They probably still have the actual geometries of these parts, perhaps the CAD files or drawings. The biggest challenge will be taking this data, structuring it and finding algorithms. But I think they probably have a surprising amount of information that they need.
Industry Week: How does a manufacturer decide if it’s worth investing in IT resources to try to use this type of technology?
Randy Altschuler: This is a great question: how do you determine ROI? I would take something you’re already doing today, build a team around it, figure out how to collect the data, structure the data, and build the algorithms. You just start small. Take a test and see how it goes.
What is the return on investment on this? I have two people dedicated to it, they cost me improves and my margins are higher. these with the prices I selected. And then you can figure out how much additional work these two programmers can do, what the bill is for where I store the data.
Industry Week: Does your AI look at data from parts submitted for a quote and also determine what the best manufacturing modality is for that part if there are multiple options?
Randy Altschuler: It’s even a step further. This gives the customer the opportunity. It makes available to them all the different ways we can do this. But rather than saying here’s what you can do, it’s more like we’re giving you every possible option and excluding what we can’t.
Industry Week: How much did you have to retrain your AI and improve accuracy each time you added a new manufacturing method to the options?
Randy Altschuler: There is absolutely a curve. The trough of the curve is now higher at the exit of the gate. We have become smarter and better at how we do things. This also applies to geography. Even if we actually use the same algorithms, training on the new data, things can be more or less expensive (in Europe or Asia).
Industry Week: So what can a manufacturer learn from Xometry’s experience in terms of what they need to know to adopt AI?
Randy Altschuler: Data capture is essential. Historically, this hasn’t necessarily been a priority for businesses, especially manufacturers, but be sure to capture your data even if you don’t yet know what you’re going to do with it.
Industry Week: But how do you know what data you’ll need to train the AI at the very beginning?
Randy Altschuler: Some of these processes, as we launch them, we were doing them manually before, we didn’t score them automatically. We have created prices for customers. They said yes, they said no, we saw what the profitability is. We’ve seen what suppliers accept, and then we use that data to develop an automatic quoting AI model.
Industry Week: Manually organize the data, for starters.
Randy Altschuler: Exactly. And if you’re a manufacturer, you have that manual data as long as you’ve collected it. So you almost have an advantage compared to when we started the company, where we had no data. …Manufacturers will already have this data. It’s just a matter of accessing it.
Industry Week: How does a manufacturer prepare to adapt more complex data to more complex AI? Should they work with third parties? Can they do it in-house?
Randy Altschuler: Start simple, take a simpler process that you do where you’re confident you know what data is already clean, what’s representative, and start there. If you succeed, you can invest. Maybe it’s going external, maybe it’s hiring people internally, but start small. Prove it. Show that you can achieve this or what the ROI of this is. Then you can expand.
Industry Week: Let’s say a manufacturer starts using AI for simple tasks, develops in-house expertise in machine learning, and now wants to use AI for something more complicated. To what extent do the lessons learned from these relatively simple tasks apply to more complex AI? Are you considering entirely new training, significant adjustments, or does existing expertise cover the most complex tasks?
Randy Altschuler: I think the technology, techniques and knowledge are very scalable across different sectors, but you will need an SME to help the AI programmer. If you try to use AI to help people decide what to eat for dinner rather than using AI to help you make manufacturing decisions, I think it will still be the same AI programmer , but the SME could be someone who serves school meals rather than someone. who has worked in a machine shop for 30 years.