276d07bb

Update from the GDS Responsible AI Advisory Panel

Thu, 18 Jun 2026 12:08:00 GMT

The UK Government Digital Service’s Responsible AI Advisory Panel had its first meeting at the beginning of March and its second just last week.

The panel was announced in the Blueprint for modern digital government to “bring together expert insight from the public sector (including frontline workers), industry, academia and civil society groups to provide constructive challenge and advice, and shape standards based on best practice.”

What do we mean by ‘Responsible AI?’

"Responsible AI”, like “ethical AI” and “trustworthy AI”, is hard to define, except that clearly the public sector shouldn't be irresponsible as it adopts AI. Responsibility includes, in no particular order, aspects of effectiveness, equity, security, resilience, environmental sustainability, democratic legitimacy and value for money. Building and buying responsible AI, in responsible ways, should avoid harms arising from AI, collapses in public trust, and wasting public money.

How this fits with existing government activity

The Responsible AI Advisory Panel is working closely with DSIT teams involved in the Data and AI Ethics Framework ; the Data and AI Ethics community of practice ; and the Algorithmic Transparency Recording Standard (ATRS) , which mandates which information government departments have to share about algorithms (including AI) in deployment.

Early Insights

Over just the last few months of the panel, we’ve already seen some of the promises and challenges that the public sector faces as it adopts AI. For example, we've seen a clear commitment to only introduce AI if it genuinely improves services for the public, and examples where it could reduce waiting times and improve accessibility of public services for people for whom English is a second language or who need support outside working hours.

We've also seen how hard it is to test and evaluate the outputs of large language models (such as ChatGPT, Gemini or Claude), even when they use retrieval-augmented generation to incorporate authoritative factual material into their context prior to responding to a question, especially when public bodies lack datasets holding examples of queries being answered well and correctly. In the private sector it might (arguably) be sufficient to use a disclaimer to remind users not to rely on the veracity of AI-generated responses; in the public sector, inaccurate AI-generated advice is not only potentially harmful for citizens who follow it, but also further diminishes trust in public services.

The panel’s focus

The panel is operating as a pilot, and we’re attempting to learn as much as we can about what kind of support an external advisory panel on responsible AI can provide to GDS and the public sector more generally. In the last few months, we've identified three broad categories of advice that could be useful:

Advice to individual projects and programmes focused on AI-based products and services.
Advice to DSIT Ministers around strategic challenges such as effective public transparency, and how to procure AI systems and services with an eye on digital sovereignty and public trust.
Advice to the GDS AI team about how to increase adoption of responsible AI practices across the public sector.

The panel is already beginning to engage with government AI projects, offering early advice and challenge. This includes work with teams developing Gov Voice, a reusable AI-enabled capability to improve service efficiency, improve quality and accessibility, and reduce duplication across government, and with programmes exploring AI tutoring tools to support pupils through personalised learning. These engagements help ensure that considerations around effectiveness, fairness and trust are built into services from the outset.

We’re also exploring how the panel can be a useful interface between public bodies, and between the public and the public sector. For example, as well as providing recommendations about what public bodies should be doing, we can highlight some of the good practices that the public sector is already adopting.

At the same time, we’re clear on our limits. The Responsible AI Advisory Panel is not a governance body and cannot provide assurance around whether or not public bodies are acting responsibly as they adopt AI. While the panel is diverse and approaches questions around responsibility from several different angles, we do not hold deep expertise on every responsible AI topic, nor can we represent the varied interests of all the communities affected by AI in public services.

What happens next

We will continue to experiment over the remainder of this year, building towards a final independent report at the end of our initial term. In the meantime, while we can’t promise to respond, I would love to hear from people inside and outside government about your experiences with responsible AI adoption in the public sector, and what you would like to see from us. You can contact us through the panel secretariat at responsibleaipanel@dsit.gov.uk .

Data maturity: the foundation for AI ready public sector data

Thu, 04 Jun 2026 14:04:00 GMT

Data maturity is essential for any organisation thinking about using AI. Before data can be considered AI ready, there needs to be confidence that it is well managed, thoroughly understood, and supported by the right people, processes and culture.

Government Digital Service (GDS) has partnered with The National Archives to explore whether and how their legal data can be prepared for AI use. Our partnership, which completed its discovery phase in April 2026, used data maturity as a framework for understanding The National Archives’ strengths and opportunities before committing to any technology approach. We are now moving into an Alpha phase. This will explore how data maturity mitigates the risks of exposing data to AI, and test potential approaches as the work progresses.

What is data maturity, and why does it matter?

Data maturity is about ensuring data is good enough to do what it needs to do, and that the people, processes and governance around it are capable of meeting those needs. AI depends on good data for good outputs. AI readiness requires confidence that data can be used safely, ethically and responsibly.

Data is only as effective as the people who manage it. Poor quality data produces unreliable AI outputs, but good data that is poorly understood or poorly governed cannot be considered AI ready either. Data maturity is where any AI readiness assessment should start.

What we set out to do with The National Archives

The purpose of the discovery phase was to explore whether The National Archives' legal data, which includes legislation and case law, could be optimised for AI use, and whether AI could be relied on when it matters most to citizens and businesses. This is part of GDS' mission to ensure public sector data is managed as a strategic national asset.

What we found

The discovery phase produced three main findings, the third of which represents the most significant opportunity for government. First, The National Archives' legal data is already close to being AI ready. The data quality is high, and The National Archives demonstrates strong data maturity, with the right skills, leadership and cultures in place. Good data managed poorly is not AI ready. The National Archives' strength is that both the data and the organisation surrounding it meet the bar.

Second, technology is moving very quickly. Large technology companies and other governments are working on the same problems. Public sector organisations building AI services need to account for how rapidly the landscape is changing and avoid over-investing in solutions that may be superseded.

Third, we believe the most significant opportunity for government in the next phase is exploring how AI generated outputs can be evaluated and validated. This is where government can add real value by building the tools and standards to assess whether AI outputs are trustworthy, rather than replicating what the private sector is already doing.

Technology moves faster than data projects

Had the GDS and The National Archives partnership gone straight to building an AI chatbot for legislation and case law, we could have found ourselves solving a problem the world's largest AI companies are already working on, with a fraction of the budget.

Instead, we used data maturity as a framework for understanding The National Archives' strengths and areas for development, helping us identify longer-term opportunities less vulnerable to rapid shifts in technology. This included further work on Model Context Protocol (MCP), an open source standard for connecting AI applications to external systems such as databases or other tools and documents.

Work with partners to solve shared problems

One of The National Archives' important organisational strengths, which would not appear in any analysis of its data sets alone, is its culture of responsible innovation. Staff with deep expertise and strong leadership are willing to explore new opportunities while being aware of the risks of moving too fast.

To test how legal data could be exposed via MCP, we partnered with the Department for Business and Trade and the Ministry of Justice to run a hackathon bringing together 40 lawyers, policymakers, data scientists and engineers from academia, the Civil Service and small and medium-sized enterprises (SMEs).

Early evidence indicated that MCP makes a significant positive difference to the quality of AI outputs. The next phase of the GDS and The National Archives partnership will explore this further, building on GDS guidance and best practice for making government data sets ready for AI.

A model with potential for wider application

This discovery is a pilot that can be replicated. It provides an example of how a data maturity-led approach can support decisions about AI. GDS will shortly launch a revised data maturity service, building on the framework published in 2023.

Building for the future: Making change simple on GOV.UK Pay

Tue, 02 Jun 2026 10:06:00 GMT

GOV.UK Pay has now processed over £9 billion through over 135 million transactions in its nearly 10 years processing payments for the public sector. Throughout that time, we’ve seen the number of services we support increase substantially and we're continually improving the service for them.

As part of the next stage of our journey we're pleased to announce Adyen as our new payment services provider (PSP) for non-Crown card payments and pay by bank, replacing Stripe.

The procurement covered non-Crown card payments and pay by bank with a specific requirement to at least maintain the current level of functionality offered to services. Suppliers were invited to bid in an open and fair tender process.

This change affects local authorities, armed forces, police forces and other organisations currently using GOV.UK Pay with Stripe as their provider. Central government, NHS and arm's length bodies will still use Worldpay for payment processing, with no change for them.

We’re doing the hard work to make it simple

A core part of the proposition of GOV.UK Pay right from the start in 2015 is that service teams should have to do as little as possible to take payments. We are already saving government significant processing fees as well as countless hours and days of people's time by handling service provider relationships and doing the technical work to keep GOV.UK Pay safe, secure and accessible.

We will now begin migrating around 1,000 services to Adyen, while making sure people can continue to pay for government services with no interruptions.

We will make migration as straightforward as possible while complying with Know Your Customer legislation that protects everyone from fraud. Most importantly, there will be no discernible difference for paying users and no loss in functionality.

Building for the future

This supplier migration is also the next big step towards introducing new ways for people to pay for services. This includes pay by bank which will reduce fraud and lost payments, as well as saving significant money for government. For paying users, it saves typing in card details and is safer and easier than a typical bank transfer.

Other improvements we're making at the moment

We have recently made a series of changes to prepare GOV.UK Pay to support other payment methods. For example, we've reorganised our settings section in the admin tool to make it easier to find what you need, and we've made it clearer when your service has gone live.

We've also upgraded our Web Application Firewall to make it easier to support in future while making sure GOV.UK Pay remains stable and secure.

You can find out more about what we've been up to and what we've got planned on our public roadmap .

Stay informed

If you are a current user of GOV.UK Pay, we will be in direct contact regularly and often to make sure you have all the information about the migration and any effect on your services (we expect this to be minimal).

If you’re a public sector organisation not currently using GOV.UK Pay, find out more or get in touch with us to see how we could help you improve the way you take payments from service users.

For any immediate questions, email us at govuk-pay-support@digital.cabinet-office.gov.uk.

Why Most Startups Struggle to Scale

Mon, 01 Jun 2026 11:38:31 GMT

Every new business starts with a bold vision and ambitious goals. But for many small and medium-sized enterprises (SMEs) – including tech startups in emerging markets – the greatest challenge is not launching the company; it’s figuring out how to scale and sustain growth without the organization becoming unmanageable. In fact, global surveys show that two out of three SME executives worldwide identify basic survival and scaling as their foremost concern. These concerns are especially acute for startups in emerging economies seeking to expand internationally, where operational complexity and resource constraints can be even greater.

