WEBVTT 00:00:12.879 --> 00:00:14.759 Good morning, good afternoon, good evening. This 00:00:14.759 --> 00:00:16.980 is John Marchiando, and this is another episode 00:00:16.980 --> 00:00:19.920 of the Two Guys and a PLC podcast. I'm joined 00:00:19.920 --> 00:00:22.719 by my colleague, brother from another mother, 00:00:23.079 --> 00:00:25.839 Dave Gutshall. Good morning, good afternoon, 00:00:25.920 --> 00:00:29.199 good evening. Great to see everybody, hear everybody 00:00:29.199 --> 00:00:33.539 out there. Just a reminder, this is for entertainment 00:00:33.539 --> 00:00:37.539 purposes only. And all opinions expressed, are our 00:00:37.539 --> 00:00:41.619 own, not reflective of our day jobs, night jobs, 00:00:41.820 --> 00:00:44.159 jobs that we kind of forget because we took too 00:00:44.159 --> 00:00:48.179 long to wake up that morning, spouses, old flames, 00:00:48.579 --> 00:00:52.670 dogs, cats, everybody else. The opinions are 00:00:52.670 --> 00:00:54.969 just our own, John. Isn't that right, sir? Yeah, 00:00:54.990 --> 00:00:56.549 I never know what you're going to come up with 00:00:56.549 --> 00:00:58.070 there, so I'm always waiting to get a new one. 00:00:59.149 --> 00:01:02.549 The job section always scares me, though. I never 00:01:02.549 --> 00:01:05.290 know what's going to come out after that. Yeah, 00:01:05.370 --> 00:01:07.329 one of these days legal is going to come along 00:01:07.329 --> 00:01:10.750 and say, you know, you need to say it this way. 00:01:11.069 --> 00:01:13.409 Yeah, probably. That hasn't happened yet. So, 00:01:13.409 --> 00:01:18.060 Dave, we have PhD number two. on the podcast, 00:01:18.280 --> 00:01:21.019 which I'm shocked we ever got to a PhD number 00:01:21.019 --> 00:01:24.239 one. I'm shocked we ever got to a PhD candidate. 00:01:24.920 --> 00:01:28.680 Yeah, exactly. Exactly. Yeah. Do you believe 00:01:28.680 --> 00:01:31.319 that? So yeah, I guess PhD number two, I'm not 00:01:31.319 --> 00:01:35.420 sure exactly what order this will air as it relates 00:01:35.420 --> 00:01:38.319 to the second one. So probably a couple in between, 00:01:38.459 --> 00:01:42.859 but in any event, yeah, let's kick it over to 00:01:42.859 --> 00:01:47.560 our esteemed guest here. So I said, mister, I 00:01:47.560 --> 00:01:49.159 think on the last one, but I really should say 00:01:49.159 --> 00:01:53.500 doctor. So Dr. Jay Agarwal, sir, do you want 00:01:53.500 --> 00:01:55.879 to introduce yourself? You know, I'm equally 00:01:55.879 --> 00:02:01.480 surprised I got to PhD as well. But yeah, this 00:02:01.480 --> 00:02:05.920 is Jay Agarwal. I am an executive who leads a 00:02:05.920 --> 00:02:07.739 lot of digital manufacturing transformations. 00:02:08.180 --> 00:02:13.370 And I did come from academia at one point. And 00:02:13.370 --> 00:02:16.610 my PhD is in chemistry, and I fondly remember 00:02:16.610 --> 00:02:22.069 those times. True fondly? Yeah. Yeah, fondly. 00:02:22.129 --> 00:02:25.870 It also said on the little bit of research I 00:02:25.870 --> 00:02:29.180 did, Jay, that you have a PhD in chemistry. chemical 00:02:29.180 --> 00:02:32.099 physics i'm like those are the two subjects in 00:02:32.099 --> 00:02:34.479 college that i did the worst in chemistry and 00:02:34.479 --> 00:02:37.139 physics so combining them together for a phd 00:02:37.139 --> 00:02:40.180 man you you blow me away i think it actually 00:02:40.180 --> 00:02:43.759 made it more fun um you know chemistry you've 00:02:43.759 --> 00:02:46.180 got to mix all of the chemicals together in physics 00:02:46.180 --> 00:02:49.719 you have to learn about uh you know all the celestial 00:02:49.719 --> 00:02:52.060 mechanics but when you combine the two you can 00:02:52.060 --> 00:02:54.360 just kind of focus on math of small molecules 00:02:54.360 --> 00:02:58.069 and somehow that was a sweet spot that that gave 00:02:58.069 --> 00:03:01.789 me a smile for a number of years. Absolutely 00:03:01.789 --> 00:03:04.289 incredible. What was the, what was your impetus? 00:03:04.330 --> 00:03:07.569 Like why those two? Because it, to John's point, 00:03:07.590 --> 00:03:10.330 I feel like the combo is unique. So what, what 00:03:10.330 --> 00:03:13.550 from your background or your history, things 00:03:13.550 --> 00:03:17.050 that you ran into as a, as a youngin, what, why 00:03:17.050 --> 00:03:19.849 those two, Jay? You know, I kind of fell into 00:03:19.849 --> 00:03:24.000 it. You know, my father had a, a company a computer 00:03:24.000 --> 00:03:26.300 company you know through the dotcom boom and 00:03:26.300 --> 00:03:29.139 you know at the time I was exposed a lot to you 00:03:29.139 --> 00:03:31.500 know what was thin clients at the time which 00:03:31.500 --> 00:03:33.580 had went out of favor then kind of came back 00:03:33.580 --> 00:03:37.219 in favor and vice versa and you know when I entered 00:03:37.219 --> 00:03:40.139 into undergrad I really wanted to be a doctor 00:03:40.139 --> 00:03:43.319 and I was an EMT at the time and I think I probably 00:03:43.319 --> 00:03:47.439 got exposed to maybe some situations I shouldn't 00:03:47.439 --> 00:03:49.479 have been in an ambulance at a young age and 00:03:49.479 --> 00:03:52.860 decided that was not the career I wanted. And 00:03:52.860 --> 00:03:56.259 so I ended up trying chemistry, but I really 00:03:56.259 --> 00:03:59.500 loved the computer science side. And through 00:03:59.500 --> 00:04:02.759 the 80s and 90s, as computers started to become 00:04:02.759 --> 00:04:05.800 a lot more powerful, people started applying 00:04:05.800 --> 00:04:08.500 it to physics problems, particularly quantum 00:04:08.500 --> 00:04:10.400 mechanical problems, which are challenging to 00:04:10.400 --> 00:04:12.560 solve. Classical mechanics problems are also 00:04:12.560 --> 00:04:15.400 tough to solve, but we're a little bit more tractable. 00:04:16.399 --> 00:04:19.379 And, you know, there was a set of professors 00:04:19.379 --> 00:04:23.779 who I really admired who were emerging in the 00:04:23.779 --> 00:04:25.560 space. And I had the opportunity to work with 00:04:25.560 --> 00:04:28.199 one when I was getting my undergrad as a summer 00:04:28.199 --> 00:04:31.199 student, Fritz Schaefer, who was the, I think 00:04:31.199 --> 00:04:32.860 at the time, the sixth most cited chemist in 00:04:32.860 --> 00:04:35.819 the world. And I ended up working with him again 00:04:35.819 --> 00:04:38.920 for my PhD. And it just combined a lot of things 00:04:38.920 --> 00:04:42.180 that I loved. I was really energized by the work, 00:04:42.259 --> 00:04:44.600 and I think that's kind of the base requirement 00:04:44.600 --> 00:04:47.920 for getting a PhD. And on the quantum mechanics 00:04:47.920 --> 00:04:50.019 side, we're talking about orbits of electrons 00:04:50.019 --> 00:04:53.160 around the nucleus, right? You nailed it. Okay. 00:04:53.420 --> 00:04:56.819 Yeah, you nailed it, which is a fun problem. 00:04:57.439 --> 00:05:03.399 You kind of guess the position of a bunch of 00:05:03.399 --> 00:05:05.800 these electrons and then solve them iteratively, 00:05:05.860 --> 00:05:09.779 and these calculations take... um you know up 00:05:09.779 --> 00:05:12.420 to a month to run if you wanted to to get really 00:05:12.420 --> 00:05:15.220 accurate data and i think the claim to fame for 00:05:15.220 --> 00:05:17.180 a lot of folks you know when i was getting my 00:05:17.180 --> 00:05:20.100 phd was you could start to get to the accuracy 00:05:20.100 --> 00:05:22.740 or even beat the accuracy of what you could achieve 00:05:22.740 --> 00:05:25.399 with a laser in a lab so you could start to overturn 00:05:25.399 --> 00:05:28.720 experimental results if you perhaps found a different 00:05:28.720 --> 00:05:35.279 finding wow wow it um you might My physics background 00:05:35.279 --> 00:05:40.139 is just as old as I am, if that makes any sense. 00:05:40.279 --> 00:05:42.160 I don't remember a lot from those days. I'm not 00:05:42.160 --> 00:05:47.220 touching that. I was always fascinated by the 00:05:47.220 --> 00:05:51.360 fact that you had equations that showed that, 00:05:51.420 --> 00:05:53.360 well, okay, these work as long as you think of 00:05:53.360 --> 00:05:55.579 it as a particle and it's in a specific position. 00:05:56.259 --> 00:05:58.800 These equations work if you think of it as a 00:05:58.800 --> 00:06:00.540 wave, but we really don't know exactly where 00:06:00.540 --> 00:06:05.879 it is. It's kind of somewhere here. And the dichotomy 00:06:05.879 --> 00:06:09.579 of trying to wrap your head around some of that, 00:06:09.680 --> 00:06:11.740 especially in the quantum world, once we get 00:06:11.740 --> 00:06:13.860 into the spooky action at a distance stuff, it 00:06:13.860 --> 00:06:17.680 just makes my head explode. It also violates 00:06:17.680 --> 00:06:19.160 a lot of the things. I know this is probably 00:06:19.160 --> 00:06:21.300 not the intent of the podcast, but it probably 00:06:21.300 --> 00:06:23.639 violates a lot of the things that you memorize 00:06:23.639 --> 00:06:27.100 as a child. You know, you take two magnets and 00:06:27.100 --> 00:06:29.560 you learn that opposites attract. And then you 00:06:29.560 --> 00:06:31.779 get into school and you learn that there's protons 00:06:31.779 --> 00:06:33.800 in the center and electrons on the outside, but 00:06:33.800 --> 00:06:36.910 they never come together. And we just memorize 00:06:36.910 --> 00:06:39.589 it and move forward. And a lot of that is explained 00:06:39.589 --> 00:06:42.889 by quantum mechanical principles and things that 00:06:42.889 --> 00:06:45.569 are a little bit harder to teach an eight year 00:06:45.569 --> 00:06:49.370 old. And so it's really fun as an adult to kind 00:06:49.370 --> 00:06:51.730 of dig into that and see how it really works. 