The Evolution of Generative AI


In this episode, Raymond sits down with Bill Kleyman to learn more about his life and career in the data center industry and discuss the evolution of generative AI.

Le contenu des ressources est affiché en anglais uniquement.

After 15 years in the technology space, Bill has the chance to see quite the endless advancements. Today, he works with leaders in digital infrastructure to help build a more sustainable and inclusive future and support an ever-connected digital society. He is also a contributing editor to leading industry publications, including Data Center Frontier, Data Center Knowledge, ITPro Today, and InformationWeek.

Before they get to the main topic, Bill tells the fascinating story of his journey to America, fleeing a dangerous radioactive Kiev after the Chernobyl fall out in the late 1980s. They then discuss the incredible advancements we’re seeing in generative AI, and how that can help in a myriad of ways across data centers, technology, health and everything in between.

Read the full episode transcript below:

Raymond Hawkins: Welcome again to another edition of Not Your Father’s Data Centre. I am your host, Raymond Hawkins, and today you’ll not hear me talk much at all because we have the great Bill Kleyman with us. He is going to man the microphone, which will make my job really easy. He’s awesome to talk to, a friend of the programme. We love him at Compass. He is far more entertaining to listen to than I am. After I get through his illustrious array of titles, we are going to hand him the microphone and let him talk about his background and how he got in the data centre business. So a board member and advisor at Neuro, AFCOM Data Centre World programme chair and a contributing editor at both Data Centre Frontier and Data Centre Knowledge. So if you read about the data centre business, Bill’s had something to do with it. If you care about our standards in AFCOM, Bill’s had something to do with it and Neuro, he’s all over that place.

So Bill, welcome to Not Your Father’s Data Centre. You know what my kids say every time they hear that? They go, but dad, it is my dad’s data centre. So there’s only two people that this show title doesn’t work for.

Bill Kleyman: Yeah, I think my five-year-old doesn’t quite know what a data centre is yet. She just understands all the blinking lights and whatnot. But you know what, I love it and thanks for having me here, Raymond. I really appreciate it.

Raymond Hawkins: So we’re super happy you joined us and if you’re willing, we’re going to let you dive right into telling us about you because I think you’re pretty well known in our industry and our space and lots of folks know a lot about you, but I don’t think they know everything about you. So let’s go all the way back to the beginning. Where’d you grow up? Where’d you go to school? How’d you get into the data centre business? Let’s dig into some of that and see where that takes us.

Bill Kleyman: Absolutely. Well thank you so much and everybody listening, it’s a pleasure to be here on this show. The data centre industry, just technology in general is something that’s near and dear to my heart. It’s actually funny if you go on my LinkedIn profile, I think literally the byline or the very top is like, « I love technology ». And then I go on to tell you why I like it. There’s a really good book out there. It’s an Amazon bestseller that I had a chance to co-author. It’s called Greener Data that outlines the beginning of, or the nexus of my career and what I’ve been doing. I got a nice little bulky chapter in that book.

So okay, everybody, we’re going to be friends really quick. My name is not really Bill, it kind of is. I was born, my name is Vitali and I was born in Kyiv, Ukraine and I came to the United States in the early nineties.

And the reason I’m telling you this is because Raymond and I had a little quick little chat. He’s like, Bill, in the same level of excitement, Bill, how did you get into this data centre industry? How did it all start? And what’s fascinating is that my next actually started when I was very young in Soviet Ukraine and Kyiv and my brother, he used to compete in… There’s no other way to put it, telegraph competitions, the switch and everything. And he would let me sit on his lap and he would put these big headphones on my head and he taught me numbers and some letters and I was like six or seven years old and I would be able to communicate with people all over then-Soviet Russian, obviously Ukraine as well. And even at that really sort of young age, I was fascinated, fascinated by how we could bring people together and closer by using technology. And obviously as phones were available, this was in the late eighties.