Many young companies begin operations with makeshift setups – for example, using disparate spreadsheets, personal email accounts, and ad-hoc manual processes to run day-to-day tasks. This can work well enough during the early hustle. However, as teams expand, customers multiply, and operations become more complex, these gaps start to create friction and bottlenecks. A technology startup in Lagos or Jakarta might start with one or two people managing sales via spreadsheets and chat apps, but scaling to reach international markets or a larger team often reveals the cracks.

he startups that manage to scale successfully tend not to be the ones with the deepest pockets or greatest resources. They are usually the ones that lay the right operational and digital foundations early on, before growth outruns their systems. Research backs this up: a global study by Microsoft found that Startups considered “early adopters” of technology were twice as likely to have met their business objectives and achieved higher revenue growth – and four times more likely to feel confident about their future success – compared to peers who waited to modernize their operations. The lesson is clear: getting foundational processes and digital tools right from day one dramatically improves a young company’s capacity to grow with confidence and resilience.

The Hidden Cost of Operational Shortcuts

It’s common for a growing business to stick with familiar shortcuts – using Excel to manage critical processes, relying on informal communication via personal emails or messaging apps, and patching together manual workflows that depend on a few key individuals. In emerging markets start-ups, these shortcuts may feel like pragmatic ways to overcome resource limitations, but research shows they impose hidden costs as a company scales.

What seems “good enough” in the early days often leads to serious issues later on:

Reduced visibility across the business. When data is siloed in spreadsheets or personal accounts, leaders struggle to get a clear, real-time picture of operations. For example, in one survey, nearly 30% of small business owners admitted they often waste time searching for information in the wrong place, and 29% said they duplicate communications across different apps – signs of a fragmented tech environment. This lack of unified visibility can lead to missed opportunities and slower decision-making.
Increased operational risk. Workflows held together by manual steps and “tribal knowledge” invite mistakes and oversights. Spreadsheets, copy-and-paste workarounds, and other improvised fixes can result in duplicate data entry, inconsistent reporting, and errors that cascade through the business. The consequences can be costly: a single spreadsheet error cost one UK company £38,000 in lost revenue – an all-too-real example of how fragile ad-hoc processes can undermine a growing firm.
Inefficient onboarding of new staff. When processes live in people’s heads rather than in documented systems, training new employees becomes slow and inconsistent. New hires may need to rely on shadowing veterans to learn tasks, and productivity suffers until they “learn the ropes.” This makes it difficult to scale teams quickly in response to new opportunities, especially in fast-expanding tech startups where headcount can double within months.
Compliance and governance challenges. Operating with informal systems can make it hard to meet regulatory requirements or maintain proper records for audits. This risk is elevated for startups expanding across borders – each new market adds layers of local regulations and compliance needs. Regulatory complexity is frequently cited as the top growth hurdle for SMEs, named by nearly two-thirds of European SME leaders in a 2025 survey. Overlooking governance early on can lead to costly mistakes or penalties down the road, and may deter potential investors or partners who expect more structure.
Time lost on repetitive administrative tasks. A series of quick fixes can create a patchwork of disjointed tasks that consume precious time every day. Employees often spend at least 2 hours a day on repetitive manual work that could be automated. For small business owners, the burden is personal – one global study found they lose 96 minutes of productivity every day to inefficiencies like chasing information and switching between non-integrated tools, amounting to three weeks of lost time per year. This is time that could have been spent on innovation, strategy, and serving customers.

All these issues mean that founders and business owners end up spending more energy managing internal chaos than driving growth. Research by Time etc found that entrepreneurs spend on average 36% of their working week on low-value administrative tasks like data entry and routine coordination. It’s no wonder that growth stalls when leadership attention is constantly diverted. As an expert noted, “while fighting fires may keep things running in the short term, it simply is not sustainable” for a business aiming to scale.

In short, operational shortcuts have hidden costs. They quietly accumulate complexity that increases risk, wastes time, and slows everything down, leaving teams feeling like things are “harder than they should be”. For startups in emerging markets, these pitfalls can be even more pronounced if they lack the institutional support or infrastructure of more developed ecosystems, emphasizing the need to address these gaps proactively.

How to Build the Right Digital Foundations from Day One

A scalable business requires more than a great product or strong sales. It needs a solid operational backbone – the systems, processes, and governance structures that allow an organization to grow without collapsing under its own weight. In practice, this means putting in place the building blocks that reduce complexity and enable scale, even when resources are limited. Here are four critical foundations that support sustainable growth:

Business Setup and Governance
Getting the organizational basics right from day one pays huge dividends as your company grows. This includes establishing clear company structures, defined roles and responsibilities, proper financial controls, and compliant record-keeping. These governance practices provide stability and accountability that investors, banks, and partners look for. Research by the International Finance Corporation (IFC) shows that strong governance is literally valued in emerging markets – investors are willing to pay a premium of 10-20% more for shares in a well-governed SME. Moreover, companies that improve board oversight and formalize their management processes have demonstrated significant gains in profitability and valuation.

Good governance is not about bureaucracy – it’s about ensuring decisions are made transparently, risks are managed, and the company can meet regulatory obligations in any market. This foundation is especially vital for emerging market startups entering new territories like the UK, where unfamiliar regulatory and compliance requirements can otherwise derail growth. By building a culture of governance early, you reduce the chance of costly mistakes and create a sturdy platform for expansion.
Digital Workplace Infrastructure
Modern digital workplace tools – from cloud collaboration suites and secure file-sharing platforms to enterprise messaging and video conferencing – enable teams to share knowledge, communicate effectively, and work productively across locations and time zones. This is crucial for startups from regions like Africa, South Asia, or Latin America that often operate across borders or remote teams from the start.

There is strong evidence that digital infrastructure and connectivity have a direct impact on productivity and growth. The World Economic Forum highlights that embracing digital technologies helps SMEs lower barriers to entry, boost efficiency, and attract talent. In fact, over 85% of businesses globally recognize that adopting modern tech is a key driver of their transformation and future readiness. Meanwhile, a 2024 study by the IFC in Africa found that although 86% of African firms have access to basic digital tools (like mobile internet and phones), only a small fraction fully leverage them to improve operations – a missed opportunity for boosting productivity and growth.

A robust digital workplace doesn’t mean using the most tools – it means choosing the right integrated tools and platforms that make work simpler, secure, and scalable. For example, as small companies increasingly adopt hybrid or remote work, 8 in 10 SMBs plan to boost their IT budgets to ensure their workforce has reliable, secure technology for collaboration. By investing in cloud-based systems and modern communication platforms, an SME in Nairobi or Mumbai can operate with the same agility and real-time insight as a more established competitor in London or New York – turning its small size into an advantage rather than a disadvantage.
Operational Processes
Well-designed, standardized processes are the engine that allows a business to handle more customers, transactions, or employees without chaos. Streamlined workflows and automation reduce manual effort, improve consistency, and create end-to-end visibility of key activities. Conversely, fragmented processes can impose a huge drag on productivity – one study found that more than half of workers spend at least two hours per day on repetitive tasks in organizations that haven’t automated their workflows. Those lost hours add up quickly, draining resources that could be better spent on innovation or customer service.

Integrated business systems (for example, linking sales, operations, and finance data) can eliminate duplicate reporting and manual data re-entry, which are common pain points as companies grow. Real-time dashboards and analytics provide leadership with immediate visibility into performance, enabling faster decisions and proactive adjustments. The World Economic Forum notes that the digital transition of SMEs can speed up productivity growth, reduce human error, and improve the overall quality of outputs – precisely because good processes ensure that every part of the organization runs smoothly and efficiently.

For emerging market startups, designing robust processes can also mitigate challenges like talent gaps or uneven infrastructure. By automating routine steps and clarifying workflows, a company can maintain quality and customer satisfaction even as volumes increase. The goal is to spend less time “fighting fires” and more time implementing strategic improvements, which becomes achievable only when operations are under control.
Ongoing Support
Finally, recognize that building and maintaining the right foundations is an ongoing journey, not a one-time project. As your business evolves, so will your technology and operational needs. That’s why many growing SMEs turn to reliable partners for continuous guidance and support. According to a global study, **nearly half of small and medium businesses rely on external technology partners to help drive strategy and manage their IT, with 33% specifically favoring managed service providers (MSPs) to proactively recommend and support the right solutions.

Having a trusted partner or advisor – whether a technology consultant, digital transformation specialist, or operational expert – can ensure your systems stay up-to-date and aligned with your strategy as new requirements emerge. In fast-changing markets, including many developing economies where the tech landscape is evolving rapidly, these partnerships provide critical expertise and stability. They help you navigate new tools, scaling challenges, and best practices as you grow, so you can focus on your core business and your customers.

Technology Should Enable, Not Complicate

One common misstep is to adopt technology reactively, adding a new app or platform every time a problem arises without a bigger plan. Over time, this leads to a sprawl of fragmented systems that are hard to manage and costly to maintain – essentially creating a digital “spaghetti” of mismatched tools. Research confirms this pattern: over 90% of companies with mostly manual processes admitted they purchase technology in a piecemeal, ad-hoc way (addressing immediate needs rather than following a strategy), and they end up with a jumbled tech stack that doesn’t integrate or adapt well as the business grows.

A better approach is to align every technology decision with your business objectives and growth plans. Each tool should have a clear purpose, integrate smoothly with your other systems, and ultimately simplify rather than complicate operations. The goal isn’t to adopt more software for its own sake – it’s to adopt the right systems that truly enable your team to work smarter at scale. For example, standardizing on a unified cloud platform for email, file storage, and collaboration might serve a scaling company better than juggling five separate free apps.

As Microsoft’s own small business research indicates, digital early adopters not only grow faster, they also feel more prepared for the future. By choosing scalable solutions and implementing them thoughtfully, an SME can avoid drowning in tech complexity and instead create a tech environment that supports growth.

