Comsysto Reply, the Reply Group company specialised in agile software development and cloud-native solutions, has partnered with Audi on the ongoing development of its “Used Car Platform”, the company’s central digital marketplace for B2B used car trading. A tailored multi-agent solution accelerates software development, reduces repetitive activities, and enables faster delivery of new features.

The platform represents a key digital sales channel for Audi’s used car sector. Since 2025, it has accounted for nearly 100 per cent of the car manufacturer's B2B business. Building on this operational relevance, Audi and Comsysto Reply have introduced an AI-based multi-agent system designed to accelerate the development of new software features and reduce repetitive tasks across the Software Development Lifecycle.

The new solution marks the latest step in a long-standing collaboration between Audi and Comsysto Reply. Since 2021, Comsysto Reply has supported the platform from the replacement of the previous legacy system with a modern AWS-based architecture through to its continuous enhancement as a scalable, business-critical digital marketplace.

At the core of the project is a tailored multi-agent system based on GitHub Copilot. The solution includes three specialised AI agents, each focused on a specific area of the development process: requirements analysis, code generation and quality assurance, including the automated creation of documentation. Working together, the agents assist development teams in transforming requirements into structured outputs that can be reviewed, validated, and integrated efficiently.

A key element of the solution is its dedicated knowledge base. Rather than relying on generic development patterns, the agents generate outputs based on the platform’s actual architectural decisions, API specifications, and security policies. This enables the system to produce results that are technically consistent with the existing platform landscape and aligned with Audi’s development and governance requirements.

The multi-agent system introduces a human-in-the-loop development model in which AI agents take over repetitive and time-consuming activities, while developers remain responsible for review, validation, and final approval. This allows the team to focus more strongly on the evolution of value-adding capabilities for the platform, including intelligent pricing features designed to respond more precisely to market dynamics in the B2B used car sector.

Stefan Winkler, Dev Lead at Audi, commented: “The introduction of AI agents has fundamentally changed both the speed and the way our team works. Our developers can now focus on the tasks that create the greatest value: developing new features that deliver real benefits for our dealers.”