Gen AI augments Seagate’s factory-focused software development.

Seagate’s Global Factory IT team started an initiative to gain value in its manufacturing operations through applying AI. The team took the opportunity to focus on software development, using AI to help move past traditional processes and optimize Seagate’s manufacturing IT solutions in real time. They selected GitHub Copilot, an AI-powered tool developed by Microsoft and OpenAI, to boost efficiency, code quality, and productivity. The first of three project phases involved identifying suitable factory use cases and evaluating the business impact of the integration.

The Global Factory IT team sought a solution to save both costs and time in factory-based software development. They immediately identified several key technical requirements. First, the team required seamless integration of GitHub Copilot and gen AI tools with existing systems. Second, the team needed developer training to effectively use these tools. Third, sensitive code and proprietary information called for robust data security and privacy measures. Finally, the solution needed a system to measure performance metrics to assess key performance indicators. 

The team divided the project into three phases, including initial adoption, ramp-up, then full implementation with measured outcomes.

Seagate IT executives and software developers had multiple objectives for integrating AI tools into their processes. Immediate needs included creating real-time code suggestions, adapting to developer’s coding styles, and reducing cognitive load. The team also needed a tool that allowed for easy review, as developers were still encouraged to review the generated code to ensure it met their projects’ specific requirements and standards. 

Long-term goals include improving code security and reliability, aiming for a 50% reduction in rework. Another goal is to boost developer productivity and efficiency, targeting a 20-50% reduction in development time through reduced manual coding. Additionally, Seagate aims to continue modernizing legacy code, focusing on risky code migration (e.g., JavaScript to Angular/JS, Visual Basic to C#) while minimizing business risks. 

The team anticipated some obstacles to applying gen AI tools. Before the new initiative, Seagate’s Global Factory IT team used more traditional methods. Code generation was done manually, leading to inefficiencies and inconsistencies. Code quality was ensured through manual reviews, which depended heavily on the reviewer’s expertise and were not foolproof. Developers had to continuously learn new languages and frameworks, slowing down development speed and productivity. Maintaining legacy systems involved dealing with outdated code, which was resource-intensive and could lead to technical debt. In software development, technical debt is the implied cost of future reworking, as speed is often a priority over long-term design effort. These traditional methods, while somewhat effective, were often resource-intensive, time-consuming, and could not fully eliminate errors or inefficiencies. 

The new solution would need to meet additional team requirements. In document generation, they wanted to enhance developer efficiency and reduce support needs through templated documentation. They also required modernized applications to address technical debt and improve developer efficiency. Additionally, the solution would need to enhance data analysis by phasing out legacy decision-making applications, speeding up decision-making, and reducing business continuity risk.

The initial stages of the project demonstrated that an AI-based tool could assist Seagate developers in saving time and in other areas. Integrating GitHub Copilot into Seagate’s Global Factory IT software development workflow has shown promise in optimizing processes, delivering business value, and setting industry standards.  

Phase 1 involved the initial adoption of GitHub Copilot. This phase involved implementing GitHub Copilot for core developers and providing training. It also included measuring the reduction in development time for routine tasks. Phase 1 has already demonstrated early benefits.  

Phase 2 will involve a ramp-up of the project, with aims to improve overall development productivity by 20-40% and achieve a code/chat acceptance rate of over 30%. Phase 3 will include fully implementing GitHub Copilot and keeping track of key metrics for continuing success. This will include optimizing resource allocation for value-adding projects, aiming for greater output with the same or fewer resources.  

Seagate's multiphase plan to integrate generative AI capabilities into its factory-based software development.

Enlarge

Seagate's multiphase plan to integrate generative AI capabilities into its factory-based software development.

Early observations from Seagate IT executives and developers highlight several benefits since implementing GitHub Copilot. Efficiency has improved with automated code generation, speeding up task and project completion. Code quality has also been enhanced, with better reliability and security due to automatic vulnerability detection and quality assurance. Developer productivity and innovation have increased as developers focus more on strategic tasks and creative problem-solving. The initiative has also led to positive higher-level business outcomes, including cost savings, increased customer satisfaction, and a competitive edge due to improved efficiency.   
 
Phase 1 of the initiative has already garnered numerous signs of success, including in launch and adoption, and productivity and business impact. First, the rapid adoption of GitHub Copilot as the first “AI pair programmer” was crucial. Over 30 developers now use GitHub Copilot, streamlining coding workflows.   
 
In terms of productivity, Factory IT professionals utilized GitHub Copilot to generate more than 250,000 lines of code, with nearly 30% seamlessly integrated into their solutions. Many reported increased productivity due to the AI tool’s assistance. The business impact extended beyond individual developers to the entire IT ecosystem, with increased productivity translating into faster feature delivery and reduced time-to-market.  
 
In phases 2 and 3, Seagate’s Global Factory IT team plans to boost developer productivity and collaboration and identify quantifiable metrics and overall IT benefits. Phase 2 will focus on semi-automating unit testing, improving documentation efficiency, and reducing code development time. Crucial tasks like architecture design, code review, integration testing, and quality assurancewill still rely on senior developers. gen AI will handle routine coding tasks, allowing developers to focus on complex tasks, critical decisions, and strategic planning. In phase 3, the team will also evaluate the impact on overall product delivery time and market competitiveness.   
 
The team plans to automate repetitive tasks, minimize rework, enhance team collaboration, and facilitate knowledge transfer through shared code snippets. They will measure acceptance rates, lines suggested, and time saved, considering broader IT benefits like productivity gains and reduced defects.  
 
As advice to other organizations considering integrating gen AI solutions into code development, Chong Khee (CK) Tan, senior manager of Global Operations IT/Global Factory IT MES Solutions and Services, recommends evaluating use cases to identify where gen AI can have the most impact. He advises starting with a small project to assess effectiveness, and addressing ethical concerns like bias, privacy, and security. He also suggests upskilling teams and fostering a learning and iterative environment.

“Thoughtful planning and adaptation maximize gen AI’s benefits,” Tan advises.