LLM-Powered Workflow Optimization for Multidisciplinary Software Development: An Automotive Industry Case Study
Summary: arXiv:2603.21439v3 Announce Type: replace-cross
Abstract
Multidisciplinary Software Development (MSD) requires domain experts and developers to collaborate across incompatible formalisms and separate artifact sets. Today, even with AI coding assistants like GitHub Copilot, this process remains inefficient; individual coding tasks are semi-automated, but the workflow connecting domain knowledge to implementation is not. Developers and experts still lack a shared view, resulting in repeated coordination, clarification rounds, and error-prone handoffs.
Introduction
This article addresses the inefficiencies present in Multidisciplinary Software Development (MSD) by proposing a graph-based workflow optimization approach. The goal is to progressively replace manual coordination with LLM-powered services, thereby enabling incremental adoption without disrupting established practices. This innovative approach is particularly relevant in complex fields such as the automotive industry, where collaboration between domain experts and software developers is critical.
Case Study: Volvo Group
Our methodology was evaluated on spapi, a production in-vehicle API system at Volvo Group. This system involves:
- 192 endpoints
- 420 properties
- 776 CAN signals
- Six functional domains
Results
The implementation of the automated workflow yielded significant improvements:
- Achieved a 93.7% F1 score in automated testing.
- Reduced per-API development time from approximately 5 hours to under 7 minutes.
- Saved an estimated 979 engineering hours.
These results indicate that the integration of LLM-powered services into the workflow effectively streamlines the development process, enhancing overall efficiency and productivity.
Feedback from Participants
In production, the system received high satisfaction ratings from both domain experts and developers. All participants reported:
- Full satisfaction with communication efficiency.
- Improved collaboration and understanding of project requirements.
- A significant reduction in the need for repeated coordination and clarification rounds.
Conclusion
The case study conducted at Volvo Group demonstrates the potential of LLM-powered workflow optimization to transform Multidisciplinary Software Development practices. By addressing the coordination challenges that often plague such projects, our approach not only enhances efficiency but also fosters better communication among stakeholders. As industries continue to evolve and integrate advanced technologies, the adoption of similar AI-driven solutions is likely to become more prevalent, marking a significant shift in how multidisciplinary teams operate.
Future Work
Future research will focus on further refining the graph-based workflow optimization and exploring its applicability in other fields beyond the automotive industry. Additionally, we aim to investigate the long-term impacts of LLM-powered services on team dynamics and project outcomes.