But it was absolutely fascinating and I really enjoyed the concept and the solutions and just the idea behind using these tools to bring people together and closer. And I carried that excitement on through my high school career where I did a bunch of AP classes, definitely around things like coding and comp sci. And I realised that my passions lie more around physical infrastructure, that design, architecture, engineering and so on. I feel like I’m an anomaly here, Raymond, that I’m telling you this story because I am not a transplant. I started off in the data centre industry. I actually went to a trade school and got a network engineering undergrad and then a master’s in business and then another master’s in information security.

And from graduating with my network engineering undergraduate, I’ve been in the data centre industry, I had a chance to work for the country’s larger Citrix partner for a while, left them as their CTO, I had a chance to spend time in DevOps. Spent four years working with obviously co-location data centre industry with my former company called Switch. And now I get a chance to dive into and work with all of these crazy things that you’re hearing about like ChatGPT, generative AI, large language models.

But what’s really fascinating is that these tools that I’m working with right now, we can talk about them, are things that are democratised for our industry, Raymond, the data centre industry and telecommunications and how we can be capturing the power of these technologies and not just leaving it and relegating into the hyperscaler. So I feel that I’m lucky. I was at a 7×24 conference just recently and I was at a big panel. We were talking about education and young people in our industry and I asked the question, all right everybody, how many in this room are native to the data centre industry? And by the way, everybody listening to this, are you native to the data centre industry or are you a musician or a doctor, accountant? I was the only one that raised his hand because everyone else was in some way adopted.

I’m lucky. I love this industry. I love working with critical infrastructure. I’m absolutely fascinated by everything that we do every single day. And I love, love, love showcasing and sharing with the industry that we are so much more than big buildings and blinking lights. I mean this industry is really cool.

Raymond Hawkins: All right, so I’m sticking with Vitali because I like it and I think I can say it properly. So I’m sticking with Vitali for the rest of this recording. You weaved right through Kyiv and then you went into the Telegraph and then you jumped into the data centre business. Hold on. Kyiv and Ukraine are in the news a little bit for the last year. So let’s back up. How did you get from, because you’re in Chicago now I think, right?

Bill Kleyman: Yeah.

Raymond Hawkins: So there’s a big journey in there that we’d love to hear a little bit about. So get us from six year old Vitali working, communicating via telegraphs to other people and get me to how you’re in the United States. There’s got to be a fun story there, a fascinating story. So let’s hear some of that journey.

Bill Kleyman: I feel like I wish we had more time to talk about some of this because I’m giving you the Cliff’s Notes version of it. Yeah, born in Soviet Ukraine, in Kyiv. And I’d say things started to sort of go downhill when Chernobyl happened in 1986. We actually moved to Crimea for a little while just to try and escape it, making sure that I didn’t have any health issues and so on. And we stayed there for a little while and then went back right around ’88 and ’89. We pretty much saw the writing on the wall that the Soviet government wasn’t going to last very long.

Raymond Hawkins: So, Vitali, for people that aren’t as old as you and me, give them the distance from where you lived to the Chernobyl incident. And I know you moved south down to Crimea, give people a little bit of Eastern European geography.

Bill Kleyman: Pripyat to Kyiv, it’s not that long. It’s probably a 20, 30 minute drive or so.

Raymond Hawkins: So for folks who don’t appreciate this, the largest nuclear accident in our globe’s history was less than an hour from your house. So when you say you moved, you guys were getting out of town?

Bill Kleyman: Yeah, yeah, we definitely. So overall it’s between, it’s about an hour drive, hour and a half, hour drive. I would say Chernobyl’s a little bit further away. It’s probably like 50 to 60 miles from where we were. Still not very far.

Raymond Hawkins: You could, yeah. I just want folks listening to appreciate when you say you guys were, Hey, Chernobyl was an incident. It wasn’t like it was a docuseries that we got to watch 40 years later. This was your family, an hour’s drive away. This accident’s happening in Soviet at the time, Soviet-controlled Ukraine and probably not the best flow of information about the accident.

Bill Kleyman: No, no. We didn’t know what happened effectively. And as you remember, they had a parade, I think literally the next day in the city of Pripyat where somebody on the roof was taking a picture of the green glow of the graphite reactors melting and spewing out iridium and cesium into the-

Raymond Hawkins: Vitali, I’ve read about this, I was not born yet, so I don’t think I can get the past that one. Yes, I remember.