00:06:52.189 --> 00:06:55.810 What's so ironic is that in some respects, it 00:06:55.810 --> 00:06:58.430 underlines, John, the comment that we had just, 00:06:58.509 --> 00:07:02.279 I think, earlier this week. And that is the closer 00:07:02.279 --> 00:07:05.360 you get to a subject, the more everything is 00:07:05.360 --> 00:07:08.199 a matter of opinion. It's like when you're way 00:07:08.199 --> 00:07:11.459 up here, it's all facts and everyone agrees. 00:07:12.379 --> 00:07:15.579 And then the more you learn, it's like, well. 00:07:16.440 --> 00:07:20.160 It depends. When you say way up here, I'm imagining 00:07:20.160 --> 00:07:22.160 you with your arms up because I can't see you, 00:07:22.220 --> 00:07:24.839 so I suspect that's what you're doing. That's 00:07:24.839 --> 00:07:26.860 right. That's exactly what I was doing. You must 00:07:26.860 --> 00:07:28.759 be Italian in a past life, Dave, because you 00:07:28.759 --> 00:07:30.399 talk more with your hands than I do, and I'm 00:07:30.399 --> 00:07:33.560 Italian. Yeah. The motion camera, I can give 00:07:33.560 --> 00:07:35.639 people motion sickness sometimes when those cameras 00:07:35.639 --> 00:07:40.180 are moving around, for sure. So, Jay, digital 00:07:40.180 --> 00:07:43.660 transformation, help us define what that means 00:07:43.660 --> 00:07:47.759 and how you got into that. Digital transformation, 00:07:48.019 --> 00:07:51.040 again, I feel like I'm entering a field that's 00:07:51.040 --> 00:07:54.660 been born and reborn a number of times, similar 00:07:54.660 --> 00:07:58.240 to what we just talked about, the quantum computing. 00:07:58.720 --> 00:08:02.379 And there's been different versions and generations 00:08:02.379 --> 00:08:04.740 of it, whether you think we're on manufacturing 00:08:04.740 --> 00:08:08.360 4 dot oh or what have you. But really, it's the application 00:08:08.360 --> 00:08:15.040 of logic instead of human processes. What occurred 00:08:15.040 --> 00:08:17.579 was originally these distributed control systems 00:08:17.579 --> 00:08:21.300 that allowed you to control devices based on 00:08:21.300 --> 00:08:24.620 an input and then the proliferation of those 00:08:24.620 --> 00:08:27.019 across manufacturing environments that allowed 00:08:27.019 --> 00:08:30.639 you to start to automate things and remove manual 00:08:30.639 --> 00:08:35.259 sensors to now where we're probably in the 80s 00:08:35.259 --> 00:08:37.960 and 90s, the emergence of advanced process control 00:08:37.960 --> 00:08:41.500 predominantly out of oil and gas that led to 00:08:42.350 --> 00:08:45.909 the ability to run a nearly lights out factory. 00:08:46.009 --> 00:08:49.450 Probably when you go to see distillation facilities 00:08:49.450 --> 00:08:52.110 and the like and the oil and gas, you don't see 00:08:52.110 --> 00:08:55.629 a lot of people running around. And now in the 00:08:55.629 --> 00:08:59.049 future, there's a huge link to taking that with 00:08:59.049 --> 00:09:00.850 the transactional information that you have with 00:09:00.850 --> 00:09:02.789 supply planning, getting that into your automated 00:09:02.789 --> 00:09:07.629 systems at the control level. The combination 00:09:07.629 --> 00:09:10.250 of both of those would be state -of -the -art 00:09:10.250 --> 00:09:13.750 where we are today. And in large part, folks 00:09:13.750 --> 00:09:15.330 who are doing digital manufacturing or digital 00:09:15.330 --> 00:09:17.549 transformations are guiding companies through 00:09:17.549 --> 00:09:21.330 a path of that automation that's as accretive 00:09:21.330 --> 00:09:23.730 as possible. In some places, it makes a lot of 00:09:23.730 --> 00:09:26.429 sense to automate everything, especially when... 00:09:27.019 --> 00:09:28.720 you know your margins are slim and the materials 00:09:28.720 --> 00:09:32.039 are very heavy or toxic etc in other cases it 00:09:32.039 --> 00:09:34.600 might make sense to to actually keep a lot of 00:09:34.600 --> 00:09:37.940 manual processes until labor or regulations impose 00:09:37.940 --> 00:09:40.340 something different did you do some consulting 00:09:40.340 --> 00:09:42.659 around that before you you got into it you know 00:09:42.659 --> 00:09:45.299 you know got went to work for someone full -time 00:09:45.299 --> 00:09:48.539 around that to kind of hone that skill yeah i 00:09:49.080 --> 00:09:51.139 I've been lucky to learn from a lot of gifted 00:09:51.139 --> 00:09:54.820 folks. You know, at McKinsey, I did a lot of 00:09:54.820 --> 00:09:56.879 consulting in this space, you know, at Kimberly 00:09:56.879 --> 00:10:01.340 Clark, you know, and now at McCain. In many ways, 00:10:01.460 --> 00:10:04.879 it's a return to the 90s, you know, when I was 00:10:04.879 --> 00:10:06.500 working with a lot of metals and mining clients 00:10:06.500 --> 00:10:08.259 where they actually had a lot of automation. 00:10:08.980 --> 00:10:10.899 Mining tends to be, you know, boom and bust. 00:10:11.159 --> 00:10:14.039 And so as a result, some of those automated controls 00:10:14.039 --> 00:10:16.019 were turned off or weren't calibrated. They went 00:10:16.019 --> 00:10:18.059 to a very manual environment. And then, you know, 00:10:18.059 --> 00:10:21.580 they were kind of going back to the very automated 00:10:21.580 --> 00:10:23.700 facilities that they had originally intended. 00:10:24.620 --> 00:10:27.039 And that's where I learned, you know, that's 00:10:27.039 --> 00:10:29.470 where I cut my teeth, so to speak. And then it 00:10:29.470 --> 00:10:32.309 was a great combination with a lot of the computing 00:10:32.309 --> 00:10:35.490 that I had done in my PhD and as a faculty member, 00:10:35.629 --> 00:10:39.409 because you just really combine this proliferation 00:10:39.409 --> 00:10:42.190 of cheap computers with automated controls and 00:10:42.190 --> 00:10:45.429 manufacturing. It's interesting. You mentioned 00:10:45.429 --> 00:10:47.889 earlier, right, the whole, you know, on again, 00:10:47.970 --> 00:10:50.710 off again, again, zero clients or thin clients 00:10:50.710 --> 00:10:53.990 and just the. You know, we are seeing this computing 00:10:53.990 --> 00:10:56.850 pendulum swing back where if you've ever kind 00:10:56.850 --> 00:10:58.830 of heard me in the more public space, I'll talk 00:10:58.830 --> 00:11:01.649 about with this computing pendulum where, you 00:11:01.649 --> 00:11:03.710 know, we centralized compute and grease green 00:11:03.710 --> 00:11:05.870 screens. Then we decentralized compute with compute 00:11:05.870 --> 00:11:08.110 at the edge. Then we recentralized compute with 00:11:08.110 --> 00:11:12.830 VMs. And now we're we're moving it back on prem 00:11:12.830 --> 00:11:16.809 because cloud bills are expensive and we need 00:11:16.809 --> 00:11:19.590 that compute closer because the speed of light 00:11:19.590 --> 00:11:22.460 to wait for a result from a. Even a data center 00:11:22.460 --> 00:11:25.100 down the street is that latency, right? It's 00:11:25.100 --> 00:11:30.320 too high. What's your opinion on – so we kind 00:11:30.320 --> 00:11:32.200 of opened with the digital transformation item. 00:11:32.600 --> 00:11:36.519 What's your opinion on especially companies that 00:11:36.519 --> 00:11:39.139 have sort of been down the journey, but a lot 00:11:39.139 --> 00:11:42.539 of that transformation is about moving as many 00:11:42.539 --> 00:11:46.360 applications out as they can, and now suddenly 00:11:46.360 --> 00:11:50.470 they're seeing, oh, my gosh, do I have to – Am 00:11:50.470 --> 00:11:53.169 I now suddenly having to move this stuff back 00:11:53.169 --> 00:11:55.649 on -prem because I've gotten rid of a lot of 00:11:55.649 --> 00:12:00.029 those competencies? Where's your head at on that, 00:12:00.129 --> 00:12:03.169 again, that pendulum for the broader market? 