Context-Based and Event-Driven Multi-Agent Architectures for Large Language Model Assistants in Engineering Environments

This paper presents advancements in the integration of large language models (LLMs) as AI assistants within complex engineering environments. Building on prior research, the study introduces a context-sensitive and event-driven multi-agent architecture that enables LLM-based assistants to respond dynamically to user actions, system states, and changes across heterogeneous engineering tools and processes. The architecture captures information about active editors, workflow configurations, tool interactions, and ongoing tasks, allowing the assistant to adapt its reasoning to the current context and to delegate subtasks to specialized agents through a unified interface. The resulting system supports engineers in activities such as workflow generation and analysis, interaction with internal data structures, and retrieval of domain-specific information. Several application scenarios demonstrate how the architecture can improve usability and efficiency by providing targeted, context-aware assistance.