Issue link: https://iconnect007.uberflip.com/i/1534953
manage it by having multiple sup- pliers or a bit of both? You have to come up with a strategy that cre- ates financial success for you. Accept the fact that your sup- pliers won't be perfect, but you can help them be better. That's where these requirements come in—going to audit and making sure they have process con- trol. But then, with good process control of core processes, a cir- cuit board with 400 steps or pro- cesses will not be able to do 100% process control of every process. Things will fall out of control every once in a while, so you must be able to manage that. The scary part is when peo- ple put all their proverbial eggs in one basket, and then the bottom falls out of the basket. If you've been around awhile, you don't do that. You have mul- tiple material suppliers, as an example, and can switch rela- tively quickly. I could sit here for hours and talk to you about how you do that, but it really is a thought pro- cess that you have to go through. GEN3 and I are providing the tools to allow you to do this and get information that brings value back to your company as far as making decisions on your suppli- ers and your supply chain, so that you can better manage it, mak- ing sure that once your products get into the field, they are reliable and working soundly. Ultimately, this is about keeping your cus- tomers satisfied. You mentioned your software iterations and that you are many versions from your starting point. AI is a big topic right now. How will AI come into the world of test, or has it already? Neves: It's on its way in because tests generate lots of data. The HATS² test system will gener- ate tens of thousands of resis- tance data points. The system goes through and highlights the worst and the best ones. But it's not necessarily looking at all the trends and everything else that may be in all that data. I can't do that without something like AI. The benefit of AI in testing is its ability to analyze not just the individual test, but also all the parameters: board thickness, via structure sizes, the number of via structures, and various other attributes across multiple jobs. AI can interpret all of this to iden- tify, "This now points to X issue," which is something I have never been able to do in all my years of experience. There are great pos- sibilities for AI taking the data that's produced by all the tests that we do and pointing back to process changes, process con- trol, and really high-value feed- back to help the manufacturer make better decisions. When all this data is captured, where is it being stored? Is it localized or is it in the cloud? Naisbitt: Data storage has always been a significant chal- lenge because test equipment, especially with SIR, generates a lot of data. We have the capa- bility to run 4,000-hour tests at 1-minute intervals which take 18 different measurements every minute. As you can imagine, that amounts to millions and millions of lines of data. If the PCs in use are not up to spec or are older, it can be a struggle to keep up, which may result in lost data or cause issues with the test itself. Then you have the famous Micro- soft updates. We advise our cus- tomers not to connect to the