Wired In: AI Strategy and Application in Electronics & High Tech
00:09 All right, let's talk a little bit now about use cases, right? Mm-hmm. Not all AI use cases fit, you know, ne- neat- neatly and nicely into every industry, right? Sure. So what are some specific AI applications or use cases that, that you're seeing that proves most valuable for companies in the electronics and high, high-tech industry today? Yeah. One that I love is the, like, the ALM space. Such a ripe area for innovation. It's so critical as more and more, like, we're dealing with software and hardware, but, like, in the EHE space, like, you've gotta go through rapid iteration, you gotta be
00:40 compliant with regulations, you're working with lots of materials from different locations in the world. So those requirement's essential in terms of, like, what is the intent of that next delivery? And what AI is really good at is processing that type of data, 'cause requirements are written text. AI loves natural language. So in there, what I love are use cases like, one, just like basics, like, can you just check for quality? Sure. Hey, we as a company, we have standards in how we deliver it, templates. Can you review that for quality so I can have that standard ac- standardized across the company? There's all kinds of benefits in terms of
01:10 efficiencies, of upskilling your, your new engineers, but also just in terms of meeting requirement com- in terms of compliance. So that's one area I like. Where it gets even better, though, is like, all right, can you start generating things for me? Like, can you now, based on those requirements, start generating test cases? Right. So I love that use case, too, for the EHE market, 'cause you wanna validate whatever's coming out, you wanna do that quickly. It's speed, speed, speed. So can we have the AI generate test cases to quickly validate whether those requirements meet that market demand, and then go for it. Like, I like that a lot for the EHE space.
01:41 Yeah. You know, and I'm seeing, you know, I spend a lot of time in front of customers and, and at events and things, and we hear a lot about the, what you just described- Mm-hmm writing requirements and, you know, how that's becoming even more important across multiple domains and disciplines within the organization, right? So you might have an engineer who's working on a circuit board or on, you know, a, an integrated circuit or something. And there's all these different simulation requirements, mechanical, thermal- different things that would come in now that might impact software development- Yeah and firmware, and maybe the enclosure that that thing goes into.
02:12 And, you know, requirements and the, and AI being able to interpret and understand how different things impact others based on change or different variables that a human might not catch, 'cause they're in a different system and there, there's not visibility there maybe. So I s- I see AI really being able to connect the dots in ways that companies historically have struggled. Yeah. and another great example, too, is, you know, companies that are moving into, you know, like, taking design data and trying to simulate- you know, these different scenarios, where they wanna spend less money and time on
02:42 prototyping and developing, you know, early builds. Mm-hmm. And how can they do that inside of our tools and, and with their other tools they have, and use AI to give them guidance and insight to say, "Well, you know, if you move this connector over here by two millimeters, and, you know, increase the height, you can insert it, you know, twice as fast, and now you can build 5,000 more boards a month." Yeah. That makes a lot of sense. It's, as you're kind of jumping, you know, going through those use cases, you kind of trigger me to think about traceability. Like, back on the ALM front, like, and again on the product data foundation, what's
03:13 interesting is that, like, when you use a tool like Codebeamer, like, a core value prop for that is, like, it's the traceability between- requirements and test. And so when I think about the layering of value, like, set up your product data foundation, use those requirements systems, not just to write the requirements and have that passed down to test systems, but they're, behind the scenes, you're creating links and traces, and the AI, again, finds a lot of value in that. So when it's ready to now generate the next taste- test case for that requirement, it's just, it's smooth. Like, there's no, like- Yeah no manual work to be done because you have that traceability,
03:45 and I think that, that traceability then leads into other areas like systems modeling or down to design. So my whole point here is that, like, the story of layering more and more value is truly the story, like, that story of digital transformation, we're really starting to see the outcomes now. It's like- Absolutely becoming real. We're getting more speed, we're getting higher quality, and AI is just surfacing that value faster and at greater scale. Right. And to your earlier point, you know, I, I couldn't agree more. Like, the, the industry, el- electronics and high tech especially, we see extremely high levels of innovation, so they want to adopt AI
04:16 as fas- fast as possible. Yeah. Find ways to make them, you know, be more efficient, more effective, move more quickly. But they also need to compress their development times. Yeah. They need to get their products to market more quickly. I mean, just look at the demand for data centers and, and hardware that goes into those. You know, the, it's an insatiable appetite right now from a lot of these, you know, companies we're talking about that are leading the AI charge, and they can't get this hardware fast enough, or the power needed to, to generate, you know, the tons of megawatts they need and things like that. Just curious, like, you know, as, given the speed requirement for EHE, like, what have you
04:48 seen, you know, with the customers you spoke to out in market and, and your experience talking to analysts, just in general the ecosystem, like, what's been their appetite for AI? Are they, like, faster adopters? Are there certain blockers for them? I'm just kinda curious. I would say one word would be frothy. Yeah. Right? And you know, there's a lot of buzz and excitement, but there are companies who are either building some of their own things, whether it's, you know, with us in collaboration or, or on top of our tools, they're adopting their own tools in-house. Yeah. There's a lot of partnerships that are being developed quickly and, and rapidly in the EHE space, right? Yeah.
