Neural CAD
Autodesk's generative AI foundation models that replace traditional parametric CAD engines—enabling text-to-geometry generation and intelligent design transitions for buildings and products.
Neural CAD in practice
Floor Plan, annotated
An architectural floor plan with walls, doors, room tags, and the column grid every other discipline draws on.
- 1Exterior wall with window openings
- 2Door, swing, and door tag
- 3Room name and number
- 4Partition type tag
Definition
Neural CAD is Autodesk's category name for generative AI foundation models that fundamentally reimagine how CAD geometry is created. Announced at Autodesk University 2025 and planned for commercial availability in 2026, Neural CAD represents 'an entirely new, machine learning approach to generating CAD objects in contrast to the classical parametric CAD engines employed for over 40 years.' Two applications are emerging: Neural CAD for Geometry (Fusion) generates editable precision CAD geometry from text prompts and spatial constraints, handling both organic curves and flat surfaces with defined edges; Neural CAD for Buildings (Forma) enables architects to transition between early conceptual massing models and detailed building layouts, with changes to one architectural system immediately reflected in others. Autodesk's AI research foundation spans 15+ years and nearly 100 peer-reviewed papers.
Examples
Typing 'generate a cantilevered staircase with open risers for this 3.5m x 4m opening' and receiving editable CAD geometry
Using Neural CAD for Buildings to transition a massing concept into a detailed floor plan with programmatic room assignments
Neural CAD generating parametric modeling commands needed to recreate a complex organic facade form in Fusion
Nomic Use Cases
See how Nomic applies this in production AEC workflows:
Compatible Platforms
Nomic integrates with these platforms so you can use neural cad across your existing project data:
Frequently Asked Questions
Neural CAD is Autodesk's category name for generative AI foundation models that fundamentally reimagine how CAD geometry is created. Announced at Autodesk University 2025 and planned for commercial availability in 2026, Neural CAD represents 'an entirely new, machine learning approach to generating CAD objects in contrast to the classical parametric CAD engines employed for over 40 years.' Two applications are emerging: Neural CAD for Geometry (Fusion) generates editable precision CAD geometry from text prompts and spatial constraints, handling both organic curves and flat surfaces with defined edges; Neural CAD for Buildings (Forma) enables architects to transition between early conceptual massing models and detailed building layouts, with changes to one architectural system immediately reflected in others. Autodesk's AI research foundation spans 15+ years and nearly 100 peer-reviewed papers.
Typing 'generate a cantilevered staircase with open risers for this 3.5m x 4m opening' and receiving editable CAD geometry. Using Neural CAD for Buildings to transition a massing concept into a detailed floor plan with programmatic room assignments. Neural CAD generating parametric modeling commands needed to recreate a complex organic facade form in Fusion.
Project Research: Instantly access all project-critical information from a single search interface.
More Technology Terms
View allSee Neural CAD in action
Nomic is purpose-built AI for architecture, engineering, and construction. Connect your project data and start getting answers in minutes.