Bill Kleyman: Yeah?

Raymond Hawkins: I couldn’t get that one out with a straight face. « I wasn’t born yet. » Unfortunately, I was.

Bill Kleyman: I appreciate that, you try to make it with a straight face.

Raymond Hawkins: All right, so Chernobyl happens, mom says, we got to go.

Bill Kleyman: Literally mom and dad. Dad actually had to stay back and work a little bit, but me, my mom and my brother and some extended family hopped on a train literally and went to Crimea. And that’s actually where we learned what happened. We had an idea what was going on. It was either the Americans launched the bomb or… Because we were being told to get the heck out of Dodge and we weren’t sure what was going on. But when we got to the south to Crimea, you were able to tune into European radio stations and you could hear, they’re like, okay, something happened in Northern Ukraine right by the Belarus border. And we’re like, okay, we think we know what’s going on here. And that was the nexus behind a lot of the, that we had to get out, not just not out of Dodge, but I get out of the country.

And many folks stayed. We had some family members and some friends definitely stayed back. But when we decided to leave, that was right around the collapse of the Soviet Union and the Soviet government at the time that we announced that we wanted to leave, they took our passports, they tore them up and they called us traitors to the nation because you’re either Russian or you’re nothing, which is, Raymond and everybody listening, not unlike what’s happening to the Ukrainian culture right now by the Russians.

And so we basically had to get rid of everything. We had just a couple of hundred bucks. I had a couple of toys with me, a couple of backpacks, and we were political asylum refugees living in Europe for a while until we finally got a political asylum visa to the United States and came to good old Brighton Beach, Brooklyn in the early nineties and then moved into a little apartment north of Chicago. It was in the city, just the north side. Lived our extended family in one little apartment and literally built everything from scratch.

Raymond Hawkins: And so this is the late eighties, early nineties when you… Early nineties, by the time you get to Chicago?Yes?

Bill Kleyman: Yeah.

Raymond Hawkins: All right, awesome stuff. You won’t hear many data centre podcasts where you are going to get a firsthand account of fleeing Chernobyl. That is pretty amazing stuff. So that’s why I wanted to make sure you got it in there because it’s not just, Hey, we decided to leave Ukraine. This was, we got to get out of Dodge and how it led you to end up being now an American, which we’re so excited about.

Bill Kleyman: And I love being here. This is an absolutely wonderful country. Nowhere else I’d rather be. For those listening, that series Chernobyl that was done, that was very, very well done. Very good perspective, very open and honest. If that’s the one thing that you watch about what happened in my home country to really try to embrace and understand the situation. That’s a good one. That’s a good one. Get a tissue or two already, because we obviously had friends and people that we knew responding to that incident, and some of those folks are going in nothing but a gas mask and paper thin radioactive suit when studies show that in the centre of the radioactive zone, you couldn’t really spend more than 30 to 40 seconds shovelling that stuff off a roof where people were spending more than that. But kind of goes to show you the state of the things back then.

Raymond Hawkins: So I did watch, I can’t remember if it’s Netflix or what it is, but I did watch the whole miniseries and I think the biggest thing that struck me, the lack of respect for human life out of the Soviet Union. I think that’s the thing I left with, is thinking that their lack of appreciation for the safety of their people, for the sanctity of life, that the image of the state and technological advancement and the honouring the country from a perspective of, Hey, look how great we are was more important than the humans that live there. And that to me was probably the biggest impression and the biggest heartbreak watching that whole series.

Bill Kleyman: I mean, it goes to say, there’s a nuclear meltdown and the Russian government decides, Hey, let’s have a parade.

Raymond Hawkins: And let’s tell everybody it’s not our fault. It wasn’t us.

Bill Kleyman: It’s fine. It’s no problem. Don’t worry, green glow is normal.

