Preparing for Agentic AI, a completely new future of government

What if every government agency could save thousands of hours every month—time that could be reinvested in supporting vulnerable citizens and improving lives?

Agentic AI is not just the next step in automation, it's a seismic shift in how governments can achieve operational excellence and build citizen trust. Industry is moving in that direction rapidly and Gartner has predicted that by 2028, 1/3 of interactions with generative AI will use autonomous agents for task completion. The UK Government must move quickly too.

In the public sector, the East and North Hertfordshire NHS Trust is estimated to have saved over 3,300 hours of HR-related work using conversational agents. That’s a vital breakthrough amidst resource pressures. Imagine scaling such successes government-wide to achieve operational efficiency and even excellence. he agent at New Jersey’s Department of Community Affairs has saved 800 hours of work each month in relation to its Utility Debt Relief Arrearage Forgiveness Program. Water Corporation has cut development efforts and associated costs by 30% through use of a code agent.

At IBM, treating ourselves as “client zero”, we’ve seen savings with agents across procurement, supply chain, finance, legal and more, resulting in a productivity improvement of $3.5 billion over the past two years. HR is our most famous use case, with Jon Lester – VP of HR Technology – stating “You can train it do certain tasks within one process, but you can easily have it apply those same skills to other processes. It blows chatbots out of the water. It really is changing our understanding of the future of work.”

Agentic AI offers the promise of more: systems that are characterised by orchestration where one, or more, AI agents are involved in complex problem solving. Whilst AI agents are singular models with native tools that can autonomously plan and execute tasks, an agentic system includes orchestration that enables integration of an agent with other agents, models, tools and data sources.

As the rest of the world veers in this direction, we are seeing a future where today’s front-end systems will become back-end systems. There are at least three key patterns with big potential for the public sector:

However, according to a recent poll of Chief Information Security Officers, 1 in 5 UK companies experienced data leakage because of employees using gen AI. 38% of workers admitted to sharing sensitive work information with AI tools without their employers' permission.

With so much focus on productivity and a target across all departments in the recently announced Spending Review 2025 of 5% in savings and efficiencies by 2028-2029, we must ask:

What must the UK Government do to safely take advantage of Agentic AI?

Firstly, the business case for value being returned to government is still critical to success. A new shiny technology – or its latest evolution – does not negate the need for this! Furthermore, individual productivity alone will not return the savings that the UK so desperately needs; government must prioritise activity at the enterprise level.

Shadow AI is a risk when knowledge workers aren’t given the tools they need to maximise their productivity. AI assistants and agents are already being trialled and deployed in some departments and agencies. For agentic AI, teams must now be equipped with multi-agent orchestration offering an open, hybrid approach that works across apps, vendors and clouds. This will ensure existing successes can be integrated and evolved as government matures in agentic AI.

There are three attributes to look for when introducing AI agents to drive productivity:

Agentic AI brings new risks and challenges such as hallucinations about tool choice, and goal misalignment, and amplifies existing risks and challenges, such as unexplainable actions and compliance. Previous governance frameworks focused on models will be increasingly insufficient. Instead, government should begin to govern AI systems as integrated systems that interact with various components, processes and humans too.

To achieve real gains, UK Government organisations must build capability to manage and govern 100s of autonomous agents.

Four fundamental organisational capabilities, supported by technology, are necessary to build trust at scale:

The Institute for Business Value also recommends governments think like a start-up: be open and willing to break with the past. It recommends: “lean into what you want your organisation to look like in three years—even if it seems impossible today. Take a product development approach to transformation, encouraging teams to quickly adopt new strategies, measure their success, and then iterate based on what they’ve learned, to avoid executing on outdated long-term plans.”

The opportunity for government and its 5 missions is huge.

Agentic AI has the potential to revolutionise the way governments operate with huge savings that can be ploughed into addressing the 5 missions, improve the lives of citizens, putting more police on the street, addressing NHS waiting lists and more. We have the opportunity to work together to deliver it safely and responsibly.

For a deeper dive, read AI agents: Opportunities, risks, and mitigations, a deep-dive into the unique risks posed by AI agents and potential mitigations, written by the IBM AI Ethics Board. Why not try an agentic AI bootcamp designed for engineers and for line of business participants to learn more?

Central Government Programme activities

The techUK Central Government Programme provides a forum for government to engage with tech suppliers. We advocate for the govtech sector, evangelise tech as a solution to public sector challenges, facilitate market engagement, and help make the public sector an easier market to operate in. Visit the programme page here.