Where this conversation describes AI being used to build or deliver a product. Part of the AI in practice tracker.
Engineering & simulationAMAIZEcustomers able to reduce the amount of distortion to an acceptable tolerance
The AI platform predicts thermal issues and residual stress in 3D printed components, allowing engineers to adjust scan strategies and recipes to minimize distortion without redesigning the part.
“we see every week customers able to reduce the amount of distortion to an acceptable tolerance without having to anchor or compensate the design and things like that. So this is one use case.”
Engineering & simulationAMAIZEincrease in the amount of valves, closed impellers, open impellers in the oil and gas
The tool analyzes G-code and design files to generate corrected print recipes (scan strategies, timings) that make existing, non-optimized designs printable, enabling the production of spare parts without modifying the geometry.
“We have a lot of success involves, for example, we see a massive increase in the amount of valves, closed impellers, open impellers in the oil and gas, where they just literally want to recipe so that they can print it instead of having any modification.”
Engineering & simulationAMAIZEpost processing is 20 to 30%
The AI engine helps predict where thermal issues will occur and suggests solutions to reduce support structures, thereby lowering post-processing costs which account for 20-30% of the total cost.
“They tell us that post processing is 20 to 30%. If we can help them reduce the amount of support structures, then they will be able to produce that. So they use our tool to manage the thermal better and produce surfaces with reduced amount of support.”
Engineering & simulationAMAIZEworkflow is extremely fast, brutally efficient
The platform acts as a co-pilot that integrates into the design workflow, allowing engineers to drop in G-code, receive analysis of issues, and iteratively compute corrections to find a valid print recipe, drastically speeding up the process.
“You open a 1000 Kevin, platform, maze, you drop it. We analyze it, we show you where the issue is, we tell you how you want to correct it. You tell us what you want to change and whatnot. And then we will go and apply that correction iteratively computationally until we find the right new recipe that will make it print and you can download it and go to your machine.”
Engineering & simulationAMAIZEgenerate 1000000s and 1000000s of data points
The AI models are trained on millions of data points to predict microstructure and properties for specialty materials that are not available off-the-shelf, enabling the printing of new material formulations.
“In order to develop this algorithm, we need to generate 1000000s and 1000000s of data points, and that's the hardest part of our R&D activity. So they come to us and they're not asking us for the traditional titanium 64 or Econ 718, but they're coming with a new formulation. I don't have that part of the shelf. We need to go train it for them.”
Running the companyincreased their productivity massively
The company uses AI tools internally to increase the productivity of its own engineering team, with leadership actively measuring the resulting efficiency gains.
“My engineers have increased their productivity massively. I can, I actually spend my time sometimes with my co-founder measuring the productivity increase, and it's mind blowing. So I wish everyone benefits get that benefits in their organization as soon as possible.”