Willem Mast

Creating foundations for defense startups, we bring you from R&D to manufacturing.

[Willem Mast @WMDSquared]

It’s never easy to make the stuff we use everyday.

[Ilir Aliu @IlirAliu_]

Robotics companies: let us solve repetitive tasks like folding shirts.
Meanwhile, the factory floor:

[Jonathan | Fabworks 🏭 @fabworks_guy]

90% the tube parts we laser cut are simple cut to length with a few holes.

This technology is extremely unexplored and while this is an extreme example I think more engineers need tube laser tech in their toolbox.

[Willem Mast @WMDSquared]

These guys are doing something new.

[Edward Mehr @EdwardMehr]

This is honestly great.

When we first told prime defense integrators, as a Tier 1, that we wouldn’t charge facilitization costs… it genuinely broke people’s brains a bit. Program offices literally didn’t know how to model a supplier that wasn’t asking the government to fund a giant pile of bespoke tooling and dedicated factories.

Traditional defense manufacturing works like this:
One weapon system. One dedicated line. One dedicated tooling stack. One giant bill before the first part even exists.

So of course primes had to charge facilitization. Their factories were effectively hardcoded for a single missile.

The beauty of Robocraftsman is that the same factory that builds one missile today can build another tomorrow. Minimal part specific tooling. Minimal reconfiguration. Software defined manufacturing instead of cathedral building.

Defense procurement has been optimized around inflexible factories for so long that people forgot factories themselves could become reusable infrastructure.

That changes the defense manufacturing math entirely.

[Willem Mast @WMDSquared]

As a founder, this is something I have to be careful of - constant improvement. Change is only good if it’s directed at real issues. It’s called “Organizational Whiplash”, if you don’t filter it, in Rocketfuel the book.

[Jaynit @jaynitx]

Jeff Bezos reveals the moment an early Amazon executive told him he had enough ideas to destroy Amazon:

"Early in Amazon's history, Jeff Wilke came to me one day and said, Jeff, you have enough ideas to destroy Amazon. You have enough ideas per minute, per day, per week to destroy Amazon."

"I was like, what do you mean?"

"He said, you have to release the work at the right rate that the organization can accept it."

"Every time I released an idea, I was creating a backlog, a queue, work in process. It was just stacking up, it was adding no value. In fact, it was creating distraction."

"So I started prioritizing the ideas better, keeping lists of them, keeping them to myself until the organization was ready for the ideas."

[Willem Mast @WMDSquared]

@ThePeelPod @jimbelosic @sendcutsend Great story, we need more cool stories like this. Great name too btw.

[Machina Labs @MachinaLabs_]

We build factories that deliver production grade metal structures for defense and mobility.

Faster than anyone thought possible.

And we’re hiring people who want to work on the hard stuff.

• Robotic Process Engineers
• Senior Mechanical Engineers
• Senior Industrial Automation Engineers
• Senior DevOps Engineers
• Software Engineers - Testing
• And more��

Machina is the Factory. Come build with us.

[Todd Saunders @toddsaunders]

I know Silicon Valley startups don't want to hear this.....

But the combination of someone in the trades with deep domain expertise and Claude Code will run circles around your generic software.

I talked to Cory LaChance this morning, a mechanical engineer in industrial piping construction in Houston. He normally works with chemical plants and refineries, but now he also works with the terminal

He reached out in a DM a few days ago and I was so fired up by his story, I asked him if we could record the conversation and share it.

He built a full application that industrial contractors are using every day. It reads piping isometric drawings and automatically extracts every weld count, every material spec, every commodity code.

Work that took 10 minutes per drawing now takes 60 seconds. It can do 100 drawings in five minutes, saving days of time.

His co-workers are all mind blown, and when he talks to them, it's like they are speaking different languages.

His fabrication shop uses it daily, and he built the entire thing in 8 weeks. During those 8 weeks he also had to learn everything about Claude Code, the terminal, VS Code, everything.

My favorite quote from him was when he said, "I literally did this with zero outside help other than the AI. My favorite tools are screenshots, step by step instructions and asking Claude to explain things like I'm five."

Every trades worker with deep expertise and a willingness to sit down with Claude Code for a few weekends is now a potential software founder.

I can't wait to meet more people like Cory.

[Willem Mast @WMDSquared]

This is pretty wild. Gotta love Lego.

[Willem Mast @WMDSquared]

@thesamparr Yep

[Willem Mast @WMDSquared]

@zanehengsperger Agreed, great company! I’m always amazing at how easy companies like this make it seem.

[Willem Mast @WMDSquared]

@durinmining Looking good.