How Koplio Helps

At Koplio, we specialize in helping ambitious startups and SMEs – including those expanding from emerging markets to the UK – establish the right foundations to operate efficiently and scale with confidence. Our services cover the full spectrum of what a growing business needs:

Company setup and operational readiness: We guide you in creating the governance structures and compliance processes that investors and regulators expect, laying the groundwork for trust and sustainable growth.
Digital workplace implementation: We deploy modern, secure collaboration tools (like Microsoft 365 and other cloud platforms) so your teams can communicate seamlessly and access information anytime, anywhere – a critical enabler for productivity and global expansion.
Business systems and process design: Our experts design integrated workflows and data systems tailored to your operations. By eliminating redundant manual tasks and consolidating fragmented tools, we help you achieve the efficiency and consistency needed to scale without added complexity.
Ongoing technical and operational support: We act as your long-term partner, providing maintenance, training, and strategic guidance. This ensures your technology evolves with your business – echoing the best practice that many high-performing SMEs leverage external specialists to support continuous improvement.
Strategic technology consulting: We align tech investments with your business goals. From digital strategy to vendor selection, we make sure every tool serves a clear purpose and supports your future growth – so technology remains an enabler, not an obstacle.
We work alongside founders and business leaders as partners. Our focus is on practical solutions that balance immediate needs with long-term scalability, drawing on global best practices and our experience in emerging markets.

Final Thoughts

Growth shouldn’t require constant firefighting. With the right operational and digital foundations in place, businesses can spend far less time managing complexity and much more time delivering value for customers. A modern, well-governed SME is agile and resilient: it can adapt to new market demands, engage a distributed workforce effectively, and pursue expansion (even into new regions) with fewer growing pains.

By contrast, companies that ignore foundational systems often hit a wall where problems scale faster than revenue, leading to chronic inefficiencies and stalled growth. The evidence is piling up that investing early in robust processes, digital infrastructure, and governance pays off. For example, digital technologies not only improve internal efficiency but also unlock new market opportunities – 75% of companies globally expect to adopt e-commerce platforms, opening avenues for international expansion. And mobile and digital transformation in emerging economies could boost global GDP by $11 trillion by 2030, underscoring the immense economic potential of digital-enabled scaling.

Whether you’re launching a new venture, expanding into competitive markets like the UK, or modernizing existing operations, investing in strong foundations today can save enormous time, cost, and risk tomorrow. Founders in Africa, Latin America, or South Asia who start building these capabilities from day one will be better positioned to overcome the unique hurdles of their environments and thrive on the global stage.

At Koplio, we believe that every growing business should have access to the systems, expertise, and support needed to scale sustainably. The data and examples are clear: the right digital and operational foundations aren’t a “nice to have” – they are the differentiator between continuous firefighting and confident growth. The sooner those foundations are in place, the sooner you can get back to what inspired you to start the business in the first place – creating value for your customers and leading your company into the future with confidence.

Putting customers first:  DVLA and  CustomerFirst  partnering to improve Drivers Medical service

Thu, 28 May 2026 12:44:00 GMT

When CustomerFirst launched in January , we announced that DVLA was our first partner. Since then, we have been working closely with the DVLA team to understand the Drivers Medical service, its customers and how it makes licensing decisions.

Last month we shared our NewCo approach to transforming services. Building on that, CustomerFirst and DVLA have now formed a NewCo team that are working on improving the Drivers Medical service.

Our team brings together people from both organisations to look at every element of Drivers Medical across policy, operations, technology, and customer experience. Our NewCo approach helps us move faster, test ideas, and scale what works while maintaining continuity for customers.

What is Drivers Medical and why it matters

Drivers Medical matters most at moments that can feel uncertain or stressful for people. When someone has a medical condition that could affect their ability to drive, they are legally required to notify the DVLA. For many drivers, this raises questions about what happens next, whether they can drive in the meantime and what it means for their independence, livelihood or daily life.

The Drivers Medical service exists to support people through this process and to make fair, evidence-based licensing decisions. It considers medical information provided by customers and, where needed, by healthcare professionals, alongside clear legal and medical standards that support the safety of all road users.

The number of people with driving licences is increasing. People are also living longer and driving later in life. This means the annual number of medical notifications DVLA receives is rising – they expect to receive about 900,000 in this financial year. These decisions can have far-reaching effects on people’s lives and often come at a vulnerable time. It is vital that the service can meet growing demand and make timely, safe, and fair decisions.

Why DVLA and CustomerFirst are working together

Drivers Medical plays a critical role in keeping people safe on the road, but the service is under increasing pressure. Growing demand has led to a high number of cases being in progress at any one time and the average waiting time for customers to receive a licensing decision consistently exceeds the target of 50 days. For people who rely on driving for work or everyday life, wait times even within that target can be difficult.

Drivers Medical is complex. The hundreds of thousands of medical licensing decisions it makes range from straightforward to multi‑condition assessments. Up until recently, most conditions were only notifiable to the DVLA through paper-based forms. Customers often call DVLA to ask for updates on their application instead of being kept up to date. Many decisions also depend on information from third parties, including GPs and other medical professionals, which can delay decisions.

DVLA has already introduced new improvements to the service, making it more accessible for customers and simpler for staff making licensing decisions. Our NewCo team is now working to make sure the ongoing transformation delivers the best possible experience for Drivers Medical customers.

A joined-up approach to improvement

DVLA’s wider transformation ambition is to move away from fragmented services towards more seamless, end‑to‑end journeys that make sense from a customer’s perspective. Any changes to the service must keep all road users safe. The service must be easy to use and trusted. Medical licensing decisions can be complex, and people use the service in different ways.

Our priority is to make the Drivers Medical experience as smooth and joined-up as possible for customers and staff making licensing decisions. We will look at:

how people interact with Drivers Medical at different points in their journey, from telling the DVLA about a medical condition, to receiving updates on their case, to receiving a licensing decision
how information flows between Drivers Medical, other government services, and third parties, such as medical practitioners, that contribute to licensing decisions

We will take a test‑and‑learn approach, starting small to build a shared understanding of user needs and service complexity. We will then test improvements carefully before making any decisions about wider change.

What happens next

Now that we have established a NewCo team, we are working quickly to improve the Drivers Medical experience, with a clear focus on improving outcomes for customers, colleagues, and road safety.

We will do this by:

prioritising areas of the Drivers Medical journey where delays and uncertainty have the greatest effect on customers
running test‑and‑learn activities to reduce the number of cases awaiting a licensing decision, cut waiting times and improve communication with customers
rethinking the processes behind licensing decisions to better support staff and simplify decision-making

When tests are successful, we will scale these changes.

We will share our progress openly. You can follow our work on the GDS blog  and on our campaign page .

If you want to get in touch, email us at  customer.first@dsit.gov.uk

Shaping the future Fast Stream: digital and cyber security schemes

Thu, 28 May 2026 08:30:00 GMT

The Civil Service Fast Stream is the UK government’s flagship graduate programme, designed to develop the next generation of leaders across the Civil Service, and offers a structured, fast-track route into a wide range of careers, combining hands-on experience with formal training and support.

With a variety of specialist schemes available, the Fast Stream provides opportunities for graduates and civil servants to contribute to meaningful work that shapes public services and policy. Two of these are the Digital and Cyber Security Fast Stream schemes.

The Digital and Cyber Security Fast Stream schemes

The Digital and the Cyber Security Fast Stream programmes are two separate specialist schemes. Each serve as a comprehensive leadership development pathway for the next generation of innovators in the digital and cyber security professions.

The Fast Streamers are at the forefront of crucial work on high-impact projects, from improving how people access government services, to improving the UK's cyber resilience. Their work plays an important role in modernising public services and protecting the UK’s digital infrastructure.

Both the Digital and Cyber Security Fast Stream run for three years and follow a common structure used across the wider Fast Stream. Over the course of the programme, participants typically rotate through different roles and departments. They gain insight into how government works, while building the skills needed to take on senior positions.

Where they differ is in focus, skills and the types of roles participants undertake. Digital Fast Stream participants work in digital roles across government to improve how public services are designed, delivered and experienced by users. This stream has six roles. Four technical roles that include product manager, delivery manager, business analyst and user researcher. From 2025, we have also introduced two non-technical roles, that of the software developer and data scientist.

Cyber Security Fast Stream participants work in various government areas, covering both technical and non-technical roles, depending on the chosen pathway. On the Cyber Security Fast Stream programme, participants learn the practice of protecting systems, networks, programmes, the data on them, and the services they provide, from unauthorised access, digital attacks, harm or misuse.

Life at Government Digital Service (GDS)

One of the most exciting places a Fast Streamer might work is at GDS. As part of the team, they could work on areas such as:

Building platforms: Working on services which millions use in their day to day lives.

Innovation: Exploring how artificial intelligence (AI) and emerging technologies can make public services faster and simpler.

User-centric design: Ensuring that government services work for everyone, regardless of their technical ability.

Who can apply to the Fast Stream?

People with analytical minds and problem solvers from all walks of life can apply to the Fast Stream. Whether this is someone just finishing university, or already working in government, the door is open.

You need to have or be expecting a 2:2 degree or higher in any subject. It doesn’t necessarily need to be a STEM degree. We value diverse perspectives and a passion for tech.

If you are already a Civil Servant, you can apply to the Fast Stream to accelerate your career, without a degree, provided you meet the internal eligibility criteria .

Why choose the Digital or Cyber Security Fast Stream?

The digital and cyber landscape is evolving at an unprecedented pace, and the Civil Service is keeping stride. By joining either of these streams, you can actively participate in the conversations and decisions that shape the future of the UK’s digital landscape.

Here, your voice and expertise will help bring about meaningful change, placing you at the heart of innovation and improvement in public service.

Keep an eye out! Over the coming weeks, we’ll be publishing a series of case studies featuring our current digital and cyber security Fast Streamers. They’ll be sharing their experience and insights, the challenges they've tackled, and what a typical day looks like on the Fast Stream programme.

Answers in seconds, 24/7: GOV.UK Chat launches in the GOV.UK app

Thu, 14 May 2026 21:30:00 GMT

People’s expectations of digital services are changing - including their expectations of government. Research consistently shows that more people are becoming familiar with AI-powered tools in their everyday lives. Recent figures published by Ofcom suggest that just over half of adults (54%) say they use AI tools, compared to 3 in 10 (31%) the year before, and that three-quarters of online adults (75%) read AI-generated search summaries at least sometimes. This growing familiarity means many users now expect services to meet them in a more conversational, responsive way.

Today, we’re officially launching GOV.UK Chat in the GOV.UK app to do just that. This is a new way for people to get answers from government using everyday language, through the convenience of their phones, at whatever time suits them.

This is the biggest change to how people can interact with government content since the GOV.UK site was launched in 2012. Instead of navigating or searching 80,000 pages of guidance, users can now use the app to ask a question and get a conversational response, grounded in official government information and remaining firmly rooted in the principles of clarity, accuracy and trust.

This supports the vision set out in the roadmap for a modern digital government - we are creating a new generation of personalised, joined-up public services, which make it easier for people to find what they need, when they need it.

