not only technical skills but also the confidence to apply them.

Beyond formal programmes, cultural change is essential. AI champions and peer networks can provide trusted guidance, show colleagues how tools apply to their roles, and address anxieties about workload or job security. Transparency about where and how AI is used builds trust, while diverse teams help spot risks and mitigate bias.

The goal is not just to ‘train’ but to empower civil servants – embedding digital capability into every role, normalising continuous learning, and giving staff the confidence to see AI as an ally that enhances their expertise.

Can you share examples of how your team is working with government clients to bridge the gap between AI potential and legacy reality? What’s worked – and what hasn’t?

Our approach focuses on understanding current pain points, then proving value through iterative, data-driven projects.

What’s worked:

• Improving public-facing services, such as the NHS App transformation, which freed clinical capacity and saved costs while creating foundations for future AI use.
• Automating administrative burdens, such as the MoJ’s AI-powered legal tool, which helps families resolve disputes and streamlines court data. Similarly, auto-redaction pilots in policing have shown significant potential to reduce repetitive tasks.
• Enhancing operational efficiency and security, such as organising Border Force’s x-ray image databases and applying AI to detect anomalies, balancing trade flow with national security.

These projects follow a ‘scan, pilot, scale’ methodology: identify specific pain points, prove quick wins, then scale up with confidence.

What hasn’t always worked:

• Poor data quality and fragmentation, which frequently limits the effectiveness of AI. Without proper governance, AI risks becoming underpowered or insecure.
• Cultural resistance, where staff anxiety slows adoption. Overcoming this requires clear communication, change management, and proof that AI augments rather than replaces roles.
• Outdated funding models, which deprioritise remediation of legacy systems and create barriers to integration.

The lesson is clear: AI transformation depends as much on addressing foundations – data, culture, funding – as it does on the technology itself.

Finally, if you were to advise a departmental leader today on three immediate actions to accelerate AI-driven change without creating more complexity, what would they be?

1. Prioritise data readiness. Launch a programme to clean, connect and govern data, creating a single source of truth. Without this, AI will only amplify complexity.
2. Adopt ‘scan, pilot, scale.’ Start small with high-impact use cases like document automation, prove measurable value, and then expand systematically.
3. Build an AI-confident workforce. Invest in applied capability-building and AI champions, empowering staff to see AI as a tool that enhances their roles and makes work more meaningful.

These three actions—data readiness, iterative scaling, and human empowerment—create a solid, sustainable path to AI-driven transformation.

WATCH: Digitise, Automate, Innovate — Paving the Way for AI. Join Chad and senior leaders from the Home Office, Cabinet Office, and Department for Business and Trade as they discuss building the right foundations to serve government’s AI ambition.