EUI Agent
Classification: Sustainable Design Research / Computational Design
Research Topic: AI-Assisted Building Energy Prediction and Design Optimization
Location: Seattle Climate Zone 4C
Zoning: DMC 240/290-440
Research Type: Collaborative Research
Date: Spring 2026
WHY THIS MATTERS?
Designers need faster and more accessible energy feedback during early-stage design.
Buildings are responsible for 40% of global energy consumption
Seattle-based firms are actively engaged
1,300+ firms joined AIA 2030, yet a significant performance gap remains.
WHY CURRENT METHODS FALL SHORT?
Existing studies focus on simulation, prediction, or interaction separately, but rarely integrate all three into a single workflow.
Physics-Based Simulation
Accurate
Too slow
ML Surrogate
Fast
Not physically grounded
AI Interface
Interactive
Not connected to simulation
"How can an EnergyPlus-generated surrogate machine learning model, integrated with an agentic AI workflow, support rapid and physically consistent energy prediction during early-stage architectural design?"
ANN
Data-driven prediction model
Learns annual EUI from 8 design variables
Soft physical regularization
Encourages physically plausible energy trend
Agentic AI
Natural-language design interaction
Converts design intent into model inputs and feedback
Start-to-End Methodology
Agentic AI Workflow
Model Demo