AI-ACCELERATED MES TESTING: 100% more automation output, validation up to 7 weeks earlier

Our client 

Our client is a global manufacturer of advanced composite materials, supporting critical aerospace programmes through an international network of factories. 

As part of a major technology transformation, the business was moving away from its legacy ERP platform and modernising its Manufacturing Execution System (MES). Having previously helped the client mitigate the risks associated with this complex programme, Roq was asked to accelerate the delivery of automated UI testing for the MES application. 

The client challenge 

Developing an end-to-end feature within the MES could span several sprints, with some features taking up to eight weeks before they were ready for automation. As a result, automated testing often began late in the development lifecycle, delaying feedback and increasing the likelihood that defects would become more costly and time-consuming to resolve. 

The project team comprised three automation testers and nine manual testers. Although the manual testers brought valuable business and testing knowledge, they did not initially have the automation experience needed to contribute code confidently. 

Creating and maintaining automation code was also a largely manual process. Repetitive coding, refactoring duplicated scripts and checking compliance with project standards all consumed valuable time. As automation output increased, the growing demand on experienced team members to review code risked creating a further bottleneck. 

The challenge was to expand the team’s automation capability, accelerate testing and maintain code quality without introducing delivery risk or making costly, disruptive changes to the client’s established ways of working. 

What we delivered 

Roq integrated GitHub Copilot into the team’s Visual Studio development workflow, using AI to accelerate code creation, improve consistency and reduce repetitive manual effort. 

Copilot supported the team throughout the automation development process, enabling them to work more efficiently while improving the consistency, quality and reusability of the code they produced. 

We combined this AI support with a structured upskilling programme for the client’s nine manual testers, enabling them to become hybrid testers capable of contributing to automation. This allowed the team to adopt a component-based approach, creating UI validation scripts as soon as individual pages were delivered rather than waiting for an entire end-to-end feature to be completed. 

AI was also used to support code reviews and assess whether new scripts adhered to the project’s coding standards. This helped experienced automation engineers manage the increased volume of contributions while giving newer team members relevant, contextual guidance as they worked. 

Copilot was used as an accelerator, not an autonomous decision-maker. Every AI-assisted change was manually reviewed and tested before acceptance, ensuring that human expertise and quality controls remained central to delivery. 

Client impact 

The AI-enabled approach allowed the client to begin receiving automated tests from the first week of development, rather than waiting up to eight weeks for a complete feature. By bringing testing forward, the team could identify issues when they were quicker, easier and less costly to resolve, reducing the risk of defects reaching production and avoiding more substantial rework later. 

The client achieved measurable improvements in capacity and delivery: 

• Automation capability expanded from three specialist automation testers to 12 team members able to contribute code. 

• Automated test output increased by 100%. 

• Time to begin UI validation was reduced by up to 88%, from eight weeks to one week. 

• Faster code completion and refactoring reduced time spent on repetitive development tasks. 

• Cleaner, more reusable code improved the quality and maintainability of the automation solution. 

• AI-supported reviews helped the team absorb increased output without allowing code quality or project standards to slip. 

Crucially, Copilot gave Roq greater confidence in enabling nine less experienced automation contributors simultaneously. Its context-aware suggestions reduced the risks associated with this rapid expansion, while mandatory human review ensured that accuracy and safety were never delegated to AI. 

The client gained more than a short-term increase in test output. By upskilling the wider team and embedding AI-assisted automation into the delivery process, Roq left the client with stronger internal capability and a more scalable approach to Quality Engineering. 

This provided a pragmatic, cost-effective use of AI, materially improving speed, capacity and code quality without requiring a specialised AI model, significant additional investment or disruptive changes to the client’s established working practices. 

GET IN TOUCH

If you'd like to understand how Roq can combine AI, automation and Quality Engineering expertise to reduce delivery risk and help you move faster with confidence, reach out to us via ask@roq.co.uk or click here.