Why we’re launching GOV.UK Chat now

We’re releasing GOV.UK Chat in the GOV.UK app following a period of careful testing that has given us confidence to make it more widely available. We used the Scan > Pilot > Scale approach set out in the AI Opportunities Plan , a way of working that included one of the largest user research studies GDS has ever undertaken. We’ve previously written about the key things we’ve learned during our research and testing over the last couple of years, and how these insights have informed our delivery approach.

Since the soft launch on 26 March 2026, GOV.UK Chat has been available to GOV.UK app users without any proactive communications. This allowed us to observe how people used the service in real conditions, ensure our safety measures are working as planned, learn where Chat is working best and identify potential improvement areas. In the weeks that have followed the soft launch:

more than 7,800 people have used GOV.UK Chat

it has handled more than 15,000 questions

we're seeing the same level of performance that we achieved in our recent pilot.

What people are using GOV.UK Chat for

During our testing, GOV.UK Chat has proved particularly helpful for people navigating common but complex life moments, such as:

becoming a new parent and understanding childcare support

buying a home for the first time

exploring apprenticeships or understanding income tax

planning for retirement and understanding the State Pension.

Since the soft release demand has been strongest for answers relating to tax, driving and transport, and benefits.

In user research sessions, people told us they valued:

being able to ask questions ‘in plain English’

getting a clear starting point without feeling overwhelmed

having direct links back to the source guidance on GOV.UK.

The insights we are gaining are shaping how we continue to develop the service, so that we can expand the features and functionality of GOV.UK Chat in the ways that will help people most.

Built with safety, trust and transparency at its core

We know trust matters, especially when introducing new technology into public services. That’s why GOV.UK Chat has been designed with strong safeguards:

it clearly signposts when users should check the original guidance

it does not attempt to provide advice

it is monitored and evaluated to ensure accuracy and safety.

Before releasing GOV.UK Chat, we worked with security experts, including the AI Security Institute (AISI), to carry out thorough safety checks and test the system's strength against potential risks. We also made privacy a key priority throughout development - GOV.UK Chat lets people know they should not share personal information, as well as ensuring the system filters out any personal information that may be provided.

What’s next

This launch marks the start of a change in how people can choose to interact with government. Right now, GOV.UK Chat helps users find and understand trusted information. In the future, we plan to build on this foundation and add further features to help give people their time back.

We’ll continue to learn from how people use GOV.UK Chat in the app, improve the quality of responses, and explore how it can better support onward journeys when people need to contact departments or services directly.

Most importantly, we’ll keep listening to users - and using what we learn to make public services work better for everyone.

To access GOV.UK Chat, download the GOV.UK app. More information on how to do this is on GOV.UK

Testing a different way to improve complex public services

Thu, 30 Apr 2026 11:51:00 GMT

We all depend on public services, but too often those services fall short. Most public service transformation does not fail because teams lack skill or commitment. It fails because no single team can act on the constraints that shape outcomes for users.

In many services, responsibility is spread across policy teams, arm’s-length bodies, regulators and third parties. Decisions made in one place affect outcomes elsewhere. Teams are often asked to take responsibility for end-to-end services but lack the authority or space to address the constraints outside their control that matter most.

The result is familiar: teams improve individual parts of the system but the overall experience for users changes little.

CustomerFirst has been set up in DSIT to tackle this problem: to transform public services and test whether taking NewCo approach is the best way to do it.

Seeing and solving the whole service

Government is clear about what good looks like. The Service Standard asks teams to solve whole problems for users, not just optimise parts of a system. This intent is reinforced in the 2025 Blueprint for Modern Digital Government , which starts with the need to join up public services.

In practice, delivering on this intent is hard.

While teams are expected to think end-to-end, most delivery models remain organised around functions, funding streams and accountabilities. Responsibility for outcomes is often shared, but authority to change how the service works is not. Teams may understand what needs to change without having a clear route to make it happen.

The intent of the Service Standard is clear. The operating model for delivering it in complex, federated services is much less so.

Trying a different approach

CustomerFirst is exploring how teams can solve whole problems in practice when authority, incentives and accountability are fragmented, and how that learning can be shared so others can build on it.

We describe this way of working as NewCo .

In our model, NewCo is not a formal organisation or a fixed approach. It is a deliberate way of creating space for small, multidisciplinary teams to look at a service differently and act on what they learn.

In the context of public services this does not begin by creating permanent parallel teams or new organisations. Instead, it creates bounded space to surface constraints, test assumptions and generate evidence. Where that leads is not predetermined. Learning may be reintegrated into existing services, or it may point towards more structural change. The focus is on outcomes first, not organisational form.

This way of working is not entirely new. In its early years, GDS brought policy, design, technology and operations together in small teams with a clear mandate to redesign services end to end. A similar pattern appeared during the Universal Credit programme , which created space to re-examine assumptions after early implementation challenges. More recently, mission-focused bodies such as the Infected Blood Compensation Authority have been established to move at pace where existing structures would have constrained progress.

At CustomerFirst, we are testing whether this way of working can be applied deliberately and repeatably, rather than emerging only in exceptional circumstances. We are not changing the Service Standard; this work focuses on understanding what it takes to meet those principles in complex services and feeding that learning back to inform how they evolve.

What NewCo makes possible

At its core, this approach shifts the work from designing solutions to testing constraints.

It enables teams to focus on a small number of questions:

Which constraints most shape outcomes for users?
Where do those constraints come from? (for example, policy, funding models, governance)
Which constraints are genuinely fixed versus unchallenged?

Rather than starting with large-scale redesign, the emphasis is on creating conditions to explore these questions safely and practically, using real services.

Experience from earlier NewCo-style efforts suggests this only works when minimal conditions are in place: clear ownership of outcomes, permission to test changes within agreed guardrails, access to end-to-end capabilities, and governance that grows from the work rather than preceding it. Without these, the risk is recreating existing constraints in a new form.

This approach is therefore deliberately selective. It looks for situations where creating bounded space is likely to unlock learning that business-as-usual delivery cannot.

The test is simple: does this way of working lead to better outcomes for users?

Learning in the open

Meaningful change in complex public services rarely comes from top-down direction alone. It depends on shared understanding, practical experimentation and evidence.

CustomerFirst are committed to learning in the open — testing what it takes to solve whole problems in complex services and sharing what works, and what does not, so others can move faster.

We are already working with several government delivery bodies, including DVLA, to explore how this approach can be applied in practice. Our focus is on where targeted; evidence-led test-and-learn ways of working could unlock better outcomes for users.

We have a lot to do, and a lot to learn. Over the next two years, CustomerFirst will work differently to deliver better services for citizens and to challenge long-standing assumptions about how services are designed and run.

You can follow our work on the GDS blog and on our campaign page .

If you want to get in touch, email us at customer.first@dsit.gov.uk .

GOV.UK One Login for HMRC: how we made it happen and what comes next

Tue, 28 Apr 2026 11:52:00 GMT

After 3 years of close collaboration between the Government Digital Service (GDS) and HM Revenue and Customs (HMRC), GOV.UK One Login is now being rolled out to new HMRC users. This implementation represents a significant milestone for the cross-government project.

Since February, a proportion of users who access HMRC services online for the first time have been directed to sign in or create sign in details through GOV.UK One Login. This is another significant step towards a simpler, more consistent experience for people accessing government services online.

Over the coming months, we will be rolling this out to more new HMRC users until GOV.UK One Login becomes the standard route for everyone accessing HMRC services for the first time.

My role as the GDS team lead for the HMRC engagement is to ensure the GOV.UK One Login platform enables users’ smooth access to HMRC’s services. That means working closely with our HMRC colleagues to understand what they need, turning those needs into practical work our teams can deliver, and making sure the platform works well for everyone who uses it.

How we got here: building a shared solution

Delivering a cross-government service of this scale required deep, collaborative work including:

Framing the challenge: We started by gaining a clear picture of just how complex the project would be as HMRC services are among the largest and most-used digital services in government.
HMRC and GDS integration: The teams from GDS and HMRC partnered to understand exactly what HMRC needed, covering everything from day-to-day running of their services to the essential work of keeping our combined systems safe and reliable for everyone.
Building together: We built solutions together that work across both organisations, so that users get a seamless experience.
Testing and iteration: Rigorous testing was essential. We used user feedback to iterate on page language and layout, resolving specific points of confusion identified by users.

Go live: the moment the button was pressed

The go live day for this milestone (9 February 2026) was worth celebrating. A small group of the core GDS team travelled to HMRC offices to be there in person when we switched the service on.

The joint GDS-HMRC team is spread right across the UK, making this in-person gathering a momentous occasion. Crucially, the launch went smoothly, and we celebrated the first successful users accessing HMRC services via GOV.UK One Login, complete with celebratory cake.

The next phase: monitoring and iterating

Going live was just the beginning. We have established joint reporting measures and are now actively monitoring metrics to ensure the service continues to run smoothly and provides a positive user experience. We are tracking the:

number of people using GOV.UK One Login to access HMRC services
success rate and potential areas of ‘friction’ for users proving their identity via GOV.UK One Login
contact centre call volumes related to the new service

We will use this data to ensure we continue to meet user needs and that people can complete their required interactions with HMRC easily and effectively.

What’s next

Reaching this point has taken a huge amount of work from both teams. However, there is more to come. Our next priority is bringing existing HMRC users onto GOV.UK One Login. These are users who already have and actively use Government Gateway to access HMRC services.

We are continuing to work closely with our HMRC colleagues to take forward this essential next phase of work. Ultimately, the goal is the same as it has always been: making it simpler and easier for people to access the government services they need online.

P.S. The cake was delicious – I could have had a second slice, and I am very particular about cake.

Subscribe to the GDS blog to get the latest updates about our work.

A new business analysis course is now live on the Government Digital and Data Hub

Fri, 24 Apr 2026 09:30:00 GMT

If you have ever wondered what business analysts do in government, the Business Analysis in Government: Getting Started is your chance to learn more.

As part of the governments ambition to build a modern digital government, there’s a growing need to invest in skills, talent and capability. The Government Digital and Data (GDD) profession plays a central role in this with a goal to double its size and equip civil servants with the expertise needed to deliver better public services.

To support this, the Government Digital Service is expanding access to high-quality learning opportunities, available to all, including the new introductory course in business analysis.

This course is open to everyone, whether you work as a business analyst already and would like to develop some skills, are simply curious about the field, or are maybe even considering a career in government.