00:12:03.610 --> 00:12:07.590 Yeah, maybe comment too to say even in some things 00:12:07.590 --> 00:12:10.529 that seem like they're quite nascent, we're already 00:12:10.529 --> 00:12:13.490 seeing that pendulum from large language models 00:12:13.490 --> 00:12:16.789 to small language models, you know, going into 00:12:16.789 --> 00:12:19.899 the cloud -based. APIs to bringing small language 00:12:19.899 --> 00:12:23.440 models on -prem or in our cell phones. And so 00:12:23.440 --> 00:12:27.500 I think that if we learn from that history, then 00:12:27.500 --> 00:12:30.740 for most companies, they should maintain some 00:12:30.740 --> 00:12:34.980 level of flexibility. And that's really the key 00:12:34.980 --> 00:12:38.340 is what do you need from a data and timescale 00:12:38.340 --> 00:12:40.740 piece that should be at the edge? And then how 00:12:40.740 --> 00:12:42.799 do you properly leverage the cloud so you're 00:12:42.799 --> 00:12:44.340 not locked into something that's challenging 00:12:44.340 --> 00:12:48.580 to migrate in the future? That's it. I think 00:12:48.580 --> 00:12:51.039 that's true for a lot of the data that's produced 00:12:51.039 --> 00:12:55.299 at the edge. We stream up a portion of that data 00:12:55.299 --> 00:12:58.440 to a cloud. How do you keep it in native format 00:12:58.440 --> 00:13:01.620 so that you're not constantly staffing teams 00:13:01.620 --> 00:13:04.639 to do data translation or data migration, which 00:13:04.639 --> 00:13:08.600 tends to be not very accretive, right? And then 00:13:08.600 --> 00:13:12.049 how do you make sure you're constantly... increasing 00:13:12.049 --> 00:13:14.570 the value of the data as you're transforming 00:13:14.570 --> 00:13:17.549 it rather than just replicating it is a big piece 00:13:17.549 --> 00:13:21.529 as well. For sure. I think we see a lot of folks 00:13:21.529 --> 00:13:24.509 who take data in the edge, they migrate it to 00:13:24.509 --> 00:13:26.730 the cloud or elsewhere where they do a lot of 00:13:26.730 --> 00:13:29.029 data engineering, but you have a lot of users 00:13:29.029 --> 00:13:31.850 at the edge that aren't benefiting from it. And 00:13:31.850 --> 00:13:33.889 so, you know, I really think for organizations, 00:13:33.950 --> 00:13:36.350 you have to ask the question of where is the 00:13:36.350 --> 00:13:38.710 right place to do that engineering in the long 00:13:38.710 --> 00:13:40.590 term and how do you make sure the most of your 00:13:40.590 --> 00:13:43.149 citizen data scientists benefit from it? And 00:13:43.149 --> 00:13:45.629 I think a lot of organizations originally thought 00:13:45.629 --> 00:13:49.419 they needed data scientists. in a central location 00:13:49.419 --> 00:13:52.059 working in the cloud, not realizing a lot of 00:13:52.059 --> 00:13:55.279 the latent skills of folks who did PLC programming 00:13:55.279 --> 00:13:58.039 or database programming inside of facilities 00:13:58.039 --> 00:14:01.480 that actually were really, really gifted with 00:14:01.480 --> 00:14:04.360 some training as citizen data scientists. You 00:14:04.360 --> 00:14:07.679 kind of tipped off the... I didn't say the two 00:14:07.679 --> 00:14:09.580 -letter acronym everybody wants to talk about 00:14:09.580 --> 00:14:11.460 today, AI, but you said large language model, 00:14:11.580 --> 00:14:13.580 small language model. So I figured, okay, we'll 00:14:13.580 --> 00:14:15.860 dive in there. Everybody drinks. No, I'm kidding. 00:14:16.039 --> 00:14:18.600 Yeah, exactly. We often, Dave and I often joke 00:14:18.600 --> 00:14:22.360 that AI should be a drinking game. Every time 00:14:22.360 --> 00:14:24.139 we sit during a presentation, we always go, oh, 00:14:24.299 --> 00:14:25.139 take a drink. How long would we get into the 00:14:25.139 --> 00:14:27.620 podcast until somebody says the two -letter word? 00:14:27.740 --> 00:14:29.919 We'd be incoherent walking around. That's right. 00:14:29.960 --> 00:14:34.309 Yes. So my question really is this. How do you 00:14:34.309 --> 00:14:37.730 see AI improving either what you're currently 00:14:37.730 --> 00:14:40.990 doing or into the future? Do you see it becoming 00:14:40.990 --> 00:14:47.769 a big component in that infrastructure and that 00:14:47.769 --> 00:14:51.690 solution? What are your thoughts on that? Yeah, 00:14:51.730 --> 00:14:55.309 I think we'll continue to see the growth is the 00:14:55.309 --> 00:14:56.889 short answer, and I think everybody would be 00:14:56.889 --> 00:14:59.289 aligned with that. When you talk about it from 00:14:59.289 --> 00:15:03.220 a control standpoint, I think the use of LLMs 00:15:03.220 --> 00:15:06.879 or SLMs or any large language models introduces 00:15:06.879 --> 00:15:09.720 something that's probabilistic instead of deterministic. 00:15:09.779 --> 00:15:12.700 And for a lot of folks, introducing something 00:15:12.700 --> 00:15:15.620 probabilistic into a control regime that has 00:15:15.620 --> 00:15:19.080 very discrete limits and potentially implications 00:15:19.080 --> 00:15:24.659 for human health or sort of large tragedies if 00:15:24.659 --> 00:15:26.679 you're thinking about melting steel furnaces 00:15:26.679 --> 00:15:31.659 and so forth. In that sense, I think we'll have 00:15:31.659 --> 00:15:34.460 to consider more stringent controls, but it's 00:15:34.460 --> 00:15:36.279 not like we've never done it before. I mean, 00:15:36.279 --> 00:15:38.480 weather forecasts are inherently probabilistic. 00:15:38.679 --> 00:15:40.899 We've integrated those in operating facilities 00:15:40.899 --> 00:15:43.200 for a long time. They're used extensively in 00:15:43.200 --> 00:15:46.879 refrigeration to do wet bulb control for ammonia 00:15:46.879 --> 00:15:50.440 refrigeration and the like. And so it's thinking 00:15:50.440 --> 00:15:54.070 about how do you take the output. and bring it 00:15:54.070 --> 00:15:55.929 into controls more as a disturbance variable, 00:15:56.129 --> 00:15:58.730 perhaps, that just perturbs an existing control 00:15:58.730 --> 00:16:02.450 loop. So that's, I think, where the industry 00:16:02.450 --> 00:16:04.509 will move, and there's a lot of players who are 00:16:04.509 --> 00:16:07.769 thinking about that. And then on the outside, 00:16:08.230 --> 00:16:12.470 how do we accelerate a person? That's where use 00:16:12.470 --> 00:16:15.230 as an agent or the use of agents is going to 00:16:15.230 --> 00:16:18.590 be. more and more critical because you can ask 00:16:18.590 --> 00:16:20.970 the LLM to generate a lot of things for you and 00:16:20.970 --> 00:16:22.889 then rather than having to take the output and 00:16:22.889 --> 00:16:25.649 implement it yourself or go query a database 00:16:25.649 --> 00:16:28.629 or implement the ladder logic, you can use agents 00:16:28.629 --> 00:16:31.470 with some read -write permission to do it for 00:16:31.470 --> 00:16:34.490 you. And those two areas are probably going to 00:16:34.490 --> 00:16:37.509 be the most powerful for folks in the manufacturing 00:16:37.509 --> 00:16:40.769 environment. And so it's exciting. It's really 00:16:40.769 --> 00:16:45.519 exciting to see. Yeah, I agree. thing that continues 00:16:45.519 --> 00:16:48.379 to come up well a lot of things but the one thing 00:16:48.379 --> 00:16:50.220 that continues to come up specifically around 00:16:50.220 --> 00:16:55.360 the excitement is this concept that look you 00:16:55.360 --> 00:16:59.320 know we've already had this problem finding humans 00:16:59.320 --> 00:17:01.759 to work in some of these areas whether it because 00:17:01.759 --> 00:17:05.519 of safety or similar so in some respects this 00:17:05.519 --> 00:17:08.700 has come along at the right time in that you 00:17:08.700 --> 00:17:11.400 know maybe You know, if I designed a process 00:17:11.400 --> 00:17:14.420 to require 10 humans and now I can only find 00:17:14.420 --> 00:17:17.579 half that, do I have the ability potentially 00:17:17.579 --> 00:17:20.799 in some respects to automate maybe where I wasn't 00:17:20.799 --> 00:17:26.059 able to automate before when, generally speaking, 00:17:26.259 --> 00:17:28.900 it always was, unless it's specifically exactly 00:17:28.900 --> 00:17:31.680 repeatable, you're not going to be able to get 00:17:31.680 --> 00:17:35.740 a non -human to do it. Yet, the corollary to 00:17:35.740 --> 00:17:39.589 that is there's concerns around, obviously, To 00:17:39.589 --> 00:17:41.769 your point earlier around lights out manufacturing, 00:17:42.109 --> 00:17:44.549 the really bad thing about lights out manufacturing 00:17:44.549 --> 00:17:47.490 is it doesn't require any workers or very little. 00:17:47.769 --> 00:17:51.349 So what's your feeling, Jay, on someone that 00:17:51.349 --> 00:17:54.970 is in the industry? And it's not that they don't 00:17:54.970 --> 00:17:57.789 like the promise. The concern with them is that, 00:17:57.809 --> 00:18:00.910 you know, you can say all you want about skilling 00:18:00.910 --> 00:18:03.599 up and leveling up our. factory workers, but 00:18:03.599 --> 00:18:06.640 there comes a point where there's a whole bucket 00:18:06.640 --> 00:18:08.720 ton of them that you're probably not going to 00:18:08.720 --> 00:18:11.319 be able to level up. What's your feeling there? 