Skip to main content

This is a new service – your feedback (opens in a new tab) will help us to improve it

Use AI tools in Home Office Digital product teams

Last updated: 18 May 2026
Relates to (tags): Artifical intelligence (AI), Ways of working

When you design and build products in Home Office Digital, you should use generative AI to save time, be more efficient and test your code for bugs and errors.


Rationale

The main ways Home Office Digital product teams are already using generative AI as coding assistants are to:

  • automatically complete code they write manually
  • write small pieces of code for specific functions, manually reviewing the code that generative AI helps to create
  • work in the style of a product owner, letting generative AI take the developer role and work as an agent to create software
  • generate summaries of GitHub pull requests
  • create small pieces of content, such as README files

You should experiment with the most effective ways you can use AI in your work.


Applications and Implications

Use GitHub Copilot in your code editor

To use GitHub Copilot for Home Office Digital projects, you need:

  • a GitHub account that is suitable to use in the Home Office,
  • to apply for a Github Copilot License

Find out how to do these things by reading Home Office Digital documentation.

Ways you can use GitHub Copilot

Engineers across Home Office Digital and the wider Home Office are using GitHub Copilot in many ways, from simple, everyday additions to the way they already work, to experimental projects.

At the moment, most Home Office Digital engineers use GitHub Copilot in their code editor.

Some use cases include using GitHub Copilot to:

  • find and fix errors and vulnerabilities in existing code
  • trial agentic coding, where an engineer or team gives an AI assistant a code specification, and it is allowed to code on its own
  • use documentation as prompts; build Copilot prompts directly from documentation, to help service design and prompts work closely together
  • write simple prompts to convert JSON files into clear data lists
  • convert and restructure documentation into interactive websites, complete with AI-generated chapters and sections, as well as suggestions for related topics for a user to read
  • use agents to read documentation and train itself on the information it finds, which helps to validate the way users interact with documentation
  • use Copilot to review code and feed its review back to an engineer who supervises
  • work with a business analyst (BA) to create and iterate wireframes together, rather than an engineer having to send changes to a wireframe back to a business analyst to work on alone
  • test Jira cards to find defects and help create actions to fix problems, as well as checking acceptance criteria across all stories in an epic to see how they relate to each other
  • write a simple agentic AI tool to connect to Jira and summarise tickets

To find out more suggestions of ways to code with GitHub Copilot, you can:

Content version permalink (GitHub) (opens in a new tab)