This course is the latest instalment in our “Getting Started” series and has been created in collaboration with business analyst colleagues from across government and is hosted on the new Government Digital and Data Hub where you can find a range of learning, from business analysis to user research.

Why learn about business analysis?

Business analysis is the process of solving problems and finding opportunities to deliver solutions that create value for users and stakeholders.

In government, business analyst is one of the most in-demand roles in Digital and Data, which is why it's available as one of the four Digital Fast Stream paths, alongside Product Manager, User Researcher and Delivery Manager.

The course content was produced by a cross-government team of business analysts from the Department for Education, the Department for Work and Pensions, and the Ministry of Justice, bringing real world experience to the learning.

The topics covered in the course will also be useful for anyone who would like to:

identify and develop skills and knowledge relevant to their current role
decide whether they would like to work towards a role in business analysis
collaborate effectively with business analysts in multidisciplinary teams

Learn in a way that works for you

This course is structured over three weeks with a recommended learning time of three hours per week. It is self-directed so you can learn at your own pace, and you can join and start learning at any time. The course will include:

the basic terms and concepts of business analysis
stakeholder analysis and management
working with requirements
domains of change and adding value
process modelling
how business analysis works in a government context

It’s also hosted on the FutureLearn platform, which means you can:

discuss ideas with other learners
share experiences
learn collaboratively

All courses on FutureLearn focus on learning through a mix of reading, completing tasks, and watching educational videos produced by business analysts working in government. You can keep track of your progress by taking quizzes and end-of-week tests, and by completing the learner tasks throughout the course.

Register on FutureLearn now You can join the course at any time and learn at your own pace, completely free. If you want to go further, there’s also the option to upgrade for permanent access, complete assessments, and receive a certificate to showcase your learning.

Find more learning o pportunities on the Government Digital and Data Hub

Launched in November 2025, the Government Digital and Data Hub is a national online platform bringing together digital and data professionals from across the UK public sector.

Whether you’re an experienced specialist or just starting your journey, the Hub offers a dynamic space to access world-class learning, share ideas, and engage with a growing network of professionals committed to public service innovation.

Alongside our new Business Analysis course, the hub brings together a range of community-led learning that can help you build skills and develop in your current role or kick-start your digital career in Government. Topics include accessibility, content design, product management and user research.

The hub also offers an exciting range of free, digital and data industry-led learning from leading technology suppliers. Subjects include Artificial Intelligence, Cloud and Platforms, Data, and Cyber Security.

Explore the Government Digital and Data Hub now.

Stay connected

You can also find out more about upcoming learning opportunities and events by subscribing to our updates .

If you have any queries or would like to collaborate with the team on our future courses, please contact the GDS Learning & Engagement team at gddlearningteam@dsit.gov.uk.

GDS Local: building the foundations for digital collaboration

Wed, 15 Apr 2026 15:18:00 GMT

Local government is an essential part of delivering the ambitious vision of the Blueprint for modern digital government , and creating a joined-up government that works as a joined-up system, where digital infrastructure is shared, services are interoperable, and residents – especially those who depend most on public services – experience services that work for them.

GDS Local exists to support this ambition. Our role is to catalyse, enable and champion from the digital centre. We will do this by building a shared vision for local government technology – that helps us deliver on the ambition of the blueprint – working with local councils to reduce barriers to digital transformation, and making sure the products and standards coming from the centre of government help to make this change happen on the ground.

In our work, we want to improve everyday experiences for people – especially for those with complex or overlapping needs who rely on local services – while also helping local councils respond to rising demands and budget pressures, by using new technology to build productive, efficient services.

In this update, we want to share what we’ve been learning and doing against our 4 priorities:

Co-developing a strategic technology vision for local government
Unlocking usable and shareable data, working with the Office for the Chief Data Officer in GDS, the Cabinet Office and the Ministry for Housing, Communities and Local Government (MHCLG)
Catalysing component adoption and de-risking reuse, starting with GDS products
Enabling the conditions for collaboration, meaning councils and central government can work together more effectively and strategically – underpinning our work on the first three priorities

Co-creating a strategic vision for technology

Our aim is drive towards a shared, credible strategic view of local government technology. We want to work through the design and delivery of digital council services to enable better procurement decisions, reduce duplication and shift market conditions.

This month we delivered one of our commitments from the GDS Roadmap: a technology reference model, our Local Government Architecture Model , which is part of our ongoing work to co-create a technology vision for local government.

This is the first step in building a common way of describing the components of a local government technology stack – the set of tools and products and technologies that drive the organisation – and then what good practice looks like for each of those components and how to source them well.

We’re doing this with councils – we have run 5 workshops with councils and suppliers, through monthly online webinars averaging around 60 participants per session, and run surveys of council officers to ensure we have our finger on the pulse of what councils are doing and what matters to them.

We’ve also been researching existing council technology architectures and exploring how we might provide an architecture mapping tool to councils free of charge. This is just the start; we know that, coupled with the opportunities and challenges of Local Government Reorganisation (LGR), there is a role for GDS to really help here.

Opening up the GDS offer

One of the biggest opportunities in the GDS Blueprint is common digital infrastructure: components built or designed once, used across the public sector, rather than every organisation solving the same problems from scratch. Some of the components in this stack could be products that GDS has developed or is developing. We’re working with those product teams to understand council needs, test their assumptions and help them extend their offer to councils.

For example, for the GOV.UK app, we’ve been working with local authorities to test assumptions about council needs, surfacing integration challenges around the app like fragmented data, supplier lock-in, inconsistent APIs, and questions about local service branding. With the app team, we’ve been running a proof of concept in the app that incorporates waste collection data, which has given us useful insights on data integration and what enabling scale will require.

We’ve also been helping to boost digital talent in local councils by opening up our training programmes. Around a quarter of participants on the GDS AI Accelerator, an AI training programme for the public sector , are from local government, after we helped shape and promote the offer in relevant local government forums. We will be doing lots more of this going forward.

Bringing people together

Collaboration is crucial for us to successfully deliver our ambitions – and for local government in accessing and developing modern, digital services. In November we ran a Homelessness and Rough Sleeping Hackathon in Birmingham, bringing together council teams, with both digital and service practitioners, to work on some of the hardest problems in local service delivery. We were pleased to follow that up by hosting the winning teams to present their work to Minister for Digital Government and Data, Ian Murray, from DSIT, and Minister for Building Safety, Fire and Democracy, Samantha Dixon MBE, from MHCLG, demonstrating the collaboration between departments and the interest in this work from the highest levels.

At Innovation 2026 , we ran a workshop with Liverpool’s Office for Public Sector Innovation and GDS data teams, bringing together around 50 council officers. We’ve heard clearly from hundreds of people across local government that they want more communication, more clarity on our plans, and more ways to get involved, which is why we have created the GDS Local newsletter where you can follow our work.

To help us shape that engagement offer, please do spend 5 minutes on this questionnaire so we can make sure our communications are working for you.

The vision in the Blueprint is genuinely ambitious, and we want to build it with you. If you have any thoughts you’d like to share, please get in touch with the team at gdslocal-info@dsit.gov.uk .

Subscribe to the GDS blog to get the latest updates about our work.

Designing a Configurable Platform Architecture

Fri, 30 Jan 2026 11:37:25 GMT

Designing a Configurable Insurance Platform Architecture: A Case Study

When our team joined a growing insurtech platform as architecture advisors, the challenge was clear. The platform had to be far more than a basic quote comparison tool; it had to seamlessly serve multiple stakeholders – from consumers and insurance partners to internal operations – without turning every small change into a full engineering release. This is a common scenario for many ambitious startups building insurance distribution platforms. They must handle everything from customer quote comparisons and policy purchases to partner product configuration, payments processing, and policy issuance, all in a secure and scalable manner.

To meet these goals, we adopted an architecture that is intentionally modular and independently deployable, yet unified by a shared insurance domain model. In practice, this meant splitting the system into distinct applications (for example: a consumer-facing app, a partner portal, a core API service, content management, etc.), each isolated but working together through well-defined interfaces.

Why go modular? By decoupling the platform into separate services and frontends by role and concern, each part can evolve at its own pace. The consumer-facing application can iterate quickly to improve user experience. The partner portal for insurers or agents can have its own release schedule, enabling frequent product updates without affecting customers. The core API and backend services can focus on business logic, data consistency, and integration with third-party systems. This reduces risk – a change in one area (like adding a new insurance product) doesn’t necessitate redeploying the entire platform or introducing unintended side effects elsewhere.

Keeping Business Logic Behind the API – Protecting IP and Agility

One of our first major recommendations was to keep the customer’s front-end experience as lightweight as possible. In concrete terms, the customer-facing web application was built as a modern single-page app (using React + Vite), focused on the user journey: capturing quote details, comparing insurance options, guiding the user with chat assistance, and handling checkouts and policy retrieval. However, crucial business logic – especially insurance pricing algorithms, product rules, and provider configurations – was deliberately kept out of the browser.

Why make this distinction? Pricing rules and product configurations in insurance can be commercially sensitive intellectual property. If we had pushed all of that logic to run in the customer’s browser, it might have made initial development easier, but it would compromise security, maintainability, and even competitive advantage. By keeping the complex product model and rating algorithms on the server side (behind the API layer), the platform ensures that proprietary formulas and partnership data remain protected. This approach also means that changes to pricing or product rules can be made on the server and rolled out immediately – without requiring an app update for users or exposing logic that could be reverse-engineered. In short, this was an investment in long-term flexibility and security over short-term convenience.

Decoupling Partner Configuration from Customer Flows

Our second major architectural decision was to separate the partner configuration lifecycle from the customer-facing application lifecycle. The platform’s partner portal (often called a product studio for insurance partners) allows insurers and product managers to model their offerings – defining insurance products, coverage options, question flows, pricing rules, and product states. However, the customer quoting API does not fetch these configurations in real-time from the partner portal with every quote request. Instead, the platform implements a controlled configuration synchronization pattern.

How it works: When a partner publishes a new or updated insurance product in the partner portal, the platform normalizes and packages that product configuration via an automated pipeline. Essentially, the system compares the new configuration to the current production version (using hashing and versioning), and if changes are detected, it triggers a managed sync process. This process bundles the updated product rules and data into a deployable artifact (in our case, an AWS Lambda Layer – a versioned container of code/data) through a serverless orchestration workflow. The updated configuration package gets attached to the quoting service’s runtime environment, such that the quote-generation functions load the latest product configuration at startup. The changes are then cached in-memory for fast access during quote calculations.

