GPT-6 Astra + Grasshopper Built This Floor Plan Generator (insane)

What happens when AI doesn’t just generate a floor plan but instead builds a reusable parametric system to generate hundreds of them? In this video, I use GPT-6 Astra, Rhino, Grasshopper and the Rhino MCP server to build a parametric apartment floor plan generator. Instead of asking AI to model every design from scratch, I explore a different approach: using AI to create deterministic, reusable computational design tools that architects can actually integrate into their workflows. The result is far from perfect. But after only a few hours of prompting, it already raises some interesting questions about the future of AI in architecture, computational design and Grasshopper. 🎓 Grasshopper Fundamentals for Architects I’m launching a new course teaching Grasshopper and systems thinking from the ground up through real architectural workflows. → Join the waitlist: https://www.ontologic.tech/courses In this video, I cover: • Using GPT-6 Astra with Rhino and Grasshopper • Connecting AI to Rhino through MCP • Building an AI-generated parametric floor plan generator • Generating apartment layouts from width, depth and program requirements • Why reusable systems can be more valuable than prompting AI from scratch • Deterministic vs probabilistic design workflows • Using Python inside Grasshopper • Structuring geometry with blocks and layers • Scaling Grasshopper workflows with Python and GitHub • How architecture firms can capture internal knowledge in software • Why AI could make Grasshopper — and systems thinking — even more relevant The bigger idea is that AI doesn’t necessarily replace parametric design. It can make it dramatically easier to build the parametric systems themselves. For architecture and engineering firms, that opens up the possibility of turning design knowledge, rules, standards and workflows into reusable software while still using AI where ambiguity and reasoning are actually useful. Chapters 00:00 AI-generated parametric floor plans 00:48 What we’re building 00:58 Grasshopper Fundamentals for Architects 01:28 The floor plan generator 02:14 Why combine AI and parametric design? 02:33 Building the configurator with GPT-6 Astra 03:14 Configuring apartment layouts 04:13 The Python system behind it 04:49 Scaling Grasshopper with GitHub 05:43 Why AI shouldn’t do everything 06:17 Deterministic vs probabilistic systems 07:37 Why reusable configurators matter 08:17 Connecting GPT Astra to Rhino with MCP 09:33 Generating a floor plan directly with AI 10:13 Why agent-only workflows don’t scale 10:47 Turning AI output into reusable Grasshopper systems 11:43 What this means for architecture firms