00:18:13.200 --> 00:18:16.779 I'll give you three takes on it. The first is 00:18:16.779 --> 00:18:19.380 for a lot of the manufacturing industries, they 00:18:19.380 --> 00:18:20.960 tend to grow with the size of the middle class 00:18:20.960 --> 00:18:24.400 or the population in general. You know, copper, 00:18:24.559 --> 00:18:27.480 steel, aluminum, you know, even food industries 00:18:27.480 --> 00:18:30.299 as well. So the need for workers, so long as 00:18:30.299 --> 00:18:32.359 manufacturing stays in a particular country, 00:18:32.440 --> 00:18:34.940 will continue to grow. And you'll always have 00:18:34.940 --> 00:18:37.160 a need to probably expand your manufacturing 00:18:37.160 --> 00:18:41.319 base. The other piece I would say is that the 00:18:41.319 --> 00:18:44.180 type of work that folks are going to do will 00:18:44.180 --> 00:18:49.140 probably bifurcate. You know, scan and copy, 00:18:49.319 --> 00:18:52.099 you know, or... you type of work where you're 00:18:52.099 --> 00:18:53.759 putting two excel sheets next to each other in 00:18:53.759 --> 00:18:56.079 a factory and just copying data over is it will 00:18:56.079 --> 00:18:59.680 probably go away but but you have a lot of work 00:18:59.680 --> 00:19:01.640 that has to occur on the maintenance side that's 00:19:01.640 --> 00:19:03.440 very challenging to automate and does require 00:19:03.440 --> 00:19:05.740 advanced skills and electrical and automation 00:19:05.740 --> 00:19:08.019 yeah exactly those trades are never going to 00:19:08.019 --> 00:19:10.720 go away you're you're not going to have an ai 00:19:10.720 --> 00:19:13.599 make a wiring change in a panel for example that's 00:19:13.599 --> 00:19:16.259 it yep And so you'll see, you know, the folks 00:19:16.259 --> 00:19:19.059 in the control room, some of perhaps the more 00:19:19.059 --> 00:19:21.299 junior folks who are not educated in a particular 00:19:21.299 --> 00:19:23.259 trade would have to go maybe more towards a trade 00:19:23.259 --> 00:19:26.140 route or they level up. And when you think about 00:19:26.140 --> 00:19:29.000 it in the scheme of the control pyramid, you 00:19:29.000 --> 00:19:31.740 know, you're not going to use AI probably in 00:19:31.740 --> 00:19:34.519 the bottom three rungs, which would be your distributed 00:19:34.519 --> 00:19:36.920 control systems, your advanced distributed control 00:19:36.920 --> 00:19:38.420 systems, which would be like your feed forward 00:19:38.420 --> 00:19:40.920 loops, dead time compensation loops. you're not 00:19:40.920 --> 00:19:42.460 going to use it in model predictive control. 00:19:42.640 --> 00:19:44.640 But when you start to get to real -time optimization, 00:19:44.960 --> 00:19:47.299 scheduling of the factory, et cetera, that's 00:19:47.299 --> 00:19:49.740 when AI is going to matter because you're taking 00:19:49.740 --> 00:19:53.140 something temporal and then in something transactional 00:19:53.140 --> 00:19:55.380 and maybe a disturbance such as a weather or 00:19:55.380 --> 00:19:57.500 a customer order, and you're combining those 00:19:57.500 --> 00:19:58.940 three things together. And that's inherently 00:19:58.940 --> 00:20:01.119 quite challenging for a human to do. We have 00:20:01.119 --> 00:20:03.099 a lot of tools like MES systems that help with 00:20:03.099 --> 00:20:05.980 us, but that's where AI probably will have the 00:20:05.980 --> 00:20:09.500 most acceleration. And so those folks may level 00:20:09.500 --> 00:20:12.250 up. and reduce perhaps in the number, but the 00:20:12.250 --> 00:20:14.210 trades that we just talked about will continue 00:20:14.210 --> 00:20:16.430 to be quite important. And then of course, as 00:20:16.430 --> 00:20:19.230 factories continue to expand to meet population 00:20:19.230 --> 00:20:21.930 demands, just the inherent number of people will 00:20:21.930 --> 00:20:25.609 continue to grow. Love that. Fantastic. Fantastic. 00:20:25.869 --> 00:20:27.490 Yeah. Humans are always going to be necessary. 00:20:27.750 --> 00:20:32.569 That's, I mean, every paradigm shift in this, 00:20:32.829 --> 00:20:35.430 you know, with this type of impact, everybody's 00:20:35.430 --> 00:20:38.539 afraid of losing jobs, losing workers. pushing 00:20:38.539 --> 00:20:40.539 people out. But I think it just, to your point, 00:20:40.660 --> 00:20:45.000 leveling them up or focusing on the things that 00:20:45.000 --> 00:20:46.799 you're still going to need a human for is really 00:20:46.799 --> 00:20:49.140 what's going to make the difference. For sure. 00:20:49.339 --> 00:20:52.019 Yeah. And I think what changes, what makes people 00:20:52.019 --> 00:20:55.359 a little more fearful with this is that this 00:20:55.359 --> 00:20:59.180 appears to impact all jobs. Whereas, like, you 00:20:59.180 --> 00:21:03.420 know, take the classic automobile replacing the, 00:21:03.480 --> 00:21:06.519 you know, your horse farms and your blacksmiths. 00:21:06.599 --> 00:21:08.500 Don't get me wrong. There was a lot of people 00:21:08.500 --> 00:21:10.539 involved there, but it was a it was a single 00:21:10.539 --> 00:21:13.319 trade or a single kind of slice. I mean, even 00:21:13.319 --> 00:21:16.079 computers, you could argue, were talking about 00:21:16.079 --> 00:21:19.819 specific finance and well, initially the actual 00:21:19.819 --> 00:21:22.619 human computers. Right. Whereas this feels very 00:21:22.619 --> 00:21:27.950 broad. And deep, not just wide, but broad, wide, 00:21:27.970 --> 00:21:30.609 and deep. And I think that's where the concerns 00:21:30.609 --> 00:21:35.869 come from, for sure. Jay, especially from what 00:21:35.869 --> 00:21:39.349 we've seen, discuss the things that you've talked 00:21:39.349 --> 00:21:42.160 about. My belief, by the way, is that you've 00:21:42.160 --> 00:21:45.960 been very forward on this recognition of the 00:21:45.960 --> 00:21:49.339 compute coming back on prem. And I think you 00:21:49.339 --> 00:21:52.339 were even in front of it before kind of AI suddenly 00:21:52.339 --> 00:21:56.000 made it sexy. I mean, you were a huge proponent 00:21:56.000 --> 00:22:00.769 of edge based processing. even potentially in 00:22:00.769 --> 00:22:04.589 industries and for reasons where many of your 00:22:04.589 --> 00:22:08.670 other peers out there may not have done it. Not 00:22:08.670 --> 00:22:10.990 because it's necessarily necessary, but because 00:22:10.990 --> 00:22:15.609 you needed that flexibility. If someone out there 00:22:15.609 --> 00:22:19.690 is looking at the AI wave and they're saying 00:22:19.690 --> 00:22:24.630 to themselves, we just don't see it in a timely 00:22:24.630 --> 00:22:27.690 fashion. We're going to kind of continue operating 00:22:27.690 --> 00:22:30.809 without having that flexibility of moving things 00:22:30.809 --> 00:22:34.009 back. I'm pretty sure you would probably say, 00:22:34.109 --> 00:22:38.589 you know, you're crazy. Here's why. Talk a little 00:22:38.589 --> 00:22:43.250 bit about, without using AI as the reason, why 00:22:43.250 --> 00:22:45.410 you may want to make sure you have some of that 00:22:45.410 --> 00:22:49.670 on -prem based on your experience. Yeah, on -prem 00:22:49.670 --> 00:22:53.980 for us. It allows us to collect data at time 00:22:53.980 --> 00:22:56.339 scales that are much faster than you would be 00:22:56.339 --> 00:22:58.480 able to do with the cloud. So as you're moving 00:22:58.480 --> 00:23:02.380 to checking batches of pills to every single 00:23:02.380 --> 00:23:04.940 pill, that sort of hardware acceleration at the 00:23:04.940 --> 00:23:07.039 edge is quite important. I think that's piece 00:23:07.039 --> 00:23:09.900 number one. There's a resiliency and security 00:23:09.900 --> 00:23:12.380 that you inherently get from locking down your 00:23:12.380 --> 00:23:15.700 OT environment. That's piece number two. It continues 00:23:15.700 --> 00:23:18.799 to be quite inexpensive to have compute and storage 00:23:18.799 --> 00:23:20.980 at the edge, which is maybe point number three. 00:23:21.220 --> 00:23:24.140 And then I think there's a general piece of source 00:23:24.140 --> 00:23:26.859 control that you should have in general. In water 00:23:26.859 --> 00:23:29.240 treatment, if you combine all of the polluted 00:23:29.240 --> 00:23:31.059 water together, then treat it, it's never as 00:23:31.059 --> 00:23:34.559 effective as if you can treat a particular chemical 00:23:34.559 --> 00:23:38.579 or pollutant at its source before it gets combined. 00:23:43.599 --> 00:23:46.279 remove any of the anomalies at the place where 00:23:46.279 --> 00:23:48.720 it's created or close to it at the edge. Then 00:23:48.720 --> 00:23:50.660 when you send it on to use in the cloud, you 00:23:50.660 --> 00:23:52.339 get much better use of those compute resources, 00:23:52.680 --> 00:23:56.839 etc. And I don't think that compute in the cloud 00:23:56.839 --> 00:23:59.019 is necessarily a bad thing. I think, as I said 00:23:59.019 --> 00:24:01.960 before, folks need a hybrid environment. It's 00:24:01.960 --> 00:24:04.720 just recognizing that the compute... in the cloud 00:24:04.720 --> 00:24:06.900 especially when you're using spot instances etc 00:24:06.900 --> 00:24:09.559 tends to be the cheapest way to use it whereas 00:24:09.559 --> 00:24:12.420 you know reserved instances forever and putting 00:24:12.420 --> 00:24:14.740 in lots of disaster recovery data in the cloud 00:24:14.740 --> 00:24:17.380 tend to be quite expensive and so i think just 00:24:17.380 --> 00:24:19.900 from a cost position you should figure out what 00:24:19.900 --> 00:24:24.119 the best utility is And I think a lot of that 00:24:24.119 --> 00:24:26.140 functionality is returning to the edge, as you 00:24:26.140 --> 00:24:28.680 mentioned. I know Cisco's made a push in that 00:24:28.680 --> 00:24:31.759 space, Rockwell as well, where we're starting 00:24:31.759 --> 00:24:34.019 to see the ability to take things like a Kubernetes 00:24:34.019 --> 00:24:35.819 cluster, bring it to the edge, and I'm getting 00:24:35.819 --> 00:24:37.660 a lot of the functionality I had in the cloud 00:24:37.660 --> 00:24:39.920 to start managing services at the edge, which 00:24:39.920 --> 00:24:43.700 is exciting. And then the network fabrics have 00:24:43.700 --> 00:24:46.420 become tremendously fast and inexpensive. And 00:24:46.420 --> 00:24:48.619 so bringing that timescale that I just talked 00:24:48.619 --> 00:24:52.509 about at the edge is great. I'll take maybe a 00:24:52.509 --> 00:24:55.950 step back and say, you know, why is where does 00:24:55.950 --> 00:24:58.470 AI make the most sense in manufacturing is often 00:24:58.470 --> 00:25:01.410 when you're trying to manage a timescale that's 00:25:01.410 --> 00:25:04.750 either very, you know, very big or has a lot 00:25:04.750 --> 00:25:06.710 of data points in it. You're trying to manage 00:25:06.710 --> 00:25:09.609 dimensionality that's very large or you're trying 00:25:09.609 --> 00:25:11.589 to measure manage something that's very nonlinear. 