Why do this? This approach carefully balances freshness and stability. Reading directly from the partner configuration store on each quote request might offer real-time updates, but it introduces runtime coupling between the live quoting process and the authoring system – raising risks of latency, inconsistent experiences for customers, and complicated debugging. On the other hand, forcing every configuration change to go through a full application redeployment would slow down business agility, since even minor product tweaks would need engineer involvement and operational overhead. Our configuration sync pattern provides a happy medium:

Consistency and auditability: Every published product setup is hashed and versioned, ensuring that the quoting engine uses a deterministic, traceable configuration snapshot. If something goes wrong with a new product setup, the team can roll back to a previous known-good version quickly.
Operational independence: The consumer-facing experience isn’t directly dependent on the partner portal’s uptime or performance. The quoting API always has a local, in-memory copy of the latest published product rules, unaffected by any issues or maintenance happening in the partner system.
Controlled speed of change: While not "real-time," the synchronization is automated and swift (e.g. within minutes or faster), giving partners near-immediate updates to their products in production. At the same time, the slight delay creates a buffer for validation and governance, so all changes go through a promotion pipeline with checks instead of instantly affecting users.

This decoupling of lifecycles means that the platform can incorporate continuous product innovation without compromising the stability of the core customer experience. Partners feel empowered to iterate on their offerings, and the business can onboard new insurance products or adjust pricing logic quickly – all while customers enjoy a consistent, high-quality experience.

Security and Ownership as First-Class Concerns

From day one, we treated security and data ownership as tier-1 architectural considerations. Insurance deals with sensitive personal information and financial transactions, so the architecture needed robust safety measures baked in at every layer:

Strong Identity & Access Control: We integrated a managed identity platform (like AWS Cognito) to handle authentication securely, along with role-based access controls. For data storage (for example using cloud databases like Amazon DynamoDB or relational stores), we implemented owner-based access rules. This means customer records (quotes, policies, submissions) are inherently tied to their creator’s identity, and the system automatically enforces that only authorized users – like the record owner or a permitted admin – can access each resource. This approach drastically reduces the risk of data leakage across customers or partners.
Custom Authorization for Multi-Tenancy: Some API endpoints require extra checks beyond basic identity. We used a dedicated API gateway authorizer to validate that, for instance, an insurance partner or internal operator only accesses data that they should (e.g., only quotes related to their own customers or region). This extra layer of security ensures that any attempt to bypass the front-end and directly call backend APIs still won’t expose data or functions to the wrong party. Such defense-in-depth is essential in regulated environments like insurance.

By designing with clear security boundaries , the platform can safely handle multiple roles (customers, agents, underwriters, support) in one environment without risk of cross-access – a critical requirement in financial and insurance systems.

Hiding Complexity: Integrations Behind Stable APIs

Modern insurance distribution involves a lot of moving parts – payments processing, digital wallet integration, vehicle data lookups, CRM systems, content management, and more. Rather than letting these external dependencies creep into the front-end or client side, the platform’s API layer fronts all such third-party services . This means the customer app and partner portal interact with a singular, stable, domain-specific API for all their needs, while the API itself handles orchestrating calls to external services like payment gateways, insurance policy issuers, content management systems, and so on.

This design brings two big benefits:

Simplicity for front-end development. The client applications don’t need to handle a patchwork of different external protocols or SDKs; they just speak a consistent language with the platform’s API. This allowed front-end developers to focus on a great user experience and domain-specific flows, rather than wrestling with the nuances of various third-party integrations.
Security & maintainability. By keeping all third-party credentials and interactions on the server side, we reduce security exposure – API keys and secrets never touch the user’s browser. Plus, if one external service updates their API or has an outage, the changes are isolated within the integration layer. The core platform can adapt without breaking the overall experience, as long as the internal API contract remains consistent.

Essentially, the API acts as a protective and simplifying façade , giving the platform freedom to swap out providers or add integrations without having to rewrite the front-end or risk leaking implementation details.

AI as a Bounded Feature, Not a Free-for-All Chatbot

During the project, there was also an opportunity to infuse the platform with AI capabilities, for example, to help users compare insurance options or assist with form filling through an AI agent. Our guidance was to treat this AI component as a bounded, product-specific feature rather than an open-ended chatbot. That meant carefully constraining what the AI assistant could do and see, depending on context. For instance, in the quote comparison screen, the AI can explain differences between insurance options and point out relevant policies. In a quote intake flow, it might help users by retrieving existing saved data or guiding them through the questionnaire – but always within predefined roles and using only approved data and actions.

By architecting the AI as a tool-using assistant with domain-specific modes, the team maintained control over the AI’s output and ensured compliance with regulatory boundaries (e.g., not offering unqualified advice, not hallucinating non-existent coverage details). The AI feature is implemented with robust prompt design and oversight (leveraging managed services and large language models in a controlled fashion), making it a reliable part of the product rather than a risky experiment. This approach underscores an important principle: in a serious domain like insurance, AI is most valuable when carefully integrated into the platform’s architecture with clear constraints and goals.

Architectural Outcomes and Lessons

The resulting platform architecture was built around clear ownership and separation of concerns , with each part of the system having distinct responsibilities:

Consumer Application: Handles the user journey (quote input, comparison, checkout, claims, etc.) and the overall user experience. It’s a lightweight web application prioritizing responsiveness and clarity, while delegating heavy business logic to the backend.
Partner Portal: Provides insurers and distribution partners with a product studio to define insurance products, manage pricing rules, and oversee their offerings. It supports partner operations like product configuration and lifecycle management independently of the consumer app’s deployment cycle.
Core API & Backend Services: Enforces all business logic, pricing calculations, policy issuance workflows, payments processing, and integration with external services. It acts as the central brain and gatekeeper of the platform – ensuring security, consistency, and providing a stable set of domain APIs for the frontends to use.
Data & Identity Layer: Utilizes secure identity management (for user login, partner access, admin roles) with fine-grained data ownership rules. All customer data and policy records are stored in multi-tenant databases (such as a NoSQL store), with the architecture ensuring each user or partner can only access their own data unless elevated privileges are explicitly granted.
Configuration Orchestration: A governed configuration promotion mechanism (using cloud orchestration tools, e.g. serverless functions and pipelines) that monitors product changes in the partner portal, validates and packages them, and safely deploys updated configuration to the quote engine. This ensures the system’s behavior is both easily adjustable and auditable.
Content & Extensibility: A headless content management system and developer documentation site accompany the platform. This supports marketing needs (for example, a marketing website) and provides a space for internal documentation or partner integration guides – helping the overall ecosystem evolve.

Throughout the project, we kept coming back to a guiding question: where should change happen in the system? This turned out to be the most pivotal architectural decision of all. We deliberately placed different kinds of change in different parts of the architecture:

User-facing experiences can change rapidly (through frequent updates to the web UI and app without high risk to the core business integrity).
Insurance products and business rules can change via configuration through the partner portal without requiring full deployments.
Pricing algorithms and integration logic change only on the server side , under rigorous testing and version control, preserving trust and consistency.
AI behavior evolves in controlled increments , rolled out carefully as a product feature.

By deciding these boundaries early, we provided the platform with a blueprint for sustainable evolution. Each layer can innovate at its ideal pace and level of governance. The customer-facing app can experiment and improve UX continuously. The partner experience can empower insurers and agents to iterate on products swiftly. And the core backend remains a source of truth that is secure, audited, and modifiable by the engineering team without client disruption.

In summary, the strength of this architecture lies not in any single tool or framework, but in the clarity of its separation of concerns and its strategic approach to change management. The platform is now robust enough to meet rigorous industry needs: modular enough for the engineering team to extend without major refactoring, configurable enough to onboard new insurance products and partners rapidly, fast and friendly enough to satisfy end-users with immediate quotes and smooth service, and secure enough to handle sensitive data and workflows in a regulated space.

For technical founders and CTOs, we believe this case underscores the key principle: design your architecture with a keen eye on where change will occur as you scale. By laying down modular components and clearly defining which parts of the system own which responsibilities (and thus, where future changes should be made), you create a platform that evolves gracefully. This approach not only avoids costly rewrites but also yields a durable competitive advantage — the ability to adapt quickly to market needs and opportunities without sacrificing stability or control.

Join AI Engineering Lab and help shape the future of digital government

Mon, 12 Jan 2026 09:00:00 GMT

2026 is set to be an incredibly exciting year for the digital industry, particularly across the public sector, and it all begins with a major opportunity for you to help shape the future of digital government.

To start the year, the AI Engineering Lab is driving AI adoption across public sector software teams by helping them to implement AI coding assistants by offering technical support and training for anyone who uses them.

This work builds on the positive results of the AI Coding Assistant (AICA) trial we ran in 2025, which found significant potential for improving efficiency and delivery.

Most importantly, this is a project which will evolve and adapt as we evaluate and learn along the way.

Here’s why you need to get involved, especially if you’re part of the digital and data in the public sector!

What is AI Engineering Lab?

AI Engineering Lab is being led by the Engineering Excellence team in the Government Digital Service (GDS), and its role is to drive sustained and impactful adoption of AI tooling in software engineering teams across the public sector. This includes the training and management of AI enablement, not just the technical implementation of tooling.

Its main aims are to:

integrate AI tooling for teams throughout the entire software development lifecycle

upskill government digital and data professionals to make the most of AI tooling

monitor and evaluate where AI tooling can enhance digital service delivery

AI Engineering Lab is a user-led project, which means that your input will help to determine how, where and why AI is used, and where it shouldn’t be.

What’s happening in early 2026

Starting in early 2026, we’ll roll out licences for leading AI tools, including:

GitHub Copilot

Gemini

Amazon Q or Kiro

Claude

We’ll be adding more tools later on, and these licences will be available for up to six months. During this time, we’ll also provide support for technical setup, integration and training.

These tools will be available to all public sector departments and organisations, although there are a limited number of licences available.

We’ll also be organising events such as hackathons and community meetings. You don’t need to be an expert to take part – anyone with an interest in how AI can or should be used in software engineering is welcome to join in.

