Hi everyone,
We’ve released a little tutorial series on agentic coding for design in Rhino that illustrates what is possible with current generation coding models. It shows a few techniques we’ve been exploring recently like using images and screenshots as references, switching between cheap and expensive models, refining ideas with conversation before actually writing code and a few others. A link to the series is here: https://www.youtube.com/playlist?list=PLgT771y9VleBygJ7PwMXBl4226pgSq75w
The tutorial series covers:
- writing scripts for exploring simple massing studies / optioneering
- introducing design constraints and iterating until these constraints are met
- creating simple user interfaces for exploring script parameters
- building tools for analysing geometry
- generating scripts for formal exploration from reference images
- refining these scripts through the socratic method - essentially getting agents to ask you questions about your design goals
- introducing really quite complex fabrication constraints (like uniform panel spacing and alignment over arbitrary double curved meshes)
We’re doing all of this in our Runchat plugin for Rhino. If you try it out and have any feedback let us know.