Raymond Hawkins: Yeah, nothing to see here. Incredible stuff. Well, we don’t get too many folks that got to live through what an incredible part of history and talking to us on the podcast. So grateful to hear that part of your story, Bill. And I like Vitali. I may have a hard time going back to Bill, but-

Bill Kleyman: You’re welcome to do it, Raymond. That’s part of what happened. So when we came to America, if I’d say my name to another American Vitali, people say it just fine, but if they read my name, they say Vitally or Vitality, they make me sound like an herbal supplement. And so my mom, she asked an American friend, she’s like, well, what does Vitali sound like? And they were like, it sounds like William, I guess. And my mom was like, he’s good. We go change now. And so through my American counterparts, it’s Bill in high school I was Billy and obviously William, and then all of my Slavic speaking friends that I still interact with, it’s Vitali. They say Vitalik with a K. Like Vitali Klitschko, the mayor of Kyiv right now.

Raymond Hawkins: Yeah, yeah. Former boxer. Hey, just mayor now.

Bill Kleyman: Former Boxer, him and his former-

Raymond Hawkins: World heavyweight champion, if I’m not mistaken.

Bill Kleyman: Right, Randy.

Raymond Hawkins: Yeah. Not just Boxer. That’s right. Former heavyweight champion. Yeah. Boy, the way I’ve enjoyed some of hiss interviews during this last year and a half and his passion for his homeland and his people and his city.

Bill Kleyman: I feel it.

Raymond Hawkins: Fascinating stuff. I promise we’re going to talk about the data centre business, but I got to ask one more Kyiv, Ukraine, family… So as we chatted earlier, you did mention a fascinating fact about your mom. So folks that listen know I’m a Marine, but I do want to hear you. We got to give props to your mother and we got to talk a little bit about her service in the military, just a little bit. Sprinkle some of that in that. Yeah, got to hear it.

Bill Kleyman: It was in the Soviet army. I honestly, I have not asked her very many details. I know that she was-

Raymond Hawkins: We don’t want to have to kill all our listeners, so nothing top secret here.

Bill Kleyman: I can tell you that she was very good shot « top of her class with a sniper rifle. »

Raymond Hawkins: Very nice. And-

Bill Kleyman: She literally was like, I never missed. I’m like, oh my god, mom, I don’t think I can hear this conversation from my mother. What was interesting is that while she was doing that work, my dad, he was in the Army Corps of Engineer in the Soviet army, Soviet Ukraine army I should say. I knew that he was designing missile silos. What I didn’t know it was for those missiles. Yeah.

Raymond Hawkins: Yeah, those.

Bill Kleyman: And so I had no idea. And honestly I found out a lot about their military service here in the United States when I was in my late twenties, just because I never really asked and they never got a chance to keep their medals, none of their… All of that stuff. Just again, you’re either Russian or you’re nothing. I think my dad maybe got a chance to keep one or two service medals, if I’m not mistaken. Same thing with my mom, but very, very, very little. If anything, they don’t talk about their military career. My mom does maybe less so maybe because of what she did. My dad will talk about his engineering feats all the time though.

Raymond Hawkins: Love the story of your mom, love her background, and love the story of how you guys got to America. So thank you for giving us a little piece of you, Bill. That’s super awesome. Let’s switch gears and let’s talk a little bit about the data centre industry, although it will pale in comparison to your background, which I love and enjoy getting to hear about. Like you highlighted, I was in the systems integrator business for years. I filled buildings full of computers. Walking in and out of data centres. I only ever had one question, Hey, can you tell me when the circuits are going to be provisioned? Because that’s when my servers could plug in and I could get paid because I didn’t get paid until they knew their servers worked. And then 10, almost 11 years ago now, someone said, Hey, have you ever looked into the data centre business?

And I was like, wait, you mean someone that owns those buildings? Like duh. And I got to join our friends at Digital Realty and learn the industry and it has been an incredible 10, 11 years of my life now. And I love to say that we’re really the foundation upon which digital transformation happens. Our industry occasionally, you’ll get an elbow to the ribs about, oh, you guys use a lot of power. You’re not a very green industry. What are you doing?

