Reforming government’s large and complex processes is often likened to turning around an oil tanker – it takes time and much effort to change direction. In the Ministry of Justice’s service transformation group, they acknowledge the truth of this metaphor but they’ve added another. While some of the projects in the team’s wide-ranging portfolio are like tankers, driving complex, multi-year change, others are speed boats – sent out to solve problems or make improvements quickly while the tankers steadily shift.
It’s the job of Megan Lee-Devlin, the ministry’s director general for service transformation, to oversee this fleet of reforms, ensuring resources are allocated to keep each part on course. Alongside shaping the overall direction of travel, she also leads teams delivering access to justice services, including legal aid, lasting power of attorney and criminal injuries compensation. The goal of her fleet is to bring together delivery, digital and other transformation capabilities to “achieve a step change in the performance” of the MoJ’s services.
To do this, Lee-Devlin relies on a team which – to extend the nautical metaphor – she describes as similar to a rowing crew: “Each of the different people on the boat has a different role and a different set of capabilities, and it’s not on the strength of any one seat on the boat that we succeed, but on how we work together,” she explains. “Through integrated teams – for instance, embedding engineers with our frontline staff to automate processes or embedding data scientists with our operational leaders to predict error – we are able to tackle problems creatively and quickly.”
Lee-Devlin has spent a career “tackling wicked problems” in private sector organisations such as IBM, PwC and McKinsey. In 2021, inspired by seeing the public sector respond to the Covid-19 pandemic, particularly the way in which government moved at pace to roll out new digital, data and operational capabilities, she began a six-month temporary stint in the Cabinet Office. She was soon hooked, became a permanent director, then a director general, leading the digital and data function across government. “Suffice it to say that once I started working in the civil service, I discovered how great it is to have a job that you really jump out of bed in the morning to do and where you leave each day knowing that you’ve really done work that matters,” she says.
In 2024, Lee-Devlin moved to the Ministry of Justice to get “closer to the coalface”, working on services for people who are, as she puts it, “often at the most challenging point in their lives”. Whether victims of crime, people navigating a loved one’s loss of mental capacity, those beginning a rehabilitative journey after committing an offence, or frontline staff working in complex and challenging environments, these groups “need great services”, she reflects. And she is, she says, motivated by “how life-changing it can be for people when we get those services right”.
“Justice Transcribe has fundamentally changed how frontline colleagues are using their time”
Alongside this, she loves the breadth of the work. “There is a huge amount of opportunity for digital transformation across the justice system,” she says. As a result, she is leading “a broad portfolio of transformation projects at different stages of maturity in different corners of justice, including legal aid, prisons, probation and victims services, and some that span the system more widely, including work to better join up our data, accelerate AI and embed secure-by-design principles into everything we do.
“It is important that these programmes add up to a different quality of experience for the people who interact with our services, and better value for taxpayers. What’s really exciting is being able to look across the portfolio and see those benefits being realised.”
Part of her job is to ensure that lessons, tools and capabilities are shared across the portfolio rather than reinvented project by project. This is central to any good portfolio management approach but feels particularly relevant to the department’s adoption of AI technology. Indeed, sharing lessons and tools is an explicit aim of the department’s AI Action Plan for Justice, published in July 2025.
Justice Transcribe is a case in point. Developed initially for probation, the tool automatically converts spoken interactions between probation officers and people on probation into structured notes. Early pilots showed it could reduce the time spent on notetaking by around 50%.
“The impact of Justice Transcribe is that it’s fundamentally changed how frontline colleagues are using their time,” Lee-Devlin says. But the impact goes beyond efficiency. One probation officer told the team that, for the first time in her long career, she was able to spend an entire meeting looking at the person in front of her, and really engaging with the ways she could help them, rather than at a screen or scribbling notes.
Despite the excitement around AI, Lee-Devlin is clear that it is not a silver bullet. “We are also investing in our core technology, data and our people, to get the most from more cutting-edge capabilities like AI,” she says.
That means digitising paper processes, modernising legacy technology and improving data maturity across the department. Without those foundations, AI cannot be deployed safely or effectively.
This focus on fundamentals is the first of three priorities for the ministry’s digital-enabled transformation over the current spending review period. It includes ensuring legacy systems are safe, secure and capable of iterative improvement, and that newer systems can connect across organisational boundaries.
“For example, making sure data can flow from the courts to the prison system,” Lee-Devlin says. In a system as interconnected as justice, fragmented technology can undermine both efficiency and outcomes, she points out.
