Falcon AI
From a design idea to a Grasshopper definition.
Describe a design change. Our prototype reads your Grasshopper model and helps build the components and connections to explore it.
Working research prototype. Not publicly released.Related research and articles
- 01Canvas context
Components, relationships, parameters
- 02Proposed definition
Design request to native components
- 03Check and review
Definition validation + designer judgment
From a Brief to Parametric Logic
An illustrative facade study: break the design request into inputs, relationships, and geometry that can be inspected.
Study a facade with hexagonal cells whose openings vary with a set of exposure values.
- 01 Facade Surface Set the surface and cell size as explicit inputs.
- 02 Hexagonal Grid Lay out the cells over the study surface.
- 03 Exposure Values Provide analysis values separately; check their source and units.
- 04 Remap + Scale Map the values to an editable range of cell openings.
- 05 Panel Geometry Inspect the resulting panels and their relationships.
- "Which parts of a definition explain its design intent?" Canvas context and readable relationships
- "How should a proposed change be checked?" Definition validation and designer review
The Prototype Workflow
Read the Context
The current canvas supplies component, connection, slider, and group information alongside the design request.
Build the Definition
The prototype turns generated definition data into native Grasshopper components and connections.
Check and Refine
Validation checks the definition data. The designer still reviews the geometry, assumptions, and design decisions.
Follow the Research
Falcon is a working research prototype. FraktalMCP is the separate, publicly available open-source Grasshopper plugin.
$ Design intent → canvas context → definition → review About the Research
What is Falcon AI today?
A working prototype in Fraktal’s research programme. Its focus is using model context to create and inspect Grasshopper definitions.
Is Falcon the same as FraktalMCP?
No. FraktalMCP is the public open-source plugin for Rhino geometry and GhPython generation. Falcon explores native canvas definitions and their context.
Does a valid definition mean a valid design?
Definition checks concern components, parameters, and connections. Architectural judgment, physical assumptions, and the resulting geometry still need review.
What information reaches an AI provider?
The prototype can include prompts, canvas summaries, and image input. The information shared depends on the selected workflow and provider; client material needs an appropriate data-sharing decision.
How can I learn more?
Contact Fraktal with the Grasshopper workflow you are investigating. This page describes the research; it does not announce a public plugin release or a release date.
What should an assistant understand about your model?
Tell us about a Grasshopper workflow, a recurring constraint, or a definition that is difficult to review.