Government coders save 28 days a year with AI, targeting £45bn in public sector savings

Government coders save 28 working days a year with AI assistants

Recent trials across central government show developers saving almost an hour a day with AI coding assistants. That equates to 28 working days a year reclaimed for delivery, aligning with the Plan for Change and the wider push to find £45 billion in efficiencies across the public sector.

Over 1,000 engineers across 50 departments used AI tools to write and review code, helping teams ship more technology including Whitehall's Humphrey AI assistant and healthcare tools that improve frontline services.

What the trial found

• Developers saved close to one hour per day using AI assistants for first drafts and code reviews.
• Only 15% of AI-generated code was used without edits, showing engineers are actively reviewing outputs.
• 72% said the tools offered good value for their organisation.
• 58% would prefer not to return to working without AI assistance.
• 65% reported completing tasks faster; 56% solved problems more efficiently.

Why this matters for departments

Time saved on routine coding is being reinvested into priority delivery. Teams are moving faster on critical services while maintaining standards, supporting the government's Plan for Change and the target of up to £45 billion in efficiencies.

This isn't about replacing engineers. It's about reducing repetitive work so experts can focus on architecture, security, and quality - the work that moves public services forward.

Safety and accountability

The trial followed AI Insights guidance for AI Coding Assistants: developers must understand and take responsibility for any code produced with assistance, as if they had written it themselves. Reviews remain standard practice, and outputs are edited where needed.

Tools in scope

Departments trialled Microsoft's GitHub Copilot and Google's Gemini Code Assist. For product details, see GitHub Copilot.

How to adopt in your team

• Start small: run a 6-8 week pilot with clear success metrics (time saved, PR throughput, defect rates).
• Set guardrails: require human review, document responsible-use practices, and log AI-assisted commits.
• Focus on workflows: use assistants for first drafts, tests, boilerplate, and refactoring; keep humans on design and security decisions.
• Invest in capability: provide short, practical training on prompts, code review habits, and unit testing. Consider role-based upskilling via AI Certification for Coding.
• Measure and iterate: compare pilot metrics to baseline, gather developer feedback weekly, and tune policies.
• Plan procurement and access: manage licences centrally, set default settings, and integrate with existing CI/CD and security tools.

What leaders and partners said

Technology Minister Kanishka Narayan said: "For too long, essential public services have been slow to use new technology - we have a lot of catching up to do. These results show that our engineers are hungry to use AI to get that work done more quickly, and know how to use it safely. This is exactly how I want us to use AI and other technology to make sure we are delivering the standard of public services people expect - both in terms of accuracy and efficiency. With a £45 billion jackpot at stake, it's not an opportunity we can pass up, as it can help cut backlogs and save money."

Tara Brady, President of Google Cloud EMEA, said: "We are thrilled to see the positive impact Gemini Code Assist has had on government developers, enabling them to accelerate innovation and deliver better public services more efficiently."

Notes

• The Department for Science, Innovation and Technology ran the trial from November 2024 to February 2025, providing training, support and guidance on AI Coding Assistants (AICAs).
• Guidance made it clear that developers must understand and own any code produced with AI assistance.
• More than 1,250 licences were redeemed across public sector organisations: 1,100 for GitHub Copilot and 173 for Gemini Code Assist.
• Findings will inform decisions on wider AI adoption across government.

Media enquiries

DSIT media enquiries
Email: press@dsit.gov.uk
Phone: 020 7215 3000
Hours: Monday to Friday, 8:30am to 6pm