This week delivered a chilly, brisk start to RAPID + TCT 2025 in Detroit. As I walked the show floor, I wandered the booths filled with equipment and printed parts, trying to figure out what was new and interesting in the industry. It struck me that the most likely source of rapid change to the AM industry will (as always) come from software and that the emergence of AI could make a huge difference to the holy grail of quality, speed, and cost. I decided to explore what AI-powered AM software is currently available. I spoke with a few experts in the field and took a closer look at some emerging tools. I also looked at other AM software platforms that could benefit from AI, some of which, I think, are just begging to have AI integrated.
AI Tools Worth Watching
A few obvious new AI-based products, namely Text to 3D from HP, Aibuild, and Synera, immediately popped up.
Text to 3D by HP is an ingenious little app that has stemmed from a research project in the company. Like ChatGPT or other AIs, the engineer can type in quick instructions. In the case of the demonstration, it was a steering wheel model with a specific logo in the middle. Within a few seconds, the model was created, and variations of the logo could be tested visually.
Text to 3D works as a private or semi-private implementation, and HP has been working with key manufacturers to add 3D data within these private environments, enabling AI to train on proprietary IP and create custom, domain-specific models. It works very quickly and can create tens or hundreds of iterations, depending on the user’s text instructions.
Because it is an HP product, it focuses on output format for Multi Jet Fusion (MJF), or as HP says, “HP’s manufacturing intelligence auto-orients, strengthens, and optimizes models for superior 3D printing results. No guesswork—just accurate prints.”
According to Mark Burhop, previously a researcher and principal investigator at Siemens who has now left to found a company focused on AI for manufacturing, teaching AI the “language of design” is a tricky notion. You can’t just throw an engineering textbook at the AI and expect it to rationalize design — it has to be trained. And this, it seems, is what HP is working hard to figure out.
The HP Text-to-3D project uses word-based prompts to deliver a custom 3D steering wheel design using AI. Image courtesy of Rachael Dalton-Taggart.
Another take on AI in AM comes from AiBuild, which is focused on toolpath generation and hybrid manufacturing. Aibuild uses AI to deliver hybrid 3D printing and CNC machine multi-axis toolpaths via a single user interface (UI) with a library of built-in examples and templates designed to make generating complex toolpaths easier and replace manual coding with visual programming.
It predicts the thermal behavior of a wide range of materials physically characterized in AiBuild’s R&D lab, automatically optimizing process parameters for each material in the production process.
Another company that is bringing AI into the AM workflow is Synera. Its process automation platform is designed for engineers, enabling the rapid creation of complex CAE (computer-aided engineering) workflows—without coding. Its low-code interface and seamless integration with CAx (computer-aided technologies) tools make workflow setup quick and easy. Workflows can be shared across the company via a web browser, allowing non-engineers to access and utilize engineering outputs.
Synera also claims to make it easier for engineers to integrate AI applications by automating training data generation, training Reduced Order Models, or interfacing with GenAI. According to Product Lead Andrew Sartorelli, his team is developing “agentic workflows” (which Microsoft Word thinks is a made-up word, as do I)—essentially, systems that allow a manager to run multiple AI “agents” to find, deliver, and analyze data for AM builds.
Why AI for AM?
Since AM is inherently digital (and yes, CNC is, too), and AI needs digital data, the possibilities for AM are wide-ranging. According to Burhop, within the next year, AI implementations can tell if an AM part design and build is bad.
“In five years,” Burhop says, “AI will drive and direct most if not all simulations and create statistical models for evaluation, certification, and qualification.”
He predicts that AI for engineering will need a “System 2 thinking approach,” which goes beyond the current “reading” of documents to reason an answer, which is also the basis for mathematics and design.
He predicts that as AI for AM reaches this System 2 approach, it will unlock a range of capabilities. AI could accelerate machine, materials, and process qualification through statistical models trained on real data. It might learn to follow published standards, improve part quality, speed up production cycles, and monitor and analyze manufacturing in real time. Downtimes could be managed more proactively, failures and waste reduced, and designs adapted automatically based on AI’s training. Although it all sounds great, I’m still skeptical that all of this will be possible any time soon. But it’s nice to think about.
