| ||||
| ||||
![]() Title:Orchestrating LLM-Powered Workflows for Autodesk Revit via Model Context Protocol: a Multi-Agent Framework for Intelligent BIM Automation Conference:ICCCBE2026 Tags:Building Information Modeling, Large Language Models, Model Context Protocol, Multi-Agent Systems and Retrieval-Augmented Generation Abstract: This paper presents an orchestration framework enabling intelligent, natural language interaction with Autodesk Revit through the Model Context Protocol (MCP). The system integrates retrieval-augmented generation (RAG), ReAct-style multi-agent reasoning, and multi-LLM support to transform natural language instructions into precise BIM modeling operations. By connecting to a Revit-MCP server, the framework executes complex modeling workflows while incorporating contextual best-practice documents and location-specific building standards. Evaluation across representative architectural tasks demonstrates substantial improvements: 75-80% reduction in modeling time, 90-95% accuracy in dimensional and property validation, and 90-95% standards compliance detection compared to 75-80% through manual review. Orchestrating LLM-Powered Workflows for Autodesk Revit via Model Context Protocol: a Multi-Agent Framework for Intelligent BIM Automation ![]() Orchestrating LLM-Powered Workflows for Autodesk Revit via Model Context Protocol: a Multi-Agent Framework for Intelligent BIM Automation | ||||
| Copyright © 2002 – 2026 EasyChair |
