GCS Assist

Introduction

Assist is a secure, accessible generative AI tool built specifically for Government Communications. It helps government communicators work more efficiently and effectively — from drafting communications plans and brainstorming campaign ideas, to reviewing content and sense-checking work against government standards.

This page is for anyone who wants to understand how government communications is using bespoke AI technology, whether you work in the profession, elsewhere in the Civil Service, or outside of government entirely.

How Assist began

We built Assist as part of a deliberate strategy to bring AI into government communications responsibly and at scale. Our starting point was understanding what makes organisations successful at innovation – and using that to shape our approach: encourage, identify, test, and scale.

Ethics and responsible use were central from the start. So was transparency. We built the first prototype in-house in just six weeks, began testing with users in late 2023, and developed a bespoke mandatory training course to ensure everyone – regardless of AI experience – could use the tool effectively and safely.

How we built and developed Assist

Assist is the first cross-government generative AI tool to achieve ‘live’ service status, meaning it has been independently assessed and confirmed to meet all government security, accessibility, and service standards. It achieved 14 out of 14 ‘greens’ in its Government Digital Service (GDS) assessment – a rigorous and transparent process.

We have also open-sourced Assist’s underlying code, so that other departments and professions can learn from and build on our approach. The resources below offer more detail on how and why we developed Assist, and the guidance we’ve published to help others do the same.

• Introducing Assist – A brief overview of how and why we built Assist, and its positive impact across the profession.
• Guidance on a human-centred approach to scaling and de-risking AI tools (GOV.UK) – Two behavioural science-informed guides to help organisations scale AI solutions whilst managing unintended consequences, based on our experience rolling out Assist across hundreds of communications teams.
  • The People Factor: A human-centred approach to scaling AI tools – A three-stage framework (Adopt, Sustain, Optimise) to fast-track and de-risk AI adoption.
  • The Mitigating ‘Hidden’ AI Risks Toolkit – A practical guide to identifying and managing unintended consequences of AI adoption.
• Assist passes Service Assessment (GOV.UK) – The full outcome of Assist’s third and final GDS assessment.
• Assist – Algorithmic Transparency Recording Standard (GOV.UK) – A technical overview of Assist’s infrastructure, underlying model, and risk mitigations.
• Assist’s code open sourced (GitHub) – The underlying code, available for others to view, learn from, and build upon.

Assist’s impact

Assist has transformed how government communicators work. It is now embedded across Government Communications, used daily by communicators across hundreds of departments and organisations to handle some of the most time-consuming parts of the job – freeing up time for the higher-value, creative, and collaborative work that really matters.

Assist has been recognised as the Winner of the 2025 National AI Awards – Public Sector Category, and won the Creative Solution Award at the Cabinet Office 2025 Awards. It is also features as a case study in the cross-government AI Knowledge Hub.

You can see Assist in action through our growing collection of case studies:

• Using Assist to draft crisis communications – How Assist supported government communicators and departments across government to create plans for potential national crises
• Using Assist to support a return to work – How Assist helped a Ministry of Defence communicator returning from maternity leave get back up to speed

The challenge

The Government Communications profession identified lack of time as a key barrier to innovation in the 2023 Horizon Review into Responsible Innovation. Research highlighted the need for support with brainstorming, drafting, and reviewing content. Improving efficiency in these tasks would free up time for collaboration and higher-impact work.