Which AM Software Is Next in Line for AI Help?
I am sure the developers of the products below are already looking at AI, but it struck me how naturally the digital nature of AM lends itself to AI integration. These few notable products shown at RAPID this year seem ready for an AI boost—or at least deserve a mention.
One product is Flow-3D AM, a simulation tool from Flow Science that is tailored for metal powder bed fusion (PBF). It tackles common print challenges like porosity, overheating, lack of fusion, and balling. It also claims to predict in-situ material behavior that can cause keyhole instabilities—an issue known to lead to serious defects. Beyond that, it promises to help engineers optimize print parameters for tricky features or better surface finishes, improving mechanical strength. While the software can already reduce the number of experimental runs and support stronger design of experiments (DOEs), it doesn’t yet use AI. But based on its goals, it seems like a perfect candidate.
Another standout is Mira3D, a small but promising startup from Mumbai that is building AM preparation software with impressive features. It covers everything from internal mesh repair and thickness analysis to part packing, DLP build support, and toolpath definition. It also offers part inspection software. While there’s no mention of AI yet, I am sure they will soon.
Then there’s 4D Additive by CoreTechnologie, which focuses on native CAD file import and slicing for various formats. It also handles B-rep model repair and automates nesting. The company describes it as “intelligent,” though it’s unclear whether that means true AI is involved. But if it isn’t already, it likely will be soon.
3DXpert from Oqton has a good reputation as an early metal AM print production software, and the software continues to evolve. Recently, it added a Build Monitoring feature that had been missing until now, enabling real-time oversight of build quality. This product is ripe for true AI.
A summary report page within 3DXpert shows issues found during the print. Image courtesy of Rachael Dalton-Taggart.
Materialise Magics needs no introduction. Obviously, as a long-time leader in AM build prep software, there was no mention of AI on the website, and the company rep I spoke with didn’t bring it up either. That said, it’s easy to assume that Materialise’s management is looking at it.
AMIS Pro, a Belgium-based software company, offers build prep tools for various platforms, including MJF, SLS, MJ, and BJ. It’s a focused and capable toolset, but again, the presence of AI wasn’t clear from what’s publicly available.
Then there’s Zeiss Data Management, which was a bit of a surprise. The software can combine multiple inputs from a wide range of measurement systems, including CMMs, CT scans, X-rays, manual input, and more, to deliver an archive and a scorecard on quality and measurement issues of parts throughout an organization. Currently, it is down to humans to spot and report on trends, issues, and more, but that kind of data is begging to be in an AI system for automated analysis and trend reporting.
Adaptive lattice creation in 4D Additive’s software. Image courtesy of Rachael Dalton-Taggart.
So What Next?
I believe it is a given that AI will become ubiquitous and almost invisible within AM and overall manufacturing software. From design iterations and build prep to the design of experiments (DoEs) and qualification processes, AI will quietly power much of it. But large datasets are required to make this work and the inherent secretiveness of OEMs around proprietary information can make this difficult.
“There is already a data struggle,” says Andrew Sartorelli at Synera. “To pull this off as an industry, we have to share the data; we have to use it. But OEMs rightly don’t want to give up their IP.”
CAD-based data presents its hurdles—not just in access but in complexity. AI would need to understand parametric history trees to be truly useful.
“We can maybe teach AI the CAD API,” says Mark Burhop, “but it needs to be taught how to understand geometry.”
Fear and cynicism around AI also create barriers, but Burhop advises: “Just use it!” He suggests trying out the commonly known AI platforms, such as ChatGPT and Perplexity, to understand what language-based AI can do.
As for ways to look at in-house AI for AM implementations, Burhop suggests starting with OpenAI tools and checking out Open Weights for training models. More opportunities are opening up for those looking to educate themselves or their teams. MIT now offers “MIT Artificial Intelligence – Understand AI Design Processes,” according to Sartorelli, the University of Utah also has some good courses. Naturally, there will be more soon.
Subscribe to Our Email Newsletter
Stay up-to-date on all the latest news from the 3D printing industry and receive information and offers from third party vendors.
Upload your 3D Models and get them printed quickly and efficiently.
