Project snapshot
Throughout our long-term collaboration with Travis Perkins plc, we have always admired the Group’s readiness to adopt advanced technologies that could improve productivity and enhance customer experience. In exploring transformative approaches to software development for the business, we suggested incorporating artificial intelligence into the process of creating and maintaining digital solutions.
Together, we investigated the opportunities of AI-assisted software development as a step towards digital transformation and improving code quality and team performance.
Travis Perkins is the UK’s largest distributor of building materials with a history going back more than 200 years. The business has grown to over 1,500 branches distributed across the country and more than 19,000 employees. Practicing a customer-first approach, the company maintains multiple digital products that improve various production processes and customer interactions.
Business challenge
Recognizing the steady advance of generative AI tools and their potential to transform processes in the IT domain, we committed to exploring the possibilities of artificial intelligence for software developers. At the same time, we needed to evaluate efficiency, security, and regulatory compliance to assess the value that AI-powered development can generate.
With our client’s consent, we introduced GitHub Copilot, an AI-driven tool to the standard development process that offers code completion functionality and chatbot-based suggestions. With this approach, we aimed to gather performance data and user feedback and test the effectiveness of AI-assisted development.
Solution
To test the benefits of AI-assisted development, our team used GitHub Copilot to get suggestions for methods and conditions, generate simple code, and check code for errors. Our development teams evaluated the performance of GitHub Copilot integrated with their usual IntelliJ IDEA and Visual Studio Code integrated development environments.
Our teams used GitHub Copilot as an inline code suggestion and autocompletion tool for both the project’s main programming languages (Java and JavaScript) and languages used for auxiliary tasks (Python and Objective-C), supporting programmers in languages that are not their primary field of expertise. Code suggestion and autocompletion was used in various development activities:
- Refactoring code
- Fixing bugs
- Handling support tasks
In addition, the development teams used the AI chatbot code solution adviser to quickly look up recommendations on proper use of AI libraries and framework techniques for specific use cases within the development process.
Business outcomes
During the six months of monitoring our team’s AI-assisted performance by a number of metrics, we noted an improvement in velocity resulting from the acceleration of multiple routine development processes:
- Faster implementation in a Copilot-enhanced environment
- Code syntax auto-generation and suggestions
- Faster lookup of code best practices with the Copilot chatbot
- Code style improvement suggestions
- Automated documentation generation
- Overall improvements to coding routine efficiency
This project allowed us to validate the advantages of AI-powered code completion assistants in developing custom IT solutions. By supporting engineering teams in routine and repetitive tasks and verifying code for errors, artificial intelligence tools allow developers to dedicate their time to more complex problems. With an AI tool speeding up development by suggesting code and streamlining information lookup, developers were able to focus on finding optimal ways of meeting the client’s requirements.
With all the benefits of AI-assisted development, we still recommend using GitHub Copilot or other AI-based software development tools for simple coding tasks with result verification by developers. While accelerating code creation and helping to identify errors, GitHub Copilot may have issues with complex logic and sophisticated tests.
Travis Perkins plc was also satisfied with the results of AI-assisted development, noting the high quality of software delivered by Intellias teams and appreciating the increase in project speed.
10–15%
Higher team velocity
30%
Reduction in development lead time
100%
Team members noted productivity improvement