ArtisanCAD: An Industrial-Level CAD Agent with Expert-Grounded Knowledge Distillation

arXiv 2026 58 method, system

TLDR

ArtisanCAD is an industrial CAD agent using expert knowledge distillation and CAD-IR to generate production-ready B-Rep models from text.

Reasoning

The paper introduces a novel CAD intermediate representation (CAD-IR) and expert knowledge distillation to address ambiguity in text-to-CAD. Strengths include a focus on industrial-grade parametric modeling and B-Rep execution, but the abstract lacks complete benchmark results and details on limitations.

Read-first score

Read-first score 58, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 87.

Recency 8%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=benchmark,metric,result

Topical relevance 42%

54.4

Uses existing LLM keyword relevance scores normalized to 0-100. AI for CAD,computer-aided design,neural CAD,generative CAD,parametric CAD,B-Rep,boundary representation,constructive solid geometry,CSG,sketch extrusion,CAD generation,CAD reconstruction,text-to-CAD,image-to-CAD,point cloud to CAD,CAD program

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=code

Field roles

FrontierMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 15.

Keyword Scores

computer-aided design

B-Rep

boundary representation

text-to-CAD

AI for CAD

parametric CAD

CAD generation

CAD program

generative CAD

neural CAD

constructive solid geometry

CSG

sketch extrusion

CAD reconstruction

image-to-CAD

point cloud to CAD

Deep Analysis

Innovations

CAD intermediate representation (CAD-IR): an executable procedural representation encoding parameters, ordered operations, MCP tool bindings, dependencies, generated entities, and verification rules
Skill-guided industrial CAD agent with expert-grounded knowledge distillation, distilling expert CATIA procedures into reusable parameterized skills
Retrieval of expert-derived skills, CAD-IR instantiation and revision, execution through a dedicated CATIA-MCP backend, and multi-view visual feedback for iterative refinement
Bridging ambiguous or intermediate-level textual prompts to full executable CAD operations via CAD-IR scaffold

Methodology

ArtisanCAD uses a CAD intermediate representation (CAD-IR) to distill expert CAD procedures into reusable skills and to scaffold vague prompts into executable operations. It retrieves expert-derived skills, instantiates and revises CAD-IR, executes through a CATIA-MCP backend, and applies multi-view visual feedback for iterative refinement, generating production-ready B-Rep models.

Key Results

On the Text2CAD benchmark, CAD-IR reduces mean Chamfer Distance from 14.83 to 9.88 for intermediate prompts. On four complex automotive components, expert CATIA recordings distilled into skills enable generation of editable CATIA-native B-Rep models for new variant requests.

Resources

Paper