Earlier this year, Dame Antonia Romeo stepped into the role of the cabinet secretary looking to modernise the public sector through the use of AI and automated approval processes in an effort to cut down administrative burdens.
As a former senior civil servant in the Cabinet Office, I believe this will be harder than it looks. One of the main challenges she will face is the legacy technology that still plagues the public sector.
The impact of legacy technology
I’ve seen first hand how outdated systems stand in the way of digital transformation, but they also impact civil servants’ day-to-day. Getting a straightforward response from such systems can take a very long time, or you may not get an answer at all.
Systems across Defra have complex and risky dependencies that are difficult to untangle, as highlighted in a 2023 Public Accounts Committee report. They are highly customised, often poorly documented, and make it hard for civil servants to extract reliable information.
Another barrier I encountered in government was the difficulty of building a clear, up-to-date picture of what is happening across departments. Fragmented, inaccessible systems that don’t link datasets make this problem far worse.
For example, in one project, my Cabinet Office colleagues and I needed to understand the impact of SPS (Sanitary and Phytosanitary) controls on animals and food products at the border. This data should have been readily available in Defra’s systems, since every pen of hens is registered. Yet legacy infrastructure made it slow and difficult to extract the information we needed, or validate the data we already had.
Manual-entry Excel sheets are also still very common across the government, even in 2026. A lot of reporting is based on manual entries, with data taken from a system but then written by hand into Excel, which often results in incorrect entries.
But legacy systems don’t just cause inconveniences – they are also costly to maintain; increase security and compliance risks; and can undermine AI adoption initiatives.
It is also worth noting that not all legacy systems are a nuisance. Some underpin critical operations or hold valuable data, meaning civil servants need to carefully evaluate which ones are important and which ones can be binned or updated.
Cultural resistance
Legacy systems are not the only barriers to modernising the public sector. There is also a cultural resistance among civil servants when it comes to giving up outdated systems and using AI and automation.
In my time as a civil servant, I witnessed a general reluctance among many civil servants to share their department's data with other departments.
Not only that, but I’ve even known several civil servants who are actually quite attached to their legacy systems and workarounds. Naturally, if you’ve spent your entire career working on these systems, and you were trained on them, there’s a danger of becoming the "gatekeeper", making you the only person who understands the system or the workaround you created. This personal investment can become a roadblock to legacy renewal or automation.
There is also a significant skills gap around AI which is bigger than in the private sector. To bridge this gap, the public sector needs better training and education. It’s also key to have AI specialists on the payroll, appropriate funding for specialised roles and reformed procurement rules that encourage innovation instead of long-term service contracts.
Other obstacles include unclear governance structures, vendor lock-in from legacy contracts, job security concerns and political sensitivities, all of which encourage departments to favour small pilots over large-scale transformation.
Change is on the horizon
If we do it right, AI would radically transform the delivery of public services. For example, it could automate routine administrative tasks such as issuing permits or managing records, freeing up time and resources.
It could also solve our legacy system nightmare. Instead of costly, complicated replacements, AI can rebuild old systems in the cloud using whatever artefacts exist such as old training videos, manuals, and outdated code. This cuts time, cost and risk, and lets civil servants focus on delivering actual results instead of wrestling with broken technology.
The big question now is how Dame Antonia Romeo’s ambitions play out in reality, and whether the full transformative impact of AI can be felt across the civil service.
Alex Case is a former senior civil servant, who worked at departments including No.10 and the Cabinet Office