Issue link: https://iconnect007.uberflip.com/i/1520213
12 SMT007 MAGAZINE I MAY 2024 If you think about physical industries world- wide, they're about 30–40% of our global GDP in regards to more than manufacturing and logistics. In the U.S., manufacturing is about $2 trillion—about 10–20% and 2–3% of global GDP. We have this huge part of our economy with 12–13 million employees, yet the produc- tivity levels have been stagnant or declining in the past few years, by 2–3% year-over-year in some sectors. Why is that the case? We've had digital transformation, Industry 4.0, and so forth, but we haven't increased pro- ductivity. In parallel, labor trends show there are huge problems with hiring and turnover in the sector. We have about 13 million employ- ees and more than a half-million openings in manufacturing. It's a huge gap; this problem will not go away. We need better leverage. How is your company addressing this? Arbabian: At Plato, our core thesis is that understanding and optimizing factory behav- ior is still very analog, very manual. We're still analyzing factory behavior like we did 150 years ago when Taylor and Gilbreth started the field of industrial engineering. ey did time-and- motion studies by observing with a stopwatch; we're not far removed from that today. Cur- rently, as a solution, we're sitting on the mez- zanine and watching the whole floor to gather data. Another solution is through Gemba Walks, where you're on the floor watching what's happening. Without the operator con- text, it's almost impossible to identify the right levers, let alone optimize, automate, and run it as efficiently as it's supposed to. Consequently, AI applied to manufacturing would be incom- plete without this new data set. AI is a function of your data quality and data sources. For example, you can't have a self-driv- ing car if you have no sensors to see the world. You might have all the internet data and all the big data, but it won't drive in a real environment. We believe that operator behavior unlocks the potential of AI for manufacturers. We started with semiconductor manufactur- ing because this segment understands the value of data. Every aspect of the machines and the environment is digitized. It's run with produc- tivity, efficiency, and quality in mind. We think electronics manufacturing sectors adjacent to semiconductor are similar. ey have the same bottleneck areas. It's more than building an advanced product; your process of build- ing it is also advanced. at's the idea for start- ing with a more complex and advanced man- ufacturing environment and propagating out toward a broader basis. Anders Holden: Getting more specific to PCB and SMT production, there's a sweet spot for our technology where the optimal balance between human and machine is reached. Obvi- ously, the line is automated, and it runs by itself most of the time. But when it stops, you need a specific, well-thought-out response. If a machine goes down in the middle of a run, you need to respond to the machine prompt. Maybe you need to change a reel or call maintenance, but the response and the actions are very tightly choreographed. at means we know what Amin Arbabian