00:25:11.849 --> 00:25:14.450 And a lot of those things are happening at the 00:25:14.450 --> 00:25:17.029 edge and the latency and the cost of bringing 00:25:17.029 --> 00:25:19.190 it to the cloud to manage it and then come back 00:25:19.190 --> 00:25:22.690 can sometimes be prohibitive as well. And also 00:25:22.690 --> 00:25:25.529 the cloud can make things, I mean, it's not to 00:25:25.529 --> 00:25:27.230 say that we don't want to defend, you know, bringing 00:25:27.230 --> 00:25:29.490 things back on -prem or go towards cloud, but, 00:25:29.490 --> 00:25:32.009 you know, on your comment about cloud for a second, 00:25:32.069 --> 00:25:35.970 you can make changes faster to, like, let's say 00:25:35.970 --> 00:25:38.450 you need, you know, an additional dynamo or something 00:25:38.450 --> 00:25:41.589 to spin up. You just turn it on. Whereas if you 00:25:41.589 --> 00:25:44.049 have to put something in on -prem, you've got 00:25:44.049 --> 00:25:45.950 to get the hardware, you've got to provision 00:25:45.950 --> 00:25:47.829 it, you've got to set it up correctly, and then 00:25:47.829 --> 00:25:50.730 you can... you can apply it into the system. 00:25:50.809 --> 00:25:53.329 So it's much more flexible from that standpoint. 00:25:53.450 --> 00:25:57.170 But again, it has to be the right solution for 00:25:57.170 --> 00:25:59.950 the right application, I guess is really what 00:25:59.950 --> 00:26:02.789 I'm saying. Yep. And it's easy to be unconstrained 00:26:02.789 --> 00:26:04.509 in the cloud too. I mean, you can burn a Ferrari 00:26:04.509 --> 00:26:08.470 a week in costs if you decided to. Yeah, probably. 00:26:09.190 --> 00:26:12.349 Probably. Didn't think of it that way. I like 00:26:12.349 --> 00:26:17.619 that though. I've seen it happen. Right. It sounded 00:26:17.619 --> 00:26:20.200 good until somebody got the cloud bill and they 00:26:20.200 --> 00:26:23.700 went, whoa, wait a minute. Yeah. Not a recognition 00:26:23.700 --> 00:26:26.980 of the amount of DASD resources that it requires. 00:26:27.950 --> 00:26:30.329 No, but it is super important, and it begs a 00:26:30.329 --> 00:26:33.089 lot of questions of how do we potentially extend, 00:26:33.170 --> 00:26:34.950 now we've done the IT environment to the cloud, 00:26:35.049 --> 00:26:36.890 how do we extend the OT environment to the cloud 00:26:36.890 --> 00:26:39.569 in a secure way? And a lot of those protocols 00:26:39.569 --> 00:26:41.569 and principles are starting to catch up, and 00:26:41.569 --> 00:26:43.609 you've already seen some reference architectures 00:26:43.609 --> 00:26:46.710 published by Cisco and others. So it's exciting 00:26:46.710 --> 00:26:50.329 to see that you could imbue really fast compute 00:26:50.329 --> 00:26:53.769 inside OT environment using cloud resources. 00:26:54.859 --> 00:26:57.059 Jay, I do want to talk about – this is actually 00:26:57.059 --> 00:26:58.799 a good segue into the question I was going to 00:26:58.799 --> 00:27:03.460 ask, and that is part of this is the digital 00:27:03.460 --> 00:27:05.720 transformation that we're talking about. Obviously, 00:27:05.799 --> 00:27:09.440 there's this huge people component. And ignoring 00:27:09.440 --> 00:27:13.140 the concerns around AI and up -leveling and what 00:27:13.140 --> 00:27:19.420 have you, a lot of this is shifting jobs, roles, 00:27:19.799 --> 00:27:22.960 the way folks have done business in the past. 00:27:23.579 --> 00:27:26.559 And that's one of the largest resistance to the 00:27:26.559 --> 00:27:29.519 transformation journey that we've seen generally, 00:27:29.640 --> 00:27:33.940 is the concern about shifting the work and what's 00:27:33.940 --> 00:27:37.160 being done. Talk maybe a little bit about, based 00:27:37.160 --> 00:27:41.859 upon your history up to this point, what successes, 00:27:42.079 --> 00:27:45.599 what's worked to get folks, the people folks, 00:27:45.759 --> 00:27:48.420 to go from, oh man, that's a lot of change, to, 00:27:48.440 --> 00:27:51.480 you know what, we can do that. Yeah, I think 00:27:51.480 --> 00:27:53.160 it's been exciting to see the people journey 00:27:53.160 --> 00:27:56.460 in general over the last, you know, call it decade, 00:27:56.500 --> 00:27:59.599 decade and a half. You know, first, the adoption 00:27:59.599 --> 00:28:03.460 of agile and agile flavors, you know, across 00:28:03.460 --> 00:28:06.559 not just technology companies, but in manufacturing 00:28:06.559 --> 00:28:09.759 and, you know, marketing and the like, I think 00:28:09.759 --> 00:28:13.420 has led to folks feeling like their job has a 00:28:13.420 --> 00:28:16.099 bit more flexibility to begin with. And I think 00:28:16.099 --> 00:28:18.720 a lot of those agile roles, you know, the scrum 00:28:18.720 --> 00:28:21.339 masters, product owners, developers and the like, 00:28:21.440 --> 00:28:25.500 now with AI and, you know, a lot of the digital 00:28:25.500 --> 00:28:28.140 upskilling that's occurred have become even more 00:28:28.140 --> 00:28:31.099 gray, where you now have product owners that 00:28:31.099 --> 00:28:33.220 say, well, I can reach in and check what might 00:28:33.220 --> 00:28:35.859 be bad at a code or I can go look at GitHub a 00:28:35.859 --> 00:28:38.240 little bit more easily than I could before because 00:28:38.240 --> 00:28:41.579 I have all these tools available to me. By and 00:28:41.579 --> 00:28:44.039 large, I've observed a lot of excitement in that 00:28:44.039 --> 00:28:46.440 space where people are saying my job is a little 00:28:46.440 --> 00:28:49.059 easier now and it was somewhat more challenging 00:28:49.059 --> 00:28:51.119 before and now I actually can keep track of the 00:28:51.119 --> 00:28:52.700 velocity of the team a lot better and I have 00:28:52.700 --> 00:28:54.799 a heartbeat on the team and I feel like we're 00:28:54.799 --> 00:28:57.559 generating better code and faster code and where 00:28:57.559 --> 00:29:00.700 velocities are increasing inside of our development 00:29:00.700 --> 00:29:04.220 sprints. And so I think there's a lot of excitement. 00:29:04.779 --> 00:29:08.220 At the same time, as you denoted, there's... 00:29:08.569 --> 00:29:11.569 There's a lot of need for upskilling and where 00:29:11.569 --> 00:29:14.470 there was these huge campaigns to introduce agile. 00:29:14.950 --> 00:29:18.289 Now we're finding these huge campaigns to introduce 00:29:18.289 --> 00:29:21.650 digital tooling, the use of LLMs, particularly 00:29:21.650 --> 00:29:25.309 RAG models as chatbots and the like to help folks 00:29:25.309 --> 00:29:27.490 accelerate their work. And my hope is at the 00:29:27.490 --> 00:29:29.069 end of this, it just means that there's a better. 00:29:29.690 --> 00:29:31.710 worker satisfaction or employee satisfaction, 00:29:32.109 --> 00:29:33.849 that they're able to do their job a little bit 00:29:33.849 --> 00:29:36.369 easier and with a little bit more efficiency 00:29:36.369 --> 00:29:39.369 than before. But absolutely, that mindset is 00:29:39.369 --> 00:29:42.490 critical. Pivoting to a little bit different 00:29:42.490 --> 00:29:45.759 take. What types of things do you see in the 00:29:45.759 --> 00:29:48.660 future that maybe people may not have on their 00:29:48.660 --> 00:29:51.960 radar today that are coming down the pipe that 00:29:51.960 --> 00:29:55.019 might make a difference in digital transformation? 00:29:55.380 --> 00:29:57.279 And we've talked a lot about AI and its impact, 00:29:57.500 --> 00:29:59.519 but what other things should we be looking for? 00:30:00.519 --> 00:30:02.920 Well, there's things I'm excited about. I don't 00:30:02.920 --> 00:30:04.660 know if everybody should be excited about them. 00:30:04.779 --> 00:30:09.200 Okay. But we've been pretty excited about... 