The second priority is transforming public-facing services. This includes process redesign, such as simplifying the administration faced by providers of legal aid, or through targeted use of AI, in line with the Government’s AI Opportunities Action Plan. This is where the speed boats come in, where MoJ is “using available technologies to solve problems quickly, particularly in areas of focus for ministers”.
An example of this is a digital rapid response unit, working alongside prison staff, to develop tools that help them to reduce error, while longer term work kicks off to digitise paper administration. These teams allow the MoJ to focus effort and resource on high-priority issues, without losing sight of the long-term reforms – the tankers – that underpin system-wide change.
The third priority is people. Lee-Devlin says the aim here is “to make sure that all of our staff have the skills and expertise they need to be able to use new tools effectively,” so that technology ultimately frees them to focus on work that matters most.
“AI is bringing a democratisation of digital development. You don’t need to be an expert coder to build a really useful tool that helps you with a day-to-day task”
The Ministry of Justice has taken a deliberately open approach to AI adoption. The department has enterprise licences for ChatGPT and Microsoft Copilot, allowing staff to use the tools safely and securely while protecting sensitive data.
This means that for simple use cases, MoJ staff don’t have to wait for others to build them tools, but can innovate themselves. This includes using AI to synthesise large briefing packs or turn handwritten notes from an away day into digital notes.
“What AI is bringing is a democratisation of digital development,” Lee-Devlin says. “You don’t need to be an expert coder to build a really useful AI tool that helps you to undertake a day-to-day task.”
Lee-Devlin herself uses tailored tools that distil documents and meeting packs into key points, drawing on her interests, role and institutional knowledge. She is encouraged by how many colleagues across justice are experimenting within agreed principles and parameters and then sharing what works through the department’s prompt-sharing platform.
This local innovation is supported by the Justice AI Unit: a relatively new team with a mission to put AI tools into the hands of staff, build confidence and free up time for higher-value work. The unit leads on AI learning and development, runs prompt-sharing competitions, rolls out enterprise tools and develops bespoke applications like Justice Transcribe. Over 5,000 staff have joined workshops to identify how their AI training can apply to their roles.
“What’s really special about the AI unit is not just the work they’re doing, but how they’re working,” Lee-Devlin says. First and foremost, that means working alongside colleagues and really listening to the problems they’re trying to solve. She adds that the team is also focused on finding smart ways to share good use cases, building platforms for colleagues to learn from each other, and helping them think through practical applications.
She also highlights a fresh approach to AI talent, with internships and fellowships drawing in people from a wide range of backgrounds to support the justice mission.
What excites Lee-Devlin most about the current moment is the shift from pilots to doing things at scale.
Justice Transcribe perfectly illustrates that transition. From a small pilot group, it is now being rolled out to thousands of probation officers, with a commitment made by the deputy prime minister that all relevant staff will have access within a year. “It is a powerful thing,” Lee-Devlin reflects, “to hear probation officers telling us that it’s changed the way that they work.”
In all of this, she and her team remain focused not just on changing how colleagues work, but on the impact that will have on the citizens needing their support. It’s a focus shared by teams across justice who, she says, are constantly thinking about small and large ways to improve citizen experience as well as efficiency.
“One of the things I love about my role is how every day I encounter colleagues who are thinking: ‘How do I make something a little bit better for our customers and frontline?’”
Advice for senior leaders supporting AI adoption
For senior civil servants navigating this rapidly evolving landscape, Lee-Devlin offers three pieces of advice.
First: be curious. Make the most of all the learning resources that exist, she says, and find formats that suit how you learn – whether that is reading, listening or watching.
Second: be connected. Leaders should seek out colleagues with expertise, both within their organisations and beyond, and use them as thought partners. One of the civil service’s strengths, she argues, is the breadth of shared experience across departments and agencies. “Steal with pride,” she advises.
Third: be a responsible champion. AI use must be grounded in ethics frameworks, robust governance and cross-functional input from legal, security, operational and technology teams. Leaders need to be clear with their teams about appropriate use cases and guardrails, while also role-modelling adoption once those conditions are in place.
“Knowing the limitations of the technology is really important,” she adds. This applies particularly to the risk of hallucinations. The same rigour and critical thinking applied to human advice must be applied to AI outputs.
At the Ministry of Justice, more than 80 senior civil servants now act as AI champions within their business areas. “They are not AI experts,” Lee-Devlin says. “They are experts in a whole multitude of different things. But they’re being curious, they’re connecting, and they’re championing responsible adoption.”