[Willem Mast @WMDSquared]

Guess I’ll give it a shot, because no one it watching.

[Willem Mast @WMDSquared]

Should I switch this account to my name?

Mark your opinion on the poll below.

This space we are in (mechanical engineering) is finally starting to change and I want to be a part of it.

I’ve been running this X/Twitter account for over a decade for my 19 yr old engineering firm.

I’m a mechanical engineer with 28 years of rapid R&D to manufacturing experience.

We don’t often need actual prototypes. Just digital ones. Then we make hundreds or thousands.

[Willem Mast @WMDSquared]

@Dylan_Morri Energy, Defense, Aero Derivative Turbines and Industrial.

[István Csanády @istvan_csanady]

The Difference Between Writing a Direct Modeler vs. a Parametric Modeler

Shapr3D started as a direct modeling tool on the iPad. Since then, it has evolved into a hybrid direct-parametric modeler available on Windows, iPad, macOS, and Vision Pro. But why did we start with direct modeling—and how hard was it to add parametric modeling later?

There are a couple of reasons:

1. Implementing a direct modeling CAD system is an order of magnitude easier than building a parametric one.
2. Direct modeling naturally fits push/pull-style interactions, which are ideal for the iPad.

Starting off as an iPad app was one of the best decisions we’ve ever made. Our full-featured mobile CAD remains one of our unique differentiators and a strong value proposition for our enterprise customers. It also gave us a greenfield market, enabling us to scale to $10M+ ARR within just the first few years.

However, in 2022 we decided to expand to desktop platforms and add parametric modeling—part of our journey to become the de facto next-generation CAD platform. And that transition was hard. Very hard.

Being resource-constrained in the early days, choosing not to take on the challenge of building a parametric modeler too early was the right call. But why exactly is it so much easier to build a direct modeler?

1. The Kernel Does Everything

Building a direct modeling CAD system requires far less work from the CAD developer because the kernel provides nearly all of the modeling operations, the data model, undo/redo, and many other components that are exceptionally complex to implement on top of the kernel for a parametric modeler.

In a direct modeler, “all” you need to do is add a renderer on top of the kernel, build the UI, and persist the static geometry data in the kernel’s native format. Of course, this is still challenging—but it’s about 5% of the effort compared to what’s needed for a parametric modeler.

2. Performance

High performance in direct modeling essentially comes “for free.” Similar geometry modifications in a parametric system can require 100x more kernel calls than in a direct modeler, since you’re modifying the feature tree rather than the geometry directly. This tree often has to be partially or fully re-evaluated.

As a result, everything must be updated and computed incrementally. You have to cache everything that can be cached—otherwise, the user experience becomes as sluggish as the legacy CAD systems. That’s clearly unacceptable in 2025.

3. Memory

In parametric modeling, every topological element must have a unique, persistent ID. On top of that, you need to store additional data structures and caches required for high performance.

This can lead to 2–3x more memory consumption compared to storing the same geometry in a direct modeler. While our direct modeling engine could handle tens of thousands of parts with relative ease, our team had to work extremely hard to achieve similar performance with the parametric engine.

4. Versioning

In a parametric engine, geometry is the result of a sequence of operations. When an operation changes, you need to create a new version of it to ensure that older designs—relying on the previous behavior—do not break.

This introduces significant development overhead and architectural complexity.

My this experience, my estimation is that writing and maintaining a parametric modeler requires 5-10x more effort than a direct modeling engine.

[Alexander Mackie @ZanderMackie]

Reading the NSIB from @ronaldreagan Institute feels like a TODO list! Notice this the first time they assess industrial readiness.

TG for our growing manufacturing and defense tech base like @Mach_Industries and the folks at the deep tech week panels yesterday @FilArons
@aphysicist
@taiwo_oluseun
@billythalheimer
@tfeldmann
@ExowattEnergy
@QuaiseEnergy
@regentcraft
@DiracInc
@durinmining
@hankai1998
@gbthevc
@RayyanIslam1
@SolideonUSA
@atomic_inc
@Andercot
@GaruthAcharya

etc

https://t.co/D4M2s5vvaU

[Willem Mast @WMDSquared]

@EdwardMehr Nice work. Nothing like modular machinery. Lots of work to get there, but benefits are huge !

[Willem Mast @WMDSquared]

These companies benefit from slow software = more users to complete a task = more profit.

You are correct. In 25 years, Solidworks has stayed nominal. Autodesk Inventor has gotten better in the first five years but stalled out , and Creo has stalled out.

Is Fusion the only one that has already implemented AI?

Onshape is getting a good following.

[Willem Mast @WMDSquared]

@codyaims Good story, thanks for sharing.