AILANTA
All research
GitHub · Hugging Face · Discovery evidence

Generative AI is moving from rendered output to editable assets

GitHub, Hugging Face, and research evidence measure whether AI systems produce native CAD, 3D, layered, and manufacturing-ready artifacts rather than flat media.

2026-08-0629 items4 source types3 validation posts
Main finding

Supported by the targeted cohort

29 projects and models show that generative design is extending beyond flat output. Editable or parametric artifacts appear in 13.8% of the cohort, native CAD formats in 72.4%, and scene-native workflows in 20.7%. Manufacturing checks remain limited at 6.9%, so editable production assets are emerging before dependable engineering validation.

01

Market snapshot

Comparable measurements from independent market surfaces.

29

Implementations and packages

Deduplicated and subject-filtered primary cohort.

4

Median repository stars

Calculated across repositories with at least one star.

0

Registry packages

Real npm and PyPI package records, not synthetic entries.

0

Latest-month npm downloads

Usage surface only; downloads are not equivalent to customers.

02

Search-match dynamics

Bars show monthly GitHub query matches plus captured Hugging Face models; loaded counts are the subject-filtered cohort used for feature analysis.

Jan 26275 loaded
Feb 26171 loaded
Mar 26353 loaded
Apr 26372 loaded
May 26318 loaded
Jun 26304 loaded
Jul 26136 loaded
Aug 2610 loaded
03

What exists inside the category

One item may contain more than one feature.

Editable project artifacts

4 · 13.8%

Native CAD formats

21 · 72.4%

Native 3D scenes

6 · 20.7%

Manufacturing validation

2 · 6.9%

Iterative control

9 · 31%

Flat or rendered output

8 · 27.6%

04

Representative projects

05

Cross-source validation

These publications are not part of the primary numeric cohort.