05:21 But, th- I mean, the appetite is extremely high. Uh- Yeah, I would think so with, like, you know, the big Nvidias of the world- Oh, yeah but all the other customers out there, like, they're, there's a, they're naturally in this space, and the speed advantage of AI I would think would be a, a ripe area for them. And they're coming up with really kind of unique, innovative ways to, to address things- that may, that maybe we haven't thought of, too, right? So that, 'cause the level of collaboration I'm seeing where customers are coming and saying, "You know, what if I could do this?" Or, "What, what if, you know, we took, you know, this piece of information from this tool and, and leveraged it against this information from
05:52 that tool and put an agent in between?" Right? Love that. And, and it's these kind of conversations are really fostering some new ideas that I think will, will come to market in time. Yeah, I've seen that a lot. Like, there's more exchanges and partnership 'cause we're all really trying to take advantage of AI, but, like, as you, as you talked about, it's going fast. Yes. And so we're all learning, and I do see an increase kind of exchange between PTC and its partners and, you know, I think that's really beneficial for everybody. So at PTC, we use a framework of advise, assist, and automate for AI capabilities, right? Can you walk us through how an AI agent might automate a complex workflow, for example?
06:25 Maybe something like an engineering change request in real time. Yeah. Sure. It's a good one. Like, if you look at the market and talk to our customers, you know, change management is a major pain point. Like, some of the figures I've seen is, like, half of the time that a customer spends in a change process is non-value-add time, just, like, exchanging information, checking the, the I's and the dots on the I's and crossing the T's, for example. So any time savings we get from change manage is big, and so this is a big use case. And so how can AI help?
06:55 So w- you know, this framework, advise, assist, automate, it just, it goes through an evolution of, like, what an agent can do. Mm-hmm. So at, at the advise level, it's like I can answer questions for you. I'm an advisor. Sure. At the assist level, it's, ah, I can now start assisting you with some of your work. And automate is like, I can do something end to end. So you asked about the automate side. I'll, I'll blend in automate and assist. So let's take an engineering scenario, like there's some part in a, in a, in a component that has a faulty, you know, faulty, like, quality issue- Sure and you need to repair that. Well, what if you could have an AI system that
07:28 could do an impact analysis? So let's say I wanna look at, you know, from a PLM perspective, like, what are the components that are impacted by it? What's the downstream inventory consequences of that? Yeah. Let's come back to requirements for fun 'cause that was kind of the thread earlier. Let's see if there's any requirements impacted. Let's use that traceability. So let's just call that the scenario. So let's call this person Roy. Roy, the engineer, is doing this change impact analysis and gets on this AI system. What an automate assist experience looks like is that Roy is talking with that environment,
08:02 that AI system, and, and, and task it to go do something. Mm-hmm. And then Roy goes back and does other work or go grabs a coffee or whatever, whatever is the case may be, and the AI can go off and start running that impact analysis, start pinging the PLM system, start pinging the ERP system- Sure start looking at the requirements, and then come back and give an initial report and say, "Hey," just like in ChatGPT format to start. Yeah. Like, "Hey, here's all Here's what I found. These items were impacted. This, in this warehouse, I see this much inventory here, this much inventory there. From a requirements and trace perspective, I'm seeing this parent assembly impacted, but
08:35 actually, I, I have some gaps. I don't see the child items that you're referring to. I see a traceability gap. You might wanna look into that." So in that story, there was some automation in the sense of, like, it went off and, like, did some work on its own and came back with a report, but, it's assisting you, too, in the sense like, "Let me give you the report," and have the human in the loop and say, now it's up to Roy to start, "Okay, what do I actually wanna do with this information?" Right. "And do I wanna bring that back into Windchill, for example, in that CR object?" So that's a combination of assist and automate. There's gonna be scenarios, too, where, like, automate's fully there, like where you have
09:08 agents in the background. I think a classic one is, like, monitoring a scenario. So, like, and again, let's keep it with change. Kind of imagine you have some kind of change manager or some kind of program manager. Their responsibility is, like, to move the change through its stages. Sure. What if you had an agent doing that, an agent that's, like, monitoring that ecosystem and saying, "Throughout this process, I'm noticing that some documentation isn't filled out here"? And it's not a domain expert, but it's a process expert, and so that agent can automate the oversight and actioning a- against gaps or, or identify risk and then communicate that out