And I always joke with folks, I say, Hey, hey, grab your phone. Just tell me what on here you’d like to stop doing. Well, we’ll shut it down. You don’t want to have food brought to the house anymore? No problem. Postmates, we’re done with that. Uber Eats, we’re done with that. You don’t want to catch a ride to the airport? No problem. We’ll kill Uber. You don’t want to order plane tickets online? No problem. We’ll kill that, right? You don’t want to watch Netflix. No problem. I mean, it is really the foundation for digital transformation. The stuff that we’ve all become accustomed to here all lives in our buildings, and what a joy to get to help provide changing the technology landscape for the world and building the place for that to live. I really think that’s what we do and it’s been a joy.

Bill Kleyman: No, I hear you and we’ll probably talk about this a little bit more. I feel like we’ve seen a transition happening in this industry where we used to say that we are the foundation of the internet. Now I firmly believe that we are the foundation for humanity, and I think that’s really special. I think that is a change in perspective, especially with all this crazy stuff that we’re hearing about AI, generative solutions. We are shifting the kinds of things that we are supporting for everyday life. I mean, it’s really special.

Raymond Hawkins: Hear, hear. I don’t think it’s too far a reach that the foundation that humanity is using to launch itself into all of that’s next is really built on the critical infrastructure industry, the data centre and the business. So talk with us a little bit. Everybody wants to talk about what’s happening, how’s our industry changing, what’s going on? And clearly the thing that makes the news and then everyone hears and sees is ChatGPT and generative AI, and it’s changing the world and machine learning and all the great acronyms. But talk a little bit about from your perch, what you see as the technology industry and really mankind shifting what happens inside those ones and zeros and blinking lights.

Bill Kleyman: So at the highest level possible, I want to make sure people understand what these systems are and certainly how they work. Generative AI is a type of machine learning. So AI is at the very top overarching umbrella definition term. And generative AI is a type of that which at its core works by training models, software models to make predictions based on data without requirement for explicit programming. That sounds really detailed and complicated. Traditional AI models, the old school stuff from a while ago was designed to identify underlying patterns, data sets based on probability distribution and find similar patterns and so on. Think of it as trying to train a child as to what a picture of a dog looks like. That’s what it is, right? The more pictures you show, the more accurate it is and the more it’s able to define using a neural network, what a dog looks like, what it should be, the different kinds of dogs.

Generative AI is the same thing, but now instead of teaching the child to recognise the dog, you’re teaching a child to both recognise the dog and then draw it, paint it, your own interpretation. So it’s still based off that same concept and model of the data, except now you’re creating images or text or videos or all of these other kinds of other systems. And these are done through different kinds of transformative, generative kind of technologies. We’ve got generative adversarial networks, transformers, and some of these other ones.

Now here’s the big difference. I’ll say this in terms that we can understand. A single Google search can power a hundred watt light bulb for 11 seconds, consuming about 0.3 kilowatt-hours of energy per one Google query. There’s roughly between 80, 90 to a hundred thousand queries in a second. A single ChatGPT query. So Raymond asks ChatGPT a question just one time is about 50 to a hundred times more powerful than that, consuming anywhere between as little as maybe one to two, upwards of three, four, and five kilowatt-hours per query.

Now, if Raymond comes out and says, I want a 50 slide deck built in PowerPoint, that’s going to be a little bit more power consuming. When we start to take a look at AI generative, everybody listening, please understand this is not Zoom, this is not EVs. You’re going to flood the data centres with all this information. We don’t have a precedent for this technology. And just to give you an idea, I think it was the first two months that they entered a million users, active users, and over the course of nine months, they’re just shy of a billion active users, unique users on ChatGPT. We don’t have anything to compare this to. Not Instagram, not the most popular TikTok applications. Nothing.

Raymond Hawkins: Has Facebook hit a billion users? I mean, now I think about it. Are they even, and they’ve been here 20 years, right? I mean, a billion users just-

Bill Kleyman: We can check as we talk. But in that sense, Raymond and everybody listening, when the foundation of the internet came out, traditional data centres kind of lost the race. We were relegated to traditional workloads, VMware, Citrix, Microsoft, all of these other tool sets. And again, the building of the internet was relegated to those big hyperscalers, the larger ones. They were able to build a lot of that foundation. Right now it feels like we’re losing that race again to the large hyperscalers.

