Just a year ago, German startup 1000 Kelvin introduced AMAIZE, its groundbreaking AI co-pilot for metal laser powder bed fusion (LPBF) 3D printing. At a time when many companies were adding “AI” to their branding, AMAIZE stood out as a genuine application of artificial intelligence in additive manufacturing (AM). Using high-powered computing and physics-informed machine learning, AMAIZE optimized one of the untouched areas of the 3D printing industry, toolpaths for LPBF machines, and effectively cut out the costly trial-and-error cycle. In some cases, the software was even able to eliminate the need for support structures. After years of industry efforts to achieve “first-time-right” prints, 1000 Kelvin seemed to have found the solution.

Now, at Formnext 2024, 1000 Kelvin is demonstrating the next logical evolution in the usage of advanced physics-based AI. AMAIZE 2.0 builds on the success of the original by expanding its capabilities across the entire metal 3D printing workflow. Beyond toolpath and exposure strategy optimization, AMAIZE 2.0 features intelligent AI models trained on physics data to enable designers to perform printability checks at the design stage in seconds. Then, they can use AI to automate build preparation by identifying the right orientation, generating optimized support structures where needed, optimizing exposure strategy, and, finally, providing users with an accurate cost estimation.

In an interview with 3DPrint.com, CEO Dr. Omar Fergani, PhD., expressed his mission, “We are committed to solving the most pressing challenges of our customers. Thus, AMAIZE is evolving at a fast pace to address these urgent challenges. That said, our vision is not just to automate LPBF workflows, but to reshape entire manufacturing processes across industries.”

AMAIZE 2.0

1000 Kelvin claims that its toolpath optimization, AMAIZE 1.0, has reduced print failures and minimized distortions by 80 percent in some cases. Some customers could even print parts that were previously unachievable, allowing them to tackle larger, more complex supply issues. With the latest update, the company is inching ever closer to perfection. The updated AMAIZE software incorporates several new features aimed at reducing costs and failure rates while empowering users, regardless of their expertise level. These include:

• Printability Checker: Automatically validates and optimizes designs for LPBF, reducing redesign cycles by 40%.
• Cost Estimator: Provides accurate, upfront cost estimations, improving quotation accuracy by 30%.
• Automated Support Structures: Leverages physics-based build preparation to save up to 20% in material costs.
• Exposure Strategy Optimization: Ensures first-time-right prints with AI-optimized parameters, cutting failure rates by 50%.

“One customer submitted a design that required several manual steps: build preparation, simulation, printing, and redesign. After a failed print, they had to call the customer, request design changes, and go through multiple rounds of calls, emails, authorizations, and machine setup. This process took days and wasted significant resources—material, machine time, energy. With our technology, however, the first design was analyzed automatically. We identified an issue with the recipe, corrected it, and printed successfully on the first try.”

If all of this is true, AMAIZE 2.0 is essentially taking the manual labor out of what have typically been some of the most time- and labor-intensive tasks in design for AM. A print job that once took seven or eight tries managed by four or five engineers can now be left to a single employee.

“One of our customers is a service bureau with five quotation engineers—sales engineers,” Fergani explained. “Their whole day involves opening emails, seeing what part is coming in, and putting it into the de facto build preparation software—which has, in effect, only geometric features without any intelligence. The engineers then do some model orientation, add supports, calculate the cost, write the quotation, and send it to the customer without considering complexity or risk of failure. From the sales step, most of these service bureaus are losing money. Now, with AMAIZE 2.0, all but one of those engineers can be freed up to focus on other tasks. And they come away with a better understanding of the offers they send to their customers while saving time and reducing cost immediately by removing numerous software licenses.”

Beyond AM

At the moment, 1000 Kelvin seems to be focused on the world of AM, but it’s easy to dream beyond the 3D printing industry. AMAIZE was originally trained on countless points of physics-based LPBF data to optimize laser toolpaths. Now, it’s grown to encompass support generation, price quoting, and more. It wouldn’t be surprising for the technology to be applied to, say, electron beam metal 3D printing or directed energy deposition. And if it can succeed in the difficult world of metals, could it work with polymers?

We also already know that AI is being applied to nearly every field around the world. There’s no reason not to believe that, given the resources, 1000 Kelvin could ultimately begin automating entire manufacturing sectors.

“The sky’s the limit,” Fergani concluded. “When we discuss these ideas, we set a North Star, but I think what’s most important is that I have a path. I have a way to it. I know exactly where the road is, where we need to turn and at what speed. I wish I could accelerate time and space, but the reality is it takes time. It takes effort. If there’s one thing the team at 1000 Kelvin is really good at is that we are consistent and, every day, we’re getting closer to that North Star.”