09:42 to the r- to the individuals by, like, sending a notification, whether that's a, a toaster in, you know, your UI or an email- Right you know, in your Outlook or something like that. Yeah, and then, I mean, I think a lot of it, too, is kind of this aggregation of information- Mm-hmm across var- a variety of systems. That takes a person a while. Yeah. I mean, as a former engineer, I remember sitting and looking through change requests and going through where used and doing impact analysis and doing dispositions- against inventory and ERP. Exactly. If an agent can do that in, you know, while I'm on lunch break, and I come back, and I have an entire summary all, you know, facilitated through natural language- Yep and
10:13 then I, you know, human in the loop, then I get the opportunity to say, "This is right. This needs more, you know, a little more depth here. This one we can ignore," and send it, send it along, right? Yeah, absolutely. I think it- I mean, that's, that's an incredible use case. 100%, and I think the numbers are somewhere, like, in the 15 to 20% of your weekly hours are spent searching for information. Sure. Totally. So, like, if we cut that down, that'd be good. Yeah, I agree. All right. Well, let- let's look now, looking ahead, right? How do you see AI further transforming, you know, these cross-functional collaboration opportunities between groups like engineering-
10:43 quality, and supply chain teams? Yeah. I mean, I get really excited about this space. Like, I think this is, this is the North Star for PTC. Like, how can we enable cross-product AI that elevating the enterprise IQ? And I'm, I'm just gonna ground it, one, in terms of, like, what we're doing, but also what I'm seeing, like, in my own personal and professional work. So I'll, I'll start there and then bring it over to, like, what we're doing at PTC 'cause it's, it's a super interesting time. So we were just talking earlier, like, in our pre kind of, like, coffee break about, some of the tools that we're using to do collaboration, our own, our own business processes. Sure.
11:14 And, what's fascinating to me is that I'm able to use these tools and, like, from a product manager, like, connect to Jira as a system of record or connect to my Microsoft Office productivity tools, but I've got this, like, home base, like this coworker that I'm collaborating with, and it's- Right I'm tasking it to do certain work. I'm creating skills or whatever. That's what we're gonna be doing at PTC too. So we're gonna be having this, you know, engineering cockpit, this interface where you can interact with all the tools you have in PTC's products portfolio, but also your own,
11:45 your other systems. And through that interface, it's gonna be purpose-built for engineering. Right. There's gonna be these skills for engineering. There's gonna be this interface that's, like, task based. And so where things are going and what we're working on is that single pane of glass and engineering cockpit and engineering workbench that can then interact with all the agents we've been building that are in our individual systems of record. So coming back to that analogy, like the contractor and the specialist, like, imagine that cockpit being kind of the control station, and then down below are all our products.
12:18 They're ac- they have these agents working in them, the specialists, the CAD agent, the requirement agent, et cetera, and they're now gonna be working together. There's this orchestrator that can interact with those specialists and then surface up- Cross product use cases, like the one we just talked about, like the impact analysis. So now I can ask a question like, "What's the impact of this change?" Sure. Go reach out into Codebeamer, go reach out into Windchill, go reach out into another product like SAP, for example, and get that answer for me. That's where things are going in terms of the experience, and then what that means for the
12:48 EHT market is that we're gonna get more collaboration. Like, walls are gonna break down between these different business units through these interfaces, and where we're putting our bet at PTC is in that engineering and extending that engineering experience to downstream things like service, for example. Yeah. No, I love the engineering cockpit kind of concept. Yep. And, and obviously here at PTC, like, our heritage has traditionally been in engineering, but we also know the value- Yep of the data that's in engineering and providing that kind of single pane of glass, that substrate for other teams to then tap into and to build off
13:22 of, right? So there might be agents in procurement. There'll be agents in supply chain. Okay. Your material planning teams, right? Your manufacturing operations teams. And, and that product data foundation kinda coming full circle. Yep. That, you know, you really don't get to have all that without that really robust, powerful foundation. Yeah. Just to build on that, like, that's the thing that, that's really giving me conviction that we're gonna deliver on that. We're already seeing it in the use cases in our products, but, like, we're starting at that product data foundation. Get your data house in order, right? Right. Then we're putting agents inside your products,
13:54 those specialists that can go do specific jobs in that advise, assist, automate framework. And now we're going to the next level up, which is this cross-product cockpit or workbench that orchestrates across those systems. All that is about added value and keep building on existing infrastructure. That gives me a lot of conviction in terms of the success that we're gonna deliver. Hey, Aura, thank you again so much for joining us on Wired In today. This was a really excellent conversation. Thanks, Stephen. If you're ready to explore more on this topic, check the description below for additional resources, and don't forget to hit that like
14:24 and subscribe button to stay Wired In.