The data centre industry is starting to catch up. They’re seeing that their customers are asking for generative AI, large language model training capacity or just general AI support. And these customers are reluctant to go to the cloud. They don’t want their data to be trained against. They want to pay lower prices. They want to have data locality and data gravity. So for the data centre industry, AI and more specifically, this modern approach to generative AI is a massive opportunity to capture a business that’s going to be growing, that’s hungry to leverage this information.

But to that sense, we’re going to have to start thinking about maybe designing things a little bit differently. Now, you mentioned something, Raymond, about the fear, uncertainty and doubt. The not in my backyard kind of conversation. I just wrote an article on data centre knowledge. It’s titled, Get Off My Lawn, You Crazy Data Centres. And if you haven’t had a chance to read that, please do. Because some of those protesters that I saw for the first time in my career in Virginia responded to the article in the LinkedIn thread.

Raymond Hawkins: Very cool. Yeah.

Bill Kleyman: I love what they had to say. And I actually offline connected with one of them and she’s like, Bill, I’m not an idiot. I don’t want to go back to the Stone Age. I don’t want to write stuff on pieces of paper. I’m not against data centres. What I’m against is the following. And she gave me a really good perspective. I know we’re going on a tangent here. She’s like, Bill, I know you’re responding to a market need. And then she goes on to explain that places like Northern Virginia, for example, their government because they want that business, sometimes they will sign contracts and documents for energy consumption that’s from fossil fuels instead of renewable fuels.

And that’s what bothers them. They’re like, can you help us? We need this leadership from the data centre industry to go back to government and assign more pressure to say, yes, this business is growing here. Yes, it needs to be here, but please stop signing contracts that expand on fossil fuels. That’s not an illegitimate request. And I think that’s fair. So there’s a lot of learning and growth, and I honestly think that things like generative AI, that’s just going to exasperate the challenges. It’s going to make it a lot louder. I mean, Raymond, South Park did an episode on ChatGPT. If you can’t go mainstream than that, then there you go.

Raymond Hawkins: Yeah, yeah, you’ve arrived. Before my friends at Facebook and now Meta hammer me, 2.9 billion global users. It took them 5.1 years to get to a billion. So multi years, right? Half a decade. So thank you Google for providing those answers at my fingertips while we talked about it. But to your point, I mean a billion users measured in months? It’s just the growth is astronomical.

Bill Kleyman: Absolutely wild. And what we’ve been doing at Neuro has been really special. So it’s been the democratisation of these technologies. So bringing it into data centre partners. Four years spent at Switch, I learned something, and this is actually summarised beautifully by Peter Gross, who’s one of my mentors, is that the data centre industry loves innovation as long as it’s 10 years old. Well, we don’t have 10 years to wait on ChatGPT and generative AI, we have maybe months and we’re already like six months behind.

                                 So one thing that I learned is that for these technologies to actually be adopted by co-locations that serve enterprise and different kinds of customers is that you have to allow them to continue to be good at selling space and power and nothing else. So we are a platform that sits on top as a Kubernetes engine that continues to help data centre customers, partners sell space and power, but in the sense of actually creating an architecture that’s dense, that is capable of supporting GPT, generative AI-ready data centres so that these leaders stop experiencing revenue bleed by sending these customers to the cloud, which sometimes those instances for GPUs aren’t even available. They’re greyed out. So I want our industry to capture this market and be a leader in it.

Raymond Hawkins: Boy, our friends at Nvidia, it is changing their world, isn’t it?

Bill Kleyman: Oh, I mean these guys have been experiencing the craziest roller coaster ever, right? From cryptocurrency. Yeah, we love Nvidia, coming down, and then now generative AI everything’s GC and GPUs.

Raymond Hawkins: Yeah, that’s exactly what I’m glad you connected the dots back to crypto, right? That was everybody had to have the latest miner and had to have the latest processor, put it in the farthest reaches where we can get the cheapest power. And then that whole thing went south. I’ll be tactful about how I say it. And now what AI is doing for the GPU is just off the charts. It’s crazy.

