Abstract
Abstract. Artificial Intelligence (AI) is rapidly transforming the Architecture, Engineering, Construction, and Operations (AECO) industry; however, inconsistent terminology and fragmented applications limit its practical adoption. This study addresses these challenges through two contributions: (1) a taxonomy of AI methods across the AECO lifecycle, distinguishing machine learning, deep learning, generative AI, and algorithmic design, and (2) an LLM-based Retrieval-Augmented Generation (RAG) framework for integrated facility management (FM) data access. The proposed system integrates BIM data (PostgreSQL), vector embeddings (Chroma), and IoT telemetry into a unified queryable environment. A hybrid retrieval strategy combining semantic search and structured queries enables accurate and explainable responses to natural-language inputs. A Streamlit-based interface supports user interaction and links responses to Autodesk Tandem 3D views for spatial context. Results demonstrate high retrieval accuracy and fast response times, highlighting the effectiveness of combining semantic retrieval with structured data operations. This work provides both a conceptual framework for AI classification in AECO and a practical system for intelligent facility management, contributing toward AI-enabled digital twins.
Showing the abstract — retrieve the full paper via the Exa API.