00:30:09.839 --> 00:30:13.019 The use of some traditional engineering tools 00:30:13.019 --> 00:30:15.759 that were used, I would say, like at the beginning 00:30:15.759 --> 00:30:18.759 of a build of a factory or beginning of a beginning 00:30:18.759 --> 00:30:21.259 of a build of a unit operation to bringing those 00:30:21.259 --> 00:30:24.500 into real time. So an example of that is things 00:30:24.500 --> 00:30:26.920 like computational fluid dynamics, which was 00:30:26.920 --> 00:30:29.339 traditionally like run it for a week, get all 00:30:29.339 --> 00:30:31.700 of the simulations, you know, either airflow 00:30:31.700 --> 00:30:34.160 or fluid flow, use that to design something at 00:30:34.160 --> 00:30:36.240 steady state and then kind of shelve it. And 00:30:36.240 --> 00:30:38.099 then if we make any improvements or we need to 00:30:38.099 --> 00:30:39.380 adjust the factory in the future, we'll. get 00:30:39.380 --> 00:30:42.880 back to it what we're excited about is now you 00:30:42.880 --> 00:30:45.700 know folks like nvidia and others have have accelerated 00:30:45.700 --> 00:30:49.160 so much the compute engines inside of these cfd 00:30:49.160 --> 00:30:52.980 simulations that they're now on a time scale 00:30:52.980 --> 00:30:56.160 that we can use them for real -time control and 00:30:56.160 --> 00:30:59.819 you're getting a near perfect answer which is 00:30:59.819 --> 00:31:02.819 super super cool especially when you end up in 00:31:02.819 --> 00:31:06.279 your chemical space and you're doing both mixing 00:31:06.279 --> 00:31:08.839 distillations and the like you know for us in 00:31:08.839 --> 00:31:10.940 food manufacturing where you have cooling and 00:31:10.940 --> 00:31:14.779 a lot of these large air flows it's it's really 00:31:14.779 --> 00:31:17.680 a cool unlock that's being driven by the fact 00:31:17.680 --> 00:31:20.339 that you have this orthogonal industry either 00:31:20.339 --> 00:31:22.900 graphics computing or computing in general that's 00:31:22.900 --> 00:31:26.680 just continuing to accelerate at the speed that 00:31:26.680 --> 00:31:28.740 it is and it's benefiting all of this other stuff 00:31:28.740 --> 00:31:32.980 that that uses the compute so we're excited about 00:31:32.980 --> 00:31:36.369 stuff like that I think from the network side, 00:31:36.690 --> 00:31:39.029 as I told you before, extending the OT environment 00:31:39.029 --> 00:31:42.690 into the cloud and getting cloud resources inside 00:31:42.690 --> 00:31:44.789 of it, I think is exciting for us, as well as 00:31:44.789 --> 00:31:47.970 this growth of compute inside that environment, 00:31:48.109 --> 00:31:50.609 even in switches and other hardware that now 00:31:50.609 --> 00:31:52.789 have hypervisors that you can place things inside, 00:31:53.029 --> 00:31:58.259 I think is a really exciting development. The 00:31:58.259 --> 00:32:00.880 real -time automation and control that we discussed, 00:32:01.079 --> 00:32:04.079 you know, for us picking a KPI, like how do we 00:32:04.079 --> 00:32:08.019 expose, you know, 50 % less set points to an 00:32:08.019 --> 00:32:10.140 operator so that they can do their job without 00:32:10.140 --> 00:32:12.160 having to memorize all the interactions from 00:32:12.160 --> 00:32:14.700 different devices, motors, pumps, and fans, I 00:32:14.700 --> 00:32:18.119 think is really exciting too. So, you know, we 00:32:18.119 --> 00:32:20.660 look at those trends. We try to figure out exactly 00:32:20.660 --> 00:32:23.460 where we would deploy them globally and where 00:32:23.460 --> 00:32:26.160 they're accretive, we chase them relentlessly. 00:32:27.230 --> 00:32:30.509 That's awesome. Yeah, it's really good to talk 00:32:30.509 --> 00:32:32.650 about the current state, but I always get excited 00:32:32.650 --> 00:32:34.470 about learning more about the future state as 00:32:34.470 --> 00:32:36.809 well. So thanks for sharing that. Of course. 00:32:36.950 --> 00:32:41.150 Yeah. And a lot of those heavy compute things 00:32:41.150 --> 00:32:44.490 exist all the way from water treatment to, as 00:32:44.490 --> 00:32:47.390 I described, to freezing and frying and the like. 00:32:47.549 --> 00:32:51.089 And it's just really cool to see how much compute 00:32:51.089 --> 00:32:53.970 we have available to us at the edge. If you think 00:32:53.970 --> 00:32:56.849 about it, there's so many things out there potentially 00:32:56.849 --> 00:33:01.109 that we wouldn't have even attempted because 00:33:01.109 --> 00:33:04.529 we would have needed answers in real time. The 00:33:04.529 --> 00:33:07.490 compute wasn't available to actually perform 00:33:07.490 --> 00:33:10.269 it. And now that we're going to have it available, 00:33:10.450 --> 00:33:12.849 I think what you're going to see is incredibly 00:33:12.849 --> 00:33:17.710 creative uses for that compute because it's generic 00:33:17.710 --> 00:33:20.130 and cheap. Generally speaking, it's going to 00:33:20.130 --> 00:33:23.349 be generic and cheap. It's getting cheaper. And 00:33:23.349 --> 00:33:25.190 it's getting cheaper. Yeah. So if you think about, 00:33:25.210 --> 00:33:28.589 look at it in places where we think of compute, 00:33:28.789 --> 00:33:32.170 and that is in the office environment, in the 00:33:32.170 --> 00:33:35.670 back office environment, in places other than 00:33:35.670 --> 00:33:38.009 the process environment. What this has really 00:33:38.009 --> 00:33:40.650 done is it's going to give us compute in a place 00:33:40.650 --> 00:33:43.289 where no one really started thinking about using 00:33:43.289 --> 00:33:48.500 it. combinations of skill sets coming out of 00:33:48.500 --> 00:33:52.660 academia and schooling and trades where you know 00:33:52.660 --> 00:33:56.690 you're taking a person that was exposed to the 00:33:56.690 --> 00:33:59.789 control space, the process space, the network 00:33:59.789 --> 00:34:02.930 space, more IT space. And so they're going to 00:34:02.930 --> 00:34:05.569 have the ability to go, wait, why don't we try 00:34:05.569 --> 00:34:08.369 this? When before there would have been no chance 00:34:08.369 --> 00:34:10.989 because there was no place to do it. They weren't 00:34:10.989 --> 00:34:13.230 in the correct department. It wouldn't have been 00:34:13.230 --> 00:34:15.630 available anyway. It would have been offline. 00:34:15.829 --> 00:34:18.869 And what good does looking at it offline do? 00:34:19.599 --> 00:34:21.780 You nailed it. And I would say plus with the 00:34:21.780 --> 00:34:25.980 combination of being able to code things so fast. 00:34:26.039 --> 00:34:30.280 And, you know, I kind of liken it to the next 00:34:30.280 --> 00:34:33.079 abstraction, which is, you know, you needed assembly 00:34:33.079 --> 00:34:36.460 at one point. You needed C++ at one point. You 00:34:36.460 --> 00:34:40.000 know, Python is now an abstraction. over a C 00:34:40.000 --> 00:34:42.360 compiler, it gets compiled down to C and C++. 00:34:42.559 --> 00:34:44.760 You use Python, and that was a huge accelerant 00:34:44.760 --> 00:34:48.139 for a while. And now I just tell the LLM to give 00:34:48.139 --> 00:34:50.659 me the Python, which then compiles into C++ or 00:34:50.659 --> 00:34:53.940 C, which then runs into machine code. And so 00:34:53.940 --> 00:34:57.739 it's like that next abstraction that's just helped 00:34:57.739 --> 00:35:02.800 us pick a language to talk in that's even easier 00:35:02.800 --> 00:35:05.800 to describe and requires less training. And so 00:35:05.800 --> 00:35:09.260 that's the real fun part for me. Right. Well, 00:35:09.320 --> 00:35:12.860 you nailed it on that, Jay, is that think about 00:35:12.860 --> 00:35:15.219 the number of people out there that have had 00:35:15.219 --> 00:35:18.780 a really good idea, but they don't. And they 00:35:18.780 --> 00:35:22.320 don't code. Yeah, exactly. Sorry, I'm stealing 00:35:22.320 --> 00:35:24.980 your thunder, Dave. No, but that's it. You said 00:35:24.980 --> 00:35:26.400 it. You summarized it a lot better than I did. 00:35:26.519 --> 00:35:30.639 They don't code. So all of that corporate kind 00:35:30.639 --> 00:35:34.579 of bureaucracy that you need to be able to line 00:35:34.579 --> 00:35:37.179 up in order for you to create that, you don't 00:35:37.179 --> 00:35:40.590 need it. Yeah. And it's pushed this skill set 00:35:40.590 --> 00:35:43.070 to folks who, again, have to be more specialized 00:35:43.070 --> 00:35:47.110 or develop large architectures for enterprise 00:35:47.110 --> 00:35:49.110 software. That's really, in some ways, a little 00:35:49.110 --> 00:35:51.489 bit challenging for somebody to code into unless 00:35:51.489 --> 00:35:54.329 they have that computer science background. And 00:35:54.329 --> 00:35:56.369 so, you know, it's pushing folks to think more 00:35:56.369 --> 00:35:58.309 like architects because they can program all 00:35:58.309 --> 00:36:01.130 the boilerplate so quickly. They can now take 00:36:01.130 --> 00:36:04.369 time to think about the overall enterprise. And 00:36:04.369 --> 00:36:06.289 so it's exciting to see that transformation. 