How Artificial Intelligence is Transforming 3D Modeling

Artificial intelligence is already part of how many engineers and designers work with 3D
models today. In practice, this usually means using AI to automate repetitive steps,
explore design options faster, or reduce manual rework in complex models.

Generative design, powered by AI, changes how people approach everything from early
concepts to production-ready designs. Instead of manually iterating on a single idea, AI
systems can generate and evaluate large numbers of design variations based on
predefined goals and constraints.

For CAD experts, this often means spending less time on routine modeling tasks and more
time on design decisions. AI-powered tools in platforms like Octave BricsCAD are mainly
used to support this shift. Their aim isn’t to replace designers, but to reduce the amount of
manual cleanup, classification, and trial-and-error in everyday workflows.

Why traditional CAD methods need AI

Traditional CAD software already brought major efficiency gains compared to manual
drafting. But most workflows are still built around linear processes and a lot of manual
input.

In a typical setup, designers move step by step from concept to prototype, often revisiting
earlier stages when errors appear or requirements change. This creates three common
problems.

1. Linear and time-consuming processes
   Each stage depends on the previous one, so changes propagate slowly through the model.
2. Human error
   Manual input at every step increases the risk of inconsistencies, missing constraints, or misclassified geometry.
3. Limited experimentation
   Because prototyping and testing are expensive, many design ideas never get explored.

When issues are discovered late in the process—during user testing, for example—the
cost of redesign can be significant. This naturally discourages designers from trying more
experimental or high-risk ideas.

AI is increasingly used to make these workflows more flexible. According to the 2025 Avnet
Insights survey, 56% of engineers globally have already incorporated AI into product
design, marking a 33% increase from the previous year. AI 3D design, for example, provides
a transformative solution for users.

In practice, this usually means using AI to:

• Generate multiple design variants in parallel
• Check constraints automatically
• Reduce repetitive corrections

The result is not a completely different workflow, but a less rigid one in which designers
can explore more options without paying the full cost of failed prototypes.

Generative 3D modeling with AI

At the core of most AI-driven CAD workflows is generative 3D modeling. Instead of
manually building and testing a design, the user defines a set of inputs, such as materials,
manufacturing methods, cost limits, and performance criteria. The AI system then
explores the solution space and generates a large set of possible designs that meet those
constraints.

This automates a large part of the optimization process. Designers are still responsible for
choosing and refining results, but they no longer need to manually test every variation. In
BricsCAD, this kind of workflow is supported through integrations like the Grasshopper
connection. This allows users to create parametric models and generative algorithms that
would be difficult to manage using standard manual modeling alone.

For example, architects can use generative methods to design facades that balance
daylight, thermal performance, and structural constraints without manually adjusting each
parameter.

In the real world...

Foster + Partners applied similar computational design techniques when developing the complex fin louver geometry for the Mobility Pavilion at Expo 2020 Dubai (Alif). The façade consists of thousands of unique elements, each adjusted to balance daylight, solar shading, and structural constraints.

Instead of manually modeling each variation, parametric and algorithmic design tools were used to generate and refine the geometry based on performance criteria. This approach allowed the design team to evaluate multiple configurations efficiently and manage a level of geometric complexity that would be difficult to achieve with traditional modeling methods.

AI can also optimize for commercial parameters, not just technical ones. Cost, material
usage, and manufacturing constraints can all be included in the optimization logic.
BricsCAD automates these feedback loops and helps shorten the design-to-production
cycle and reduces the amount of late-stage redesign.

AI and 3D printing: A perfect match

AI-driven design works especially well with additive manufacturing. When you create an
artificial intelligence 3D model, you can generate complex internal structures, organic
shapes, and lattice geometries that would be difficult or impossible to produce with
traditional methods like CNC machining.

3D printing can build these shapes layer by layer, turning AI-optimized designs into
physical parts without simplifying the geometry.

In practice, this combination allows for:

• Lightweighting: Using less material while maintaining strength
• Part consolidation: Combining multiple components into a single part
• Material efficiency: Reducing waste
• Performance optimization: Tuning geometry for specific loads or conditions

Prototyping also becomes faster. Instead of validating one design at a time, teams can test
several variations in parallel and quickly compare results.

In metal printing, AI-based quality control is already being used. Real-time melt-pool
monitoring systems can detect defects during the printing process with near-perfect
accuracy, reducing scrap and improving reliability.

From 2D to 3D: How AI is bridging the gap

Moving from 2D drawings to 3D models is still a bottleneck in many workflows. AI is
increasingly used to reduce this friction. One approach involves Generative Adversarial
Networks (GANs), which consist of two neural networks:

• A generator that creates new data
• A discriminator that evaluates whether the data looks realistic

Over time, the generator learns to produce increasingly accurate outputs.

Applied to CAD, GAN-based systems can learn to interpret 2D images or sketches and
generate corresponding 3D models. This means that, in some cases, a basic 2D sketch
can be turned into a rough 3D model automatically.

More recent diffusion-based systems, such as Text2CAD, show similar capabilities by
generating accurate 3D geometry from limited visual input. In BricsCAD, machine learning
is already used in smaller but practical ways, mainly to speed up repetitive workflows like
geometry classification and object recognition. While this does not directly convert 2D drawings into 3D models, it helps structure and interpret imported or existing geometry more efficiently. This reduces the manual effort required when transitioning from 2D-based inputs to structured 3D models, making the overall workflow less time-consuming.

The future of AI: Text-to-3D design

Text-to-design is one of the more experimental areas of artificial intelligence in 3D
modeling. The idea is simple: instead of modeling geometry directly, the user describes the
object in natural language. The AI system then generates a parametric model based on that
description. Early systems like Google’s DreamFusion demonstrated this concept in 2022.
Since then, newer models like the company’s video generation model, Veo, have become
more capable at understanding spatial relationships and engineering constraints.

In theory, a designer could type: "Create a planetary gearbox with a 6:1 ratio for 80 Nm of
torque" and receive a working parametric model. In practice, this is still limited for real
engineering work. But even today, text-based systems can be useful for:

• Generating early concepts
• Exploring alternative layouts
• Creating starting points for manual refinement

"BricsCAD is doing early exploration of AI-powered workflows where rule-based logic is applied to existing models and drawings to support standards, compliance, consistency and production documentation." - Piet Lelieur, Senior product owner, point cloud and productivity.

BricsCAD: Using AI in 3D design

In BricsCAD, AI is mainly used to reduce manual overhead in everyday CAD work. The goal
is not to replace modeling, but to automate tasks that users typically spend a lot of time
on.

Examples include:

• BIMIFY: Automatically classifies raw geometry into BIM elements such as walls, slabs, and columns
• Blockify: Detects repetitive geometry and converts it into reusable blocks
• Move Guided: Repositions entities while automatically adapting connected geometry

These tools don’t generate designs on their own, but they remove friction from common
workflows. That means less time cleaning up models and more time making design
decisions.