Get involved

This is a unique opportunity to help shape the future of digital government, and there are many ways that you can get involved. These include:

signing up for licences for AI coding assistants

accessing training and guidance

providing feedback and examples of how these tools are used in your organisation

coming to events and sharing your experiences

To take part in AI Engineering Lab, contact us at gdsengineeringexcellence@dsit.gov.uk

Innovation in action: highlights from the Local Government Innovation Hackathon

Thu, 04 Dec 2025 09:00:00 GMT

A data-driven tool to match homeless people with suitable temporary accommodation has been crowned the winning idea at the Local Government Innovation Hackathon on This exciting and collaborative event hosted in partnership with the Ministry of Housing, Communities and Local Government (MHCLG), the Local Government Association (LGA) and Birmingham City Council brought together 150 individuals across 70 organisations within local and central government, the private sector and academia to collaborate and innovate. Participants were asked to create solutions to challenges relating to homelessness, rough sleeping and temporary accommodation within 24 hours.

Team ‘Match’, the overall winners, impressed judges with a prototype that uses intelligent matching to reduce the time people spend in emergency and temporary accommodation. Their solution also included a voice-note feature to capture conversations with individuals facing homelessness, ensuring a personalised user experience while reducing administrative burdens through AI.

The team will now pitch their prototype to Ministers from the Department of Science, Innovation and Technology (DSIT) and MHCLG - a fantastic opportunity to showcase what can be achieved when local government, central government and tech partners innovate together.

Collaboration at its best

Over two days, participants collaborated to tackle pressing challenges around homelessness, rough sleeping and temporary accommodation, supported by Amazon Web Services - one of MHCLG’s Cloud Compute partners.

Building on the recent launch of GDS Local , a new unit within Government Digital Service (GDS) established to catalyse digital collaboration between GDS and local government, the hackathon showcased what we can achieve when we work collaboratively. The aim throughout was clear: to help councils deliver better, more joined-up support for people experiencing housing instability and homelessness.

Participants heard from speakers including Emily Middleton, Director General for Digital Centre Design in GDS, Kat Sexton from Birmingham City Council and Theo Brown from third sector organisation Rene House. The event was endorsed by Ministers from GDS and MHCLG, with Minister Dixon opening day 2 by focusing on the human impact and why this work really matters.

What we achieved

Fourteen teams made up of colleagues from central and local government, policy specialists and developers worked over 24 hours to design and prototype solutions aimed at improving support for vulnerable individuals across the UK.

Suggested solutions responded to three challenge statements:

using data and AI to predict and prevent homelessness
AI-driven outreach and system efficiency for homelessness and rough sleeping services
optimising temporary accommodation allocation through data-driven insights

The outputs were incredible, with teams showcasing working prototypes in under 24 hours.

Winning team

An impressive judging panel made up of specialists from GDS, MHCLG, Birmingham City Council and the voluntary sector chose Team ‘Match’ as the winners. Their data-driven solution helps reduce the time people spend in emergency and temporary accommodation through intelligent matching of individuals to suitable properties.

Well done to Team ‘Match’ for a brilliant presentation and demonstration of their tool, and to all the contributing teams who worked so hard and produced fantastic solutions to tackle these important problems.

Thank you

A huge thank you to everyone who took part and dedicated their time and energy to this event.

If you’d like to find out more about GDS Local or be the first to hear about future events please register for our mailing list or find out more on our GDS Local GOV.UK page .

Driving innovation through the Civil Service AI & Data Challenge – enter now

Fri, 17 Oct 2025 12:34:00 GMT

Government Digital Service has partnered with NTT DATA, Global Government Forum and the Cabinet Office for the fourth year running to deliver the Civil Service AI and Data Challenge. This competition empowers civil servants in all departments, professions and grades to develop innovative AI or data solutions that improve public services and drive government priorities forward.

With the government's focus on kickstarting economic growth through innovation and technology adoption, this challenge offers a unique opportunity to contribute directly to national objectives while developing civil servants’ skills, contacts and careers.

Transforming ideas into national impact

Previous winners demonstrate the real-world potential of this competition. Nick Tomline , Earth Observation Higher Data Scientist at Natural England, led the AI4Peat project that won in 2022. Three years on, his team has created the UK's first comprehensive map of peatland drainage networks using artificial intelligence.

"The AI4Peat project filled a critical national evidence gap," explains Nick. "Previously, mapping drainage systems in peat bogs was impossible at national scale due to manual mapping constraints. Our AI approach now enables us to prioritise restoration efforts for maximum environmental benefit."

The project's impact extends far beyond environmental mapping. By identifying where drainage networks exist, the team supports biodiversity net gain and reduces CO2 emissions through targeted peatland restoration. They've also established valuable partnerships with technology leaders Microsoft and Databricks while building specialist AI expertise within government.

"We've created an end-to-end pipeline that gives Natural England and Defra capability to solve other geospatial problems at national scale," Nick adds. "The 'AI4' approach we pioneered is now being applied across other government use cases, offering substantial efficiency opportunities."

John Saunders, a product manager at the Ministry of Justice, led Project Constellation to victory in 2024. His experience highlights how the challenge accelerates both professional and personal development.

"The competition proved to be a catalyst for growth," John reflects. "I stepped outside my usual remit, took on unfamiliar challenges, and developed new skills – particularly in crafting compelling presentations for senior judge panels."

Career-defining opportunities

Both winners emphasise the Challenge's unique professional development benefits. Participants gain hands-on project management experience, team leadership skills, and the ability to present confidently to senior audiences. The competition also creates networking opportunities to work with senior Civil Service leaders you wouldn't normally encounter.

"The collaborative, cross-department approach enabled us to break down traditional silos and address government priorities more effectively," explains the AI4Peat team. "The competition provided a springboard for turning our innovative idea into a high-profile project, opening doors that couldn't have come about through conventional means."

John's experience mirrors this sentiment: "Winning created opportunities to connect with senior leaders across the Civil Service and exchange ideas in a collaborative, high-energy environment. It's had a tangible impact on my career trajectory."

What you need to know

Applying for the Challenge is simple, civil servants only need to draft a 400-word submission, exploring an idea to make better use of AI or data in government. Participating in the programme requires commitment and support. Civil servants will need backing from a line manager, as you will need to dedicate part of your working week to developing an entry and collaborating with team members.

However, both winners strongly recommend participation. "It's a rare opportunity to strengthen stakeholder engagement skills, sharpen strategic thinking, and test your ability to take concepts from idea to delivery in a high-profile, fast-paced setting," says John. "The connections you make could shape the next stage of your Civil Service career."

Ready to innovate?

The Civil Service AI and Data Challenge 2025 is now open for entries. Whether your idea tackles economic growth, public service efficiency, or emerging technology adoption, this competition offers an unparalleled platform to drive meaningful change.

Entrants can also apply to join one of the interdisciplinary, cross-departmental project teams, supporting a data or AI idea. The deadline for applications is Wednesday 5 November 2025.

Don't miss this opportunity to transform innovative thinking into national impact while accelerating your career development. Enter the Civil Service AI and Data Challenge today at https://datachallenge.uk/. 

Turning research into action at the AI in Government and Academia Summit

Fri, 03 Oct 2025 09:15:00 GMT

Artificial intelligence (AI) is transforming how we live, work, and deliver public services. The UK government is committed to ensuring this transformation is responsible, effective, and beneficial to society. Last week, we took another important step with the first AI in Government and Academia Summit .

Co-hosted by the Government Digital Service (GDS), part of the Department for Science, Innovation and Technology (DSIT), and MetroPolis , the policy think tank of Manchester Metropolitan University (Manchester Met), as part of Dr Spinelli’s Chancellor’s Visiting Fellowship in AI, the event brought together around 300 public servants, technologists, and researchers from across UK universities.

The summit marked the third in-person event of the AI community for the UK government, which has grown to over 4,500 members across 340 public sector organisations in the past year. The summit was inspired by the AI Opportunities Action Plan to showcase how collaboration between government and academia can translate cutting-edge research into practical tools, guidance, and services for the public good.

Building momentum together

Collaboration between government and academia is at the heart of the AI Opportunities Action Plan, which sets out a bold agenda to use AI to modernise service delivery in the UK by achieving more with less and stimulating economic growth. To enable this change, government is encouraged to establish partnerships with world-leading academics and entrepreneurs to build long-term capability, attract talent, and ensure the UK remains at the forefront of safe and ethical AI.

This approach is already delivering results. The AI Playbook for the UK government , developed by GDS in collaboration with universities, industry, and local councils, provides government departments and public sector organisations with up-to-date, high-quality guidance on a wide range of AI technologies, enabling their responsible use in the public sector.

Collaboration with academia has also enabled GDS to launch the AI Insights series by offering digital and data professionals explainers on the technical aspects of AI implementation in government. Published bimonthly, these articles are written by government technologists and peer-reviewed by academics from UK universities to ensure they reflect the latest research developments. Each of these initiatives demonstrates the value of pooling expertise across sectors.

By bringing together senior government officials and university professors, the summit provided a space to explore the potential and challenges of this collaboration and to understand how academic research developed in the UK can make a real impact on the lives of the people we serve.

Highlights from the Summit

Co-chaired by Dr Tommaso Spinelli , AI Senior Change Manager for the UK government and Research Fellow in AI, and Prof. Keeley Crockett from Manchester Metropolitan University, the day featured a rich agenda including a mix of keynote addresses by senior government officials and academic experts, a panel discussion, and the presentation of peer-reviewed research papers covering various aspects of AI technology and theory.

Professor Darren Dancey of Manchester Metropolitan University opened the event by outlining how the university has embedded ethical AI into its teaching, including a new Embedded AI Framework launching in 2026. This framework will integrate AI into disciplines such as fashion, biomedicine, and geography, while Manchester Met’s Centre for Digital Data Research (CDDR) is already applying advanced technologies and big data to challenges ranging from climate change to public health in order to demonstrate the value of multidisciplinary collaboration.

Professor Tom Crick MBE , Chief Scientific Adviser at the Department for Culture, Media and Sport (DCMS) and Professor of Digital Policy at Swansea University, offered a strategic government perspective on the future of AI. He emphasised the need to consider the human, societal, and cultural dimensions of digital transformation, not just the technical ones.

Dr Ed Steele , IT Fellow at the Met Office and Gov.uk, provided an overview of how AI and machine learning are being practically applied to enhance environmental intelligence. His talk highlighted improvements in weather prediction accuracy, alongside ongoing experimentation with new technologies and models to reduce costs and boost efficiency.