00:36:08.010 --> 00:36:12.150 Jay, back to some of the other transformation 00:36:12.150 --> 00:36:16.150 discussions we had, do you find that different 00:36:16.150 --> 00:36:20.300 areas of the world are... more accepting of some 00:36:20.300 --> 00:36:22.440 of the digital transformation strategies you 00:36:22.440 --> 00:36:25.840 guys develop? Or is there more resistance in 00:36:25.840 --> 00:36:28.420 one place versus another? And I'm not trying 00:36:28.420 --> 00:36:31.800 to get you to throw countries under the bus. 00:36:32.539 --> 00:36:38.199 But just curious if the U .S. is much more difficult 00:36:38.199 --> 00:36:41.380 to push something through on or get people on 00:36:41.380 --> 00:36:45.820 board on versus the EU or Asia, what's that typically 00:36:45.820 --> 00:36:49.909 look like? I think in general, what a lot of 00:36:49.909 --> 00:36:53.969 folks have observed in Asian markets is the flexibility 00:36:53.969 --> 00:36:57.650 of the customer, I would say, which is if you 00:36:57.650 --> 00:37:01.650 introduce 30, 40 new products into China or India 00:37:01.650 --> 00:37:04.610 or something like that, people are used to seeing 00:37:04.610 --> 00:37:06.489 new things. They're used to companies trying 00:37:06.489 --> 00:37:11.989 out new products in the marketplace. I think 00:37:11.989 --> 00:37:16.590 in the US or EU, it tends to be maybe a more 00:37:16.590 --> 00:37:18.789 slower product development. And I think that 00:37:18.789 --> 00:37:21.469 trickles down to the folks that work for those 00:37:21.469 --> 00:37:24.510 companies in those regions of how fast they're 00:37:24.510 --> 00:37:27.469 used to moving or how flexible or how innovative 00:37:27.469 --> 00:37:31.550 they're used to being. You see companies, I would 00:37:31.550 --> 00:37:35.030 say, the insular nature in some ways of Brazil 00:37:35.030 --> 00:37:38.190 after a lot of the tariffs that they've had a 00:37:38.190 --> 00:37:40.820 little protectionist tariffs. also has created 00:37:40.820 --> 00:37:44.260 a pretty innovative space inside Brazil where 00:37:44.260 --> 00:37:47.019 you don't have a lot of folks competing potentially 00:37:47.019 --> 00:37:49.199 from other companies because it's challenging 00:37:49.199 --> 00:37:53.280 to move product into that market. And as a result, 00:37:53.360 --> 00:37:55.760 you walk around Brazil and you feel like this 00:37:55.760 --> 00:37:59.059 entrepreneurial spirit. And so I think that's 00:37:59.059 --> 00:38:01.780 exciting. I think there's also places where labor 00:38:01.780 --> 00:38:05.449 has become quite expensive. the US, Australia, 00:38:05.550 --> 00:38:08.510 and the like, where AI is pushing something different, 00:38:08.590 --> 00:38:12.070 which is the labor as a cost of goods sold, as 00:38:12.070 --> 00:38:13.969 a percentage of cost of goods sold, and folks 00:38:13.969 --> 00:38:19.010 realizing that with different geopolitical environments, 00:38:19.409 --> 00:38:22.889 both tariffs in the US and elsewhere, the need 00:38:22.889 --> 00:38:25.349 to compete, and now all of a sudden having to 00:38:25.349 --> 00:38:26.909 look at that, and the digital transformation 00:38:26.909 --> 00:38:29.590 becomes a mechanism for that. I think in the 00:38:29.590 --> 00:38:33.409 EU, transformation for a while was based on energy 00:38:33.409 --> 00:38:38.309 and sustainability. And the use of AI to use 00:38:38.309 --> 00:38:40.489 less power, especially inside heavy industries, 00:38:40.530 --> 00:38:44.449 is a massive mover of both cost and the ability 00:38:44.449 --> 00:38:47.349 to meet regulations. And so I think that's a 00:38:47.349 --> 00:38:50.489 long way of saying every region, country has 00:38:50.489 --> 00:38:53.909 their dimension that's driving innovation. And 00:38:53.909 --> 00:38:56.070 some of those are closer to a digital transformation 00:38:56.070 --> 00:38:59.670 than others. But as a global company, you kind 00:38:59.670 --> 00:39:02.179 of have to adapt to all of them. Well, that was 00:39:02.179 --> 00:39:06.139 a very eloquent answer and it didn't, you did 00:39:06.139 --> 00:39:08.860 everybody justice. So I think, I think from, 00:39:08.920 --> 00:39:11.639 from, you know, no one's under the bus. I think 00:39:11.639 --> 00:39:14.320 that was a great way to, to, to explain it. Thanks 00:39:14.320 --> 00:39:16.880 for that. Sure. That was, that was, that was 00:39:16.880 --> 00:39:18.619 great. I mean, you, you had actually brought 00:39:18.619 --> 00:39:21.199 something up there and I'm kind of curious whether 00:39:21.199 --> 00:39:24.820 you've seen different geos tackle it differently. 00:39:25.739 --> 00:39:30.820 And that is obviously the AI wave needing additional 00:39:30.820 --> 00:39:35.340 utility power and the tradeoff between how much 00:39:35.340 --> 00:39:38.820 do we – I'm not going to say everybody's throwing 00:39:38.820 --> 00:39:40.900 their sustainability goals out the window. Instead, 00:39:41.000 --> 00:39:43.019 what they're looking at is, okay, what's the 00:39:43.019 --> 00:39:45.900 right balance so that we're getting a payoff 00:39:45.900 --> 00:39:48.409 in such – And oh, by the way, there's only so 00:39:48.409 --> 00:39:50.050 much of it to go around. I mean, that's why you 00:39:50.050 --> 00:39:52.829 see the utility contracts from Constellation 00:39:52.829 --> 00:39:56.230 and others trying to lock up a nuke plant for 00:39:56.230 --> 00:40:00.250 20 years kind of scenario. Are you seeing different 00:40:00.250 --> 00:40:04.090 strategies there depending upon where you're 00:40:04.090 --> 00:40:07.530 looking in the world, Jay? Or is everyone kind 00:40:07.530 --> 00:40:12.250 of looking at it similarly? Yeah, I'll start 00:40:12.250 --> 00:40:15.539 with maybe a comparison. first and then answer 00:40:15.539 --> 00:40:18.380 that question, which is there was a time when 00:40:18.380 --> 00:40:20.760 Bitcoin mining took the amount of electricity 00:40:20.760 --> 00:40:22.599 of actual mining and there were headlines that 00:40:22.599 --> 00:40:26.320 said that. And it was in large part because the 00:40:26.320 --> 00:40:28.380 algorithms behind it were based on proof of work. 00:40:28.440 --> 00:40:30.380 And then when they migrated to proof of stake, 00:40:30.519 --> 00:40:35.360 it was 99 % less energy usage. And I think the 00:40:35.360 --> 00:40:38.420 similar thing is happening in AI, which is, which 00:40:38.420 --> 00:40:41.539 was, we're starting, you know, folks have said 00:40:41.539 --> 00:40:44.139 this and I, far be it for me to. to say what 00:40:44.139 --> 00:40:46.440 it should be, because I'm not an expert in this 00:40:46.440 --> 00:40:49.340 space. But we're reaching some sort of asymptote 00:40:49.340 --> 00:40:52.840 in LLMs, and then the specialization in SLMs 00:40:52.840 --> 00:40:55.340 has started to become a lot more important. And 00:40:55.340 --> 00:41:00.019 the energy use is a fraction. And that transition, 00:41:00.059 --> 00:41:03.380 I think, is going to really define our energy 00:41:03.380 --> 00:41:05.900 use inside these data centers in the next five, 00:41:05.980 --> 00:41:08.260 10 years. So I would say that's piece number 00:41:08.260 --> 00:41:11.659 one, is just to maintain some reality check of 00:41:11.659 --> 00:41:14.119 how much energy we're going to need. At the same 00:41:14.119 --> 00:41:17.360 time, yeah, there's real power demands that are 00:41:17.360 --> 00:41:19.300 going to flex our grid that was already somewhat 00:41:19.300 --> 00:41:22.179 stressed in certain areas. That's going to be 00:41:22.179 --> 00:41:25.139 a challenge, especially with a lot of the sustainability 00:41:25.139 --> 00:41:29.559 guidelines that we have. As battery technologies 00:41:29.559 --> 00:41:34.460 continue to improve, I think we'll see some flattening 00:41:34.460 --> 00:41:38.730 of the grid demand, which will help. Inexpensive 00:41:38.730 --> 00:41:41.570 natural gas inside the U .S. has been an incredible 00:41:41.570 --> 00:41:44.289 asset, particularly for the use of turbines. 00:41:44.630 --> 00:41:47.250 And then people have started to use those turbines 00:41:47.250 --> 00:41:49.670 not just for electricity, but the exhaust from 00:41:49.670 --> 00:41:52.150 them to power absorption chillers. And therefore, 00:41:52.269 --> 00:41:53.710 they're getting the cooling from the turbine 00:41:53.710 --> 00:41:56.570 as well. So it's been a very fun push to see 00:41:56.570 --> 00:41:59.409 the efficiency of a lot of those turbines increase 00:41:59.409 --> 00:42:01.630 as well, which has helped a huge number of industries, 00:42:01.829 --> 00:42:05.699 which is great. And then there's going to need 00:42:05.699 --> 00:42:09.659 to be this consideration of data center use versus 00:42:09.659 --> 00:42:11.480 use at the edge, like we talked about before. 