Text-to-STL

Text-to-STL tools are very crude and easy to dismiss. Engineers, in particular, can easily overlook the impact of these tools because they replace them, their craft, and their tools. But these tools can democratize the creation of files far more than the others. They could take millions of people to create what they need. This could make 3D printing, though desktop machines and services, much more meaningful and accessible to millions. This could be the Cambrian explosion moment that we’ve been waiting for in additive manufacturing. Yes, these tools suck at making good geometry. And yes, they suck at technical parts. Maybe they’ll get better, but let’s imagine that they only kind of get a little better. They’re still easy, but you couldn’t make a spoiler for your car with them. But, imagine that they were good enough for earrings, brooches, toys, and creative things for around the house. Then still, it would make 3D printing far more relevant to millions than it is today.

It’s kind of like saying the Cray supercomputer is irrelevant to the common person because it is only used for nuclear fallout predictions. No, in fact, this does in a roundabout way make it very relevant indeed. And if we look at the extent of this development, computing, then the personal computer or the networked cloud computing concept will have a daily place in our lives. CAD is now in a silo, enshrined and worshipped by 2 million experts worldwide. The other 8 billion people depend on these people to make all the stuff in the world. Directly or indirectly, they contract these 2 million to make all the stuff. Just like the scribes in European monasteries were doing all the writing for all of Europe at one moment. But Text-to-STL could let around 6 billion people have access to 3D creation. Now, this doesn’t mean that these guys will all make their own cars or produce nuclear reactors. This will be a bad idea. I can not write a script as well as a medieval scribe. But through typing, computers and email I can communicate with millions of people far more easily than they can. So my skill in putting text down is much diminished when compared to theirs, while my ability to produce text and communicate it is vastly more advanced. My text is infinitely less beautiful but infinitely faster. Given the care per letter, I’m also much more likely to make a typo.

Similarly, we used to rely on professional photographers for all the photos. Then you’d take special family photos with a big camera. Now we take thousands of pictures with devices we always have with us. It’s not that there are no more professional photographers in the world. It’s just that we value the skill less, hire them less, and there are more pictures in the world, with more people spending time taking them. And the fact that so many can publish pictures has led to Instagram, citizen journalism, and a bizarre cataloging of pictures of people’s dinners. The effects of instant valueless photography are, therefore, both easy to imagine and difficult to interpret. As in, I think we can all expect many more creations and many more people making things. This will lead to more 3D printers and more filament being sold. And then? We don’t really know whether everyone will make their own toys or whether tens of thousands of candle entrepreneurs will result. That all depends on the specifics.

AI Plug-ins & Co-pilots

These are likely to become much more prevalent through the actions of Siemens, Solidworks, Autodesk, PTC, and the like. We can see a strong interest among those firms in AI. They don’t want to get replaced as they’re addicted to all those yummy subscriptions. They’ll continue to deploy a lot of co-pilot-like tools inside their ecosystems. The value of these tools will depend on how many critical errors they make and overall trust in AI. The big CAD companies could make much more money by offering extra subscriptions through validated checker tools inside their CAD packages. A few dollars a month here for compliance to a standard, a few more here to check that all your tolerances or hole sizes comply with something else, some conversion tools there could really add up for them. Instead, these firms will probably ape the latest co-pilot and agentic trends, offering tools that may be cutting-edge but will be accident-prone. More general adoption will depend on values or specific news. I lived in Eindhoven, and there was this self-driving bus line years before this was common. One crash and a couple of choice pictures for the newspapers later, there were no more self-driving buses. Agentic and AI adoption within CAD tools hinges on moments like this. It would be folly to go all in on adopting these tools, and it would be silly to ignore them. But, adoption will depend on individuals’ views and trust in AI. If these tools let me move to a higher abstraction level, so that instead of designing every brick I can design one and quickly build a wall, adoption will be broad and across all tools.

Workflow tools

The greatest financial benefits could accrue from workflow tools. If you care about a workflow or file conversion, you can code your own tool today. But, if I, with ten people, make a tool that saves you half an hour a day and costs just ten bucks a month, then you won´t even bother. Maybe in Orthotics and Prosthetics, for example, scan-to-mold tools will get broad acceptance, while in general, scan-to-3D print people will build their own tools for their own workflows. But, in this specific O&P market, a team can develop a validated workflow that is faster and easier than a tool; there could get a lot of customers. For entrepreneurship, a well-calibrated offering in workflow that steers clear of the functionalities that will become commonplace through LLMs, will be the clearest path to long-term profits.