beeConnect
Graph-smart, LLM-powered autocomplete for Grasshopper — predicting components, micro-chains and sub-graph connections from your own script patterns.
- Edwin Hernandez Gomez, Eesha Jain, Noah Rosenberg, Ramy Maher, Rick van Dijk, Daniel Khalighinejad
Built in a 24-hour sprint at AECtech NYC 2025 with a team spanning five countries.
beeConnect is a prototype autocomplete engine for Grasshopper. It combines QuickConnections and Graph-Hop with custom graph logic so the canvas can anticipate what you build next — while keeping inference local and private.
Capabilities
- Predicts next Grasshopper components and micro-chains
- Auto-connects sub-graphs from patterns in the user’s own scripts
- Uses LLMs to tag graph motifs with human-level meaning and context
- Keeps model intelligence on-device — no cloud dependency for the graph brain
A hackathon prototype, but a concrete glimpse of graph-native, context-aware parametric AI: less blank-canvas friction, more continuity with how designers already think in graphs.
Organised by CORE studio at Thornton Tomasetti / Robert Otani.