00:42:12.039 --> 00:42:14.119 How much do you put in the iPhone versus how 00:42:14.119 --> 00:42:16.659 much do you put in the data center? And that 00:42:16.659 --> 00:42:19.239 balance, I think, still needs to right -size 00:42:19.239 --> 00:42:23.719 as well. What are your thoughts on nuclear? Nuclear 00:42:23.719 --> 00:42:27.619 has always been a very promising technology and 00:42:27.619 --> 00:42:30.820 green technology. It's great to see with regulation 00:42:30.820 --> 00:42:36.579 some of the... small reactors maybe coming to 00:42:36.579 --> 00:42:40.280 bear in the next five, 10 years. I think there's 00:42:40.280 --> 00:42:45.039 also a big piece of geothermal has had its resurgence 00:42:45.039 --> 00:42:47.480 as well, deep geothermal, which is going to allow 00:42:47.480 --> 00:42:50.000 us to decarbonize and provide a lot of potential 00:42:50.000 --> 00:42:52.760 power. I think we've seen heat pumps and high 00:42:52.760 --> 00:42:54.820 temperature heat pumps in particular emerge, 00:42:55.219 --> 00:42:58.239 especially in Europe, that'll start to be a little 00:42:58.239 --> 00:43:00.059 bit more mature in the next couple of years, 00:43:00.219 --> 00:43:03.699 which is exciting. So there's a lot of um you 00:43:03.699 --> 00:43:06.059 know renewable and renewable adjacent technologies 00:43:06.059 --> 00:43:09.079 that are really cost accretive that are coming 00:43:09.079 --> 00:43:11.179 on the market that's gonna that's gonna yield 00:43:11.179 --> 00:43:14.539 a lot i think in this space awesome yeah i'm 00:43:14.539 --> 00:43:18.039 still i'm i'm i'm not against the nuclear energy 00:43:18.039 --> 00:43:20.440 push by any means it just takes so long to build 00:43:20.440 --> 00:43:24.980 a nuke it i just wonder if we're ever you know 00:43:24.980 --> 00:43:27.159 and another four years you'll there could be 00:43:27.159 --> 00:43:28.579 another president in the white house and then 00:43:28.579 --> 00:43:31.440 things could change so you just never know Certainly 00:43:31.440 --> 00:43:35.800 some of the small reactor stuff that potentially 00:43:35.800 --> 00:43:38.599 have the ability to always fail safe, they lose 00:43:38.599 --> 00:43:41.059 cooling and they simply shut down in a safe manner. 00:43:41.860 --> 00:43:47.980 The idea of small kilowatt size is really interesting 00:43:47.980 --> 00:43:51.619 to me. But yeah, the question basically is, can 00:43:51.619 --> 00:43:55.360 we get over the regulatory hurdles enough in 00:43:55.360 --> 00:43:57.199 order to streamline it to the point that it's 00:43:57.199 --> 00:44:00.800 useful? Because, man, we need it. Like we're 00:44:00.800 --> 00:44:02.900 to the point now where we really need it. We 00:44:02.900 --> 00:44:05.320 need the you know, we need it from every angle 00:44:05.320 --> 00:44:07.599 that we possibly can. And, yeah, there's a great 00:44:07.599 --> 00:44:10.239 answer. And by the way, I don't know what I was 00:44:10.239 --> 00:44:13.039 expecting on your your power utility answer on 00:44:13.039 --> 00:44:16.639 the front, but I got way more than. So thanks 00:44:16.639 --> 00:44:19.579 for that, Jay. And I'm going to steal some of 00:44:19.579 --> 00:44:21.900 that because that was very, very good. I really 00:44:21.900 --> 00:44:24.699 enjoyed that. Telling you, right. Well, we have 00:44:24.699 --> 00:44:27.900 about five minutes left, so we want to try and 00:44:27.900 --> 00:44:30.559 wrap up here. I would love to just get kind of 00:44:30.559 --> 00:44:34.219 an open -ended answer from you on, pretend you're, 00:44:34.219 --> 00:44:37.599 again, in front of a mixed audience of IT, OT 00:44:37.599 --> 00:44:41.420 professionals. They love hearing from other customers 00:44:41.420 --> 00:44:43.579 what they're seeing, the trends that are out 00:44:43.579 --> 00:44:46.440 there, what's kind of currently hot and not. 00:44:46.840 --> 00:44:49.719 If you had to, and I'm not even going to specifically 00:44:49.719 --> 00:44:54.039 mention the topic, if you had to give Five things, 00:44:54.199 --> 00:44:59.300 five pieces of advice, Jay's opinion on if you 00:44:59.300 --> 00:45:03.340 do these five things, Mr. and Mrs. Customer, 00:45:03.559 --> 00:45:06.699 you're going to save your company yourself a 00:45:06.699 --> 00:45:09.400 whole heck of a lot of time and money. What would 00:45:09.400 --> 00:45:13.239 those five things be, Jay? This is a great question. 00:45:13.980 --> 00:45:16.440 Dave's good at this. Yeah, I should have prepped 00:45:16.440 --> 00:45:21.460 for this one. I think. Especially in heavy industry, 00:45:21.579 --> 00:45:23.860 going back and investing in the control pyramid 00:45:23.860 --> 00:45:27.840 seems to be a very accretive thing to do. A lot 00:45:27.840 --> 00:45:31.500 of companies are adopting that stance and the 00:45:31.500 --> 00:45:34.400 internal rate of return of those sort of projects 00:45:34.400 --> 00:45:39.679 tends to be 50 % plus, I think, which is an absolute 00:45:39.679 --> 00:45:44.719 no -brainer. I think what was hot and is now 00:45:44.719 --> 00:45:47.780 returning to hot is a lot of the automated vehicles. 00:45:48.380 --> 00:45:50.940 I think for a while that maybe cooled and is 00:45:50.940 --> 00:45:54.199 returning. So, you know, part and parcel with 00:45:54.199 --> 00:45:58.119 the control system automation is, you know, maybe 00:45:58.119 --> 00:46:01.179 a return to seeing if that's valuable. So I would 00:46:01.179 --> 00:46:04.320 say maybe that's topic one. Topic two, we've 00:46:04.320 --> 00:46:06.840 already covered ad nauseum. And, you know, again, 00:46:06.920 --> 00:46:08.579 if we were drinking for every time we said AI, 00:46:08.780 --> 00:46:12.219 I think the podcast would be over. But, you know, 00:46:12.219 --> 00:46:14.800 adopting AI in a sensible way. Right. Which is 00:46:14.800 --> 00:46:17.420 your your company is probably not going to make 00:46:17.420 --> 00:46:21.699 LLMs and SLMs, but you should be making knowledge 00:46:21.699 --> 00:46:24.360 graphs for rag models. You should be making agents. 00:46:24.579 --> 00:46:26.780 And that's where a lot of where your IP is going 00:46:26.780 --> 00:46:28.739 to be sitting anyway. So that's a good investment. 00:46:29.280 --> 00:46:31.280 So I would say maybe that's theme number two. 00:46:32.139 --> 00:46:36.019 Boy, five is quite a few themes. I'll stop at 00:46:36.019 --> 00:46:40.219 three. But, you know, I think the other piece 00:46:40.219 --> 00:46:43.869 we talked about is. Feeling like you can have 00:46:43.869 --> 00:46:47.369 the flexibility to build a strategy that doesn't 00:46:47.369 --> 00:46:51.889 have to encompass every keyword. Which is, what 00:46:51.889 --> 00:46:55.849 is the strategy that is truly going to give you 00:46:55.849 --> 00:46:59.409 value at every stage? Yeah. You don't necessarily 00:46:59.409 --> 00:47:01.449 need to say, I'm moving one application to the 00:47:01.449 --> 00:47:03.250 cloud, therefore all applications need to be 00:47:03.250 --> 00:47:05.800 moved to the cloud. And I think that... There 00:47:05.800 --> 00:47:08.400 are a lot of leaders who maybe feel constrained 00:47:08.400 --> 00:47:11.480 that if they say cloud native, that now everything 00:47:11.480 --> 00:47:14.300 has to be that way or vice versa. And I think 00:47:14.300 --> 00:47:16.940 we're now at a place where most of these things 00:47:16.940 --> 00:47:19.800 are so commoditized that you can choose your 00:47:19.800 --> 00:47:22.380 path and choose the path that's best for the 00:47:22.380 --> 00:47:27.599 company. Yeah, be comfortable with 80 -20. Be 00:47:27.599 --> 00:47:30.599 comfortable with nuance. You don't have to go 00:47:30.599 --> 00:47:34.500 ridiculous one way. It's somewhere in the middle. 00:47:34.539 --> 00:47:38.280 So many of the things I think we talked about 00:47:38.280 --> 00:47:41.280 fall into that rule. And don't be afraid to fail. 00:47:41.380 --> 00:47:43.599 I mean, if it doesn't work, you know, pivot and 00:47:43.599 --> 00:47:47.480 make some changes and fix it. Right. And for 00:47:47.480 --> 00:47:50.400 a lot of companies, that includes defining a 00:47:50.400 --> 00:47:53.639 value path beforehand so that you know if it 00:47:53.639 --> 00:47:56.639 didn't fail. If it failed, I should say, which 00:47:56.639 --> 00:47:58.800 is, I think, sometimes challenging where, you 00:47:58.800 --> 00:48:00.780 know, maybe we want to use less energy, so forth. 00:48:00.920 --> 00:48:03.019 But what are the true secondary and tertiary 00:48:03.019 --> 00:48:05.500 KPIs that I expect to move? And then as soon 00:48:05.500 --> 00:48:08.139 as one of them doesn't move, what steering committee 00:48:08.139 --> 00:48:09.860 is going to say, great, should we continue the 00:48:09.860 --> 00:48:14.260 investment or not? Yes. Love it. All fun stuff. 00:48:15.340 --> 00:48:18.139 And I've got to thank both of you for having 00:48:18.139 --> 00:48:20.619 me on. I really appreciate the... Well, it's 00:48:20.619 --> 00:48:22.079 great to catch up with both of you, but being 00:48:22.079 --> 00:48:25.940 part of your podcast journey is really amazing. 00:48:26.019 --> 00:48:27.960 So thank you. Thank you so much. You're absolutely 00:48:27.960 --> 00:48:30.500 welcome, sir. Yes, thank you. Thank you for agreeing 00:48:30.500 --> 00:48:33.179 to be on it. Thank you for managing to fit us 00:48:33.179 --> 00:48:37.219 around your crazy travel schedule, Jay. And yeah, 00:48:37.239 --> 00:48:39.039 I think with that, we'll go ahead and wrap it 00:48:39.039 --> 00:48:41.480 up. Thank you, sir. Thank you, everybody, for 00:48:41.480 --> 00:48:44.079 listening. Good afternoon, good evening, good 00:48:44.079 --> 00:48:47.000 morning, and good night from two guys in a PLC. 00:48:47.500 --> 00:48:48.880 Thanks, everybody. Bye now.