AI for CAD Tools
Connecting Claude to SolidWorks makes a good demo. See why leading engineering teams choose Leo AI, the agentic platform for physical products.
·
⏱
7 min read
Dr. Maor Farid
Maor Farid is the Co-Founder and CEO of Leo AI, the first AI platform purpose-built for mechanical engineers. He holds a PhD in Mechanical Engineering and completed postdoctoral research at MIT as a Fulbright fellow. A Forbes 30 Under 30 honoree and former AI researcher and Mechanical Engineer in an elite military intelligence, Maor leads Leo AI's mission to transform how engineering teams design better products faster.
BOTTOM LINE
The chatbot route has viral videos. Leo has paid enterprise deployments with engineering teams at Scania, NVIDIA, BMW, Bosch and the U.S. military, plus independent recognition: Gartner recognized Leo in its research on AI in design and engineering; G2 named it the #1 New Software globally and the #12 AI Tool worldwide for 2026, based on verified user reviews; and CB Insights placed it on the AI 100. One side has demos. The other has engineers who trust it with the work.
Short answer: connecting Claude or ChatGPT to SolidWorks gives you a chatbot that can click buttons in CAD. Leo AI is the first agentic platform for physical products. It takes engineering teams from requirements to a manufacturing-ready assembly. One makes a good demo. The other runs in production at Scania, NVIDIA, BMW, Bosch and the U.S. military.
A customer asked me last week: "Everyone is connecting Claude to SolidWorks now. What's the difference with Leo?"
First, it isn't everyone. The demos fill LinkedIn, and most come from hobbyists and solo creators. The best engineering teams aren't doing it, for two reasons. They won't run their IP through personal chatbot accounts and experimental plugins. And they know drawing faster in 3D doesn't ship the product sooner or prevent a single manufacturing mistake.
CAD is not engineering
When Bertrand Sicot ran SolidWorks, he measured how much time its most active users spent inside the CAD editor. It was under 8%.
The rest of the project is where time and money go: defining requirements, running calculations, choosing parts, checking standards, reviewing designs, and handing a product to manufacturing. That is where launches slip and expensive mistakes are made.
So a chatbot that sketches and extrudes faster is optimizing the smallest step. Your launch date doesn't move. Your error rate doesn't move. There's no business ROI. For more on this, see our guide to AI for mechanical engineering.
IN PRACTICE
One of our research engineers used to use ChatGPT. He verified answers in Leo, and now he's left ChatGPT because he can trust Leo more.
Team Lead, ZutaCore
The chatbot: a blind engineer
A SolidWorks MCP plugin lets a chatbot send commands to the CAD: sketch, extrude, fillet. What comes back is names and numbers as text. The chatbot can't see the geometry, and it knows nothing about your guidelines, past projects or the standards you design to. It's a blind engineer building from a text description. It builds what it's told, right or wrong.
The tools are early, too. The public SolidWorks plugins describe themselves as alpha or early access. Anthropic's own CAD connectors cover Autodesk Fusion and Blender, and launched as "an early step." We could not find a single published enterprise deployment.
That's fine for a weekend project or a quick concept model. Some call it vibe engineering. But nobody vibe-engineers a brake system, a flight-critical bracket or a 40-ton press. In automotive, aerospace and defense, and industrial machinery, a design mistake means scrapped tooling, recalls, or worse. Those designs need an engineer's rigor, and an AI built to that standard.
The chatbot (Claude or ChatGPT + CAD plugin) | The solution (Leo AI) | |
What it is | A general chatbot that can run CAD commands | The first agentic platform for physical products, built by mechanical engineers |
What it covers | CAD clicks, under 8% of an engineer's time | Requirements to a manufacturing-ready assembly |
What it knows | Internet text. Blind to geometry and your context | A Large Mechanical Model trained on 1B+ CAD assemblies, plus your guidelines, projects and standards |
Business impact | Faster clicks. Same launch dates, same mistakes | Mistakes caught before manufacturing, parts reused, know-how kept |
Your IP | Personal AI accounts, plugins in alpha or early access | SOC 2 Type II, isolated per customer, never trained on your data |
Track record | Viral demos, no published enterprise deployments | Paid deployments at Scania, NVIDIA, BMW, Bosch, U.S. military |
Every serious profession picks AI built for it
We've seen this movie before.
Lawyers had ChatGPT and Claude. They chose Harvey and Legora. Harvey is now used by 80 of the top 100 US law firms.
Software developers had the same chatbots. They chose Cursor, which passed $3B in annualized revenue.
Harvey, Legora and Cursor all run on frontier models, including Claude. Professionals don't choose them for the model. They choose them because they were built by people from the profession, around the real workflow, with the context and trust a general chatbot can't offer.
Physical products are next. Leo is that platform for mechanical engineering.
Your IP stays yours
This is the first question every engineering leader asks, and the first reason serious teams say no to the chatbot route. A plugin on each laptop, tied to each engineer's personal AI account, is not something you can govern.
One defense engineering unit we work with had tried ChatGPT and Gemini. Every query had to be scrubbed of sensitive details first, on machines outside their secure environment. It wasn't usable. With Leo, they work inside that environment, with no scrubbing.
Leo is SOC 2 Type II certified. Each customer gets an isolated instance, and Leo never trains on customer data.
"Leo is SOC 2 certified and nothing leaves our secured environment, which cleared the procurement review without issues." Engineer, defense and space enterprise (1,000+ employees)
FAQ
MIT NANDA, "The GenAI Divide: State of AI in Business 2025."
See Leo on your own design
From requirements to a reviewed assembly, live.
Bring one real project to a 30-minute demo. We take it from requirements to a reviewed, manufacturing-ready assembly, with your standards applied.
#1 New AI Software Globally - G2 2026
Enterprise-grade security
Trusted by world-class engineering teams
