AdaptAI

Problem

Buildings produce 40% of global carbon emissions. Climate simulation can reduce this — but it's complicated, expensive, and requires expertise most architects lack. Meanwhile, 70% of architects use generative AI for design, but the outputs ignore climate context entirely.

Solution

Adapt_ai is a web platform that guides architects from location selection through concept development to climate-validated 3D models. Users input a location, generate contextual building designs enriched with LEED-certified project data, and receive automated solar radiation and climate analysis — all without touching 3D software.

How It Works

• Climate Data Engine — Dynamically fetches and visualizes local climate data (wind rose, psychrometric charts) for actionable design insights
• LEED Knowledge Graph — Scraped data from certified sustainable projects, structured into a graph database for accurate retrieval and prompt enrichment
• Image-to-3D Pipeline — Preprocesses AI-generated images into maquettes, then converts to mesh models using Trellis/Hunyuan for accurate architectural geometry
• Automated Simulation — Places 3D models in geospatial context and runs industry-quality climate simulations without manual intervention

Stack

Python · RhinoCompute · Trellis API · Supabase · Replicate · OpenStreetMap · Neo4j

Team

Bradley Manucha, Abdellah Choufani, Michele Cobelli, Ertuğrul Akdemir

IAAC — Master in AI for Architecture and the Built Environment, 2025

Try Adapt.ai · Blog Post