Powering AI transformation in the public sector
When I speak to technology leaders within the public sector, the conversation often comes back to one question: how does successful AI adoption lead to practical outcomes for customers—which is effectively citizens? As AI adoption intensifies, the race now is for scale and speed whilst staying within the guardrails of governance, organisational policy, and security.
These are the four things public sector organisations should consider to get the most out of their AI plans for 2027:
Scaling AI pilots
The most effective AI adoption I have seen starts with a focused use case and a clear understanding of how the process works end to end. A small pilot gives teams the space to establish how AI should operate: what data it needs, where human oversight belongs, how performance will be assessed and who remains accountable. That's where AI starts moving into a position where it can become scalable, quicker, with the right guardrails and parameters This subsequently enables better levels of adoption and engagement. Research shows that organisations investing in workforce digital transformation were 1.8 times more likely to report better financial results.
At FDM, our own approach to AI readiness is based on three core principles:
Curious in experimenting with AI
Confident in using AI tools
Secure in working within the right guardrails
This is shaping how we develop our workforce and support customers to explore adoption safely and productively.
The importance of good data
The success of any AI project depends on an organisation’s data quality. Poor-quality data, unclear ownership or unsuitable architecture can be detrimental to an otherwise promising use case. Industry has a role in helping government strengthen these foundations through capability in data engineering, architecture, governance and quality. Security, access controls and technical resilience need to be considered throughout that work. I see this as a connected skills challenge. Teams need to understand how decisions about data today affect the services and AI applications they want to develop tomorrow. There is also considerable knowledge within government to draw on.
In my experience, some of the government departments can be a bit siloed. However, with good collaboration, there's an opportunity to learn from different departments and strengthen their digital and data infrastructure.
Agentic engineering to modernise legacy tech
Modernising legacy tech is one of the key pillars of any transformation programme.
Agentic engineering is actively transforming how enterprises modernise legacy systems. Instead of acting as a simple chatbot, agentic AI uses networks of autonomous, specialised software agents that take ownership of complex engineering workflows—from code discovery and test generation to full architectural rewrites.
At FDM, we supported one organisation to move from a legacy Perl environment towards a modern Java stack. The experience reinforced something I see repeatedly: successful migration depends on the people who can connect the existing estate with its future iteration. Developing that capability helps organisations preserve essential knowledge while adopting new technology.
Make continuous change part of delivery
AI is already prompting teams to reconsider established workflows and tools. Realising its benefits requires people who can question assumptions, assess risks and help colleagues adopt new ways of working.
I believe we need to treat change as an ongoing capability.
FDM is involved in a lot of ecosystems that blend between multiple suppliers, technology providers and the government itself. We must unite as a collective around the needs of the customer, and their journeys with government digital services.
Government has strong examples of digital delivery to build on. Industry’s task is to help connect that experience, strengthen the workforce behind it and turn ambition into practical outcomes for citizens.