Dr Antonio Cordella , Associate Professor at the London School of Economics, delivered a thought-provoking deep dive into the implicit values embedded in technology. He illustrated how algorithms, system designs, and architectures are never value-neutral, but instead carry assumptions and choices that must be aligned with an organisation’s ethics and strategic agenda.

The final keynote, delivered by Professor Keeley Crockett of Manchester Met, offered a forward-looking reflection on the day’s insights, centred on the theme of ‘People Powered AI’. Building on Manchester’s internationally recognised People’s Panel for AI initiative, and the PEAs in PODs project, Professor Crockett demonstrated practical ways to engage communities in shaping AI technologies. The Peoples Panel for AI, led by Manchester’s Digital Strategy team in partnership with Manchester Met, promotes inclusive consultation and digital equity by serving as a replicable model for the wider public sector to ensure that new technologies are developed with community needs and values at their core.

The panel discussion, ‘Forging the Future: Cross-Sector AI Collaboration for Public Good’, emphasised the importance of public trust, the human impact of digital change, and the infrastructural barriers that still exist.

Showcasing research with impact

The conference also provided a platform for new research, with submitted papers presented across three streams: AI theory, case studies, and practical projects in local councils.

These fascinating topics included:

using large language models to transform open data searches
using AI tools supporting local authorities with temporary accommodation
real-time translation for council services
AI methods for transparent public reporting

The diversity of perspectives presented by these contributors reflected the shared challenges and opportunities across public service delivery. We’re working to make these research papers publicly available.

Looking ahead

By bringing together public servants, researchers, and technologists, the summit created a space to forge new ideas and strengthen existing ties.

The focus on showcasing real-world projects and peer-reviewed research highlighted how the UK is moving from talking about the potential of AI to actively delivering on it. Our AI community believes that collaboration between academia and government remains essential for building long-term AI capability, attracting global talent, and translating cutting-edge research into real-world impact in safe and responsible ways.

We look forward to seeing the new partnerships and projects that will emerge from this conference.

If you want to get involved, join the AI Community of Practice to connect with people across government, attend monthly meet-ups, and stay updated with newsletters.

If you’re a civil servant, you can also enrol in our free e-learning courses on Civil Service Learning to gain practical skills in safe and effective AI use.

Finally, the One Big Thing initiative later this year will have all civil servants taking part in AI training.

Ready to grow your digital skills? Get Tech Certified this Autumn

Wed, 01 Oct 2025 09:30:00 GMT

What is Get Tech Certified?

Get Tech Certified is an opportunity for anyone in the public sector to advance their digital skills. Launched by the Government Digital Service, it follows the success of last year’s Get Cloud Certified training campaign.

From 1 October 2025 to 31 December 2025 , all civil and public servants can access over 200 free learning and certification pathways across digital and technology areas including:

Cloud & Platforms technology
Machine learning Engineering & AI
Data
Software Engineering
Security
Quantum
FinOps

These range from introductory courses to advanced qualifications, and short to full courses.

The offer is made possible through partnerships with 16 leading technology training providers , and is completely free to access.

Who is it for?

This programme is open to:

anyone curious about digital and tech skills
Government Digital and Data professionals
public sector colleagues working alongside digital teams.

Whether you’re looking to build confidence, deepen your expertise, or explore a new direction in digital and data, Get Tech Certified is designed to support your development.

Real stories, real impact

Explore courses and register today

Courses cover a wide range of topics including cloud computing, cybersecurity, data analysis, agile delivery and more.

You can now access the Get Tech Certified courses and other digital learning opportunities via the Government Digital and Data Hub – a new online hub for everyone working in digital and data roles across the UK public sector. The Hub brings together learning, career pathways, events and community updates to help public sector professionals develop the skills and connections needed to deliver better public services.

This offer is available until 31 December 2025 .

How do we test and assure AI in Government?

Mon, 22 Sep 2025 09:00:00 GMT

How do we test and assure AI solutions - is one of the most common questions we hear across government right now.

AI is changing how we deliver public services. For example, the GOV.UK Chat is experimenting with generative AI to help people navigate more than 700,000 pages of GOV.UK content and find the information they need more easily. Other projects, like the Crown Commercial Service’s recommendation system, use AI to provide smarter agreement recommendations for customers based on procurement data. You can find more examples of how AI is being trialled and adopted across government in the UK Government AI Playbook . However, they are not traditional software and we can’t test them in the same old ways.

AI systems learn from data, behave in probabilistic ways and can even change over time. They might work well in a lab, but fail in real-world conditions if we don’t test them rigorously and thoughtfully.

Evaluating AI isn’t about proving it works once. It’s about continuously asking questions like:

Is it safe, fair, unbiased and effective?
How will it behave in real-world conditions?
What happens when things change?
How do we spot and fix unintended behaviour?

AI systems bring together technical infrastructure and probabilistic models. That means we need both system testing and model evaluation. Testing looks at the whole AI system (infrastructure, APIs and interfaces). Evaluation focuses on the AI model, asking how well it meets key quality attributes. Together, they give assurance that AI systems are safe, fair and accountable.

Hence, we need a framework supporting teams to embed this kind of critical thinking throughout the lifecycle of AI testing and assurance.

A shared framework for public sector teams

To help with this, the Cross-Government Testing Community has developed the AI Testing Framework for the public sector. In March, we brought people together for a workshop to pool our experience and ideas on how best to test and assure AI systems. Since then, the framework has been shaped through working group reviews and community feedback. We published a beta version in June and you can see everyone who contributed in the Review Log .

This framework offers a shared baseline for testing, model evaluation and assurance that any department can adapt to its needs. It’s designed to support everyone involved in AI projects, from policy teams scoping ideas to technical teams developing and deploying solutions.

It recognises there’s no single way to test or evaluate AI. Different types of systems (rule-based, predictive models, generative chatbots) demand different approaches.

What’s inside the framework

The framework is built around four core elements to help teams think rigorously about testing AI:

1. Principles for testing AI systems

It sets out 11 clear principles, from designing context-appropriate tests to monitoring for change over time. These principles help teams plan quality from the outset and embed responsible practices throughout.

2. Core quality attributes

AI testing and evaluation must consider things like fairness, explainability, robustness, autonomy and evolution. These attributes shape the questions teams need to ask while testing or assuring, and the risks they need to manage.

3. Continuous Defensive Assurance Model

Testing isn’t a one-off step. The framework provides guidance for every phase of delivery: planning and design, data preparation, development, deployment and ongoing monitoring. At each stage, testing produces evidence, evaluation helps to interpret it in context and assurance provides confidence that the system, or any changes, are ready to go.

4. Modular testing strategy

Not every project needs every test. The framework includes a modular strategy that lets teams choose and combine testing activities based on the type of AI system, its use case and the level of risk involved.

Why it matters

Testing or evaluation isn’t just a box to tick.

Without good testing and evaluation, we risk:

producing unfair or biased outcomes
failing in unanticipated scenarios
degrading over time without detection
eroding public confidence in government use of AI

Good testing and evaluation helps us:

catch problems early
make better design choices
build systems that are safe, fair and effective
maintain trust in government use of AI

This framework helps us do that in a clear, structured way.

Designed for public sector use

We’ve built this framework with the realities of public sector work in mind. It:

aligns with the UK Government AI Playbook and principles
encourages transparency, documentation and accountability
supports proportionate assurance, recognising that not all AI projects have the same risks
went through a cross Government workshop and working group reviews

It’s there to help teams make informed, responsible decisions about how they test and assure their AI solutions.

Learn more

You can read and contribute to the framework here:

AI Testing Framework for Public Sector

276d07bb

Update from the GDS Responsible AI Advisory Panel

What do we mean by ‘Responsible AI?’

How this fits with existing government activity

Early Insights

The panel’s focus

What happens next

Data maturity: the foundation for AI ready public sector data

What is data maturity, and why does it matter?

What we set out to do with The National Archives

What we found

Technology moves faster than data projects

Work with partners to solve shared problems

A model with potential for wider application

Building for the future: Making change simple on GOV.UK Pay

We’re doing the hard work to make it simple

Building for the future

Other improvements we're making at the moment

Stay informed

Why Most Startups Struggle to Scale

The Hidden Cost of Operational Shortcuts

How to Build the Right Digital Foundations from Day One

Technology Should Enable, Not Complicate

How Koplio Helps

Final Thoughts

Putting customers first: DVLA and CustomerFirst partnering to improve Drivers Medical service

What is Drivers Medical and why it matters

Why DVLA and CustomerFirst are working together

A joined-up approach to improvement

What happens next

Shaping the future Fast Stream: digital and cyber security schemes

The Digital and Cyber Security Fast Stream schemes

Life at Government Digital Service (GDS)

Who can apply to the Fast Stream?

Why choose the Digital or Cyber Security Fast Stream?

Answers in seconds, 24/7: GOV.UK Chat launches in the GOV.UK app

Why we’re launching GOV.UK Chat now

What people are using GOV.UK Chat for

Built with safety, trust and transparency at its core

What’s next

Testing a different way to improve complex public services

Seeing and solving the whole service

Trying a different approach

What NewCo makes possible

Learning in the open

GOV.UK One Login for HMRC: how we made it happen and what comes next

How we got here: building a shared solution

Go live: the moment the button was pressed

The next phase: monitoring and iterating

What’s next

A new business analysis course is now live on the Government Digital and Data Hub

Why learn about business analysis?

Find more learning o pportunities on the Government Digital and Data Hub

Stay connected

GDS Local: building the foundations for digital collaboration

Co-creating a strategic vision for technology

Opening up the GDS offer

Bringing people together

Designing a Configurable Platform Architecture

Designing a Configurable Insurance Platform Architecture: A Case Study

Keeping Business Logic Behind the API – Protecting IP and Agility

Decoupling Partner Configuration from Customer Flows

Security and Ownership as First-Class Concerns

Hiding Complexity: Integrations Behind Stable APIs

AI as a Bounded Feature, Not a Free-for-All Chatbot

Architectural Outcomes and Lessons

Join AI Engineering Lab and help shape the future of digital government

What is AI Engineering Lab?

What’s happening in early 2026

Get involved

Innovation in action: highlights from the Local Government Innovation Hackathon

Collaboration at its best

What we achieved

Winning team

Thank you

Driving innovation through the Civil Service AI & Data Challenge – enter now

Transforming ideas into national impact

Career-defining opportunities

What you need to know

Ready to innovate?

Data maturity: the foundation for AI ready public sector data

Putting customers first:  DVLA and  CustomerFirst  partnering to improve Drivers Medical service