Bill Kleyman: It’s absolutely wild. And just really quick, the GPU power isn’t always needed, right? You only need GPUs when you’re doing large language model training. When you’re doing slight iterative changes to ChatGPT, you don’t need that kind of horsepower. Or if you’re doing inference training, like for example, Raymond, you ask GPT a question and it retrains the model just because of what you asked. It becomes a little bit better. That also doesn’t require massive amounts of GPU. You can do all that stuff on CPUs.

That’s a part of creating a foundation built around AI ethics, AI transparency, and most of all AI sustainability. So that people using this technology have a very deep… I want to use this word, intimate understanding of what these tools are so that it’s not foreign to them. I think it’s important. That’s part of the democratisation process for people to understand what this is. The kinds of use cases that we’ve been applying to has been absolutely wild, just absolutely crazy stuff. It’s been really fascinating.

Raymond Hawkins: And changing so fast. That’s the thing that’s gotten me, is how quickly what people are doing and how they’re building tools and how they’re engaging with it. Shocking how fast. And we used to… I’m going to date myself again, we used to say that things changed at internet speed. That’s a dated term now, right? This AI speed has just been a blur. How quickly it’s become a real thing and how quickly it’s become transformative in so many parts of our economy and now to our business, right? It is changing what happens in our data centres radically.

Bill Kleyman: Agreed, a hundred percent. And one thing I want to eliminate this is fear. It’s not going to replace your jobs. It’s a copilot. It really is a copilot, right? In every single use case that we’ve been designing this, it’s there to increase value, even for junior engineers and junior lawyers that we’re working with junior service people, instead of looking at one project for eight hours, this technology allows you to look at eight projects over the course of one hour each and then keep pushing forward. The tools are so powerful, and I know we might not have enough time to even go into some of these use cases, but they’re extraordinary. I’ll give you one really crazy example, Raymond.

Raymond Hawkins: Yeah. Give us one. Yeah.

Bill Kleyman: Just one, right? So we’re working with a data centre services company. This company services both manufacturer equipment as well as physical facilities. And they have so much data, so much data that’s being ingested, and some of it’s not shared. Some of it’s in isolated repositories. The goal is to ingest all of it in a privately held model, but then also train it against all of these manufacturers’ constantly updated manuals, field manuals, drawings, recommendations, also train it against field technician notes. I mean everything from the back of a napkin drawing to specific requirements and documentations around IP addresses.

So Raymond from Compass gives me a call and starts explaining an issue to me. What I’m capable of doing is we’re actually going to be deploying an audio engine built with GPT that’s going to listen to Raymond as he’s talking. Now I’m listening to you and just to the left is a screen that’s going to be saying, okay, Raymond is discussing data centre number four.

Cool. Now he’s talking about rack row number three in that data centre. Cool. Now he’s specifically talking about this switch. Ah, he’s describing this issue. Well, it looks like this patch wasn’t updated, which actually fixes the exact issue that he’s talking about right now. So as a technician, I already have a solution. I’m just waiting for you to stop talking so I could go apply it.

And these aren’t 500 page long documents. They’re not. Two, three sentences. Highly contextual to what you are discussing and to what the specific situation is. And what happens is after we’re done, the model learns and becomes even better and then takes that documentation applies to the customer so the next technician doesn’t have to restart all this stuff over. And it’s that level of deep intelligence, multi-layered use of data that now I can apply in real time, start to solve these level one, level two, level three technical issues based on all of this. And again, if Vertiv comes out and releases a new document for their condensers that’s automatically ingested and trained against the model so that when the next service comes in, it’s already smart. A true copilot. I mean, that’s just one scratch of the surface of the use cases that we’ve been working on. They are fascinating.

Raymond Hawkins: The thing that gets me there, it makes me think of, we used to use this term institutional knowledge that that business has learned. And while that business was a collection… And we’ll stick with your technician example, this collection of technicians, there’s the guy that’s worked there three weeks. There’s the guy who’s worked there three years. There’s the guy that’s been there 35 years, and there’s this collective institutional knowledge of this firm that you go to get engineering help and you tap into all of those individuals and it’s between their ears and you collectively sort out the institutional knowledge and experience and you get your problem solved.

And what I hear you describing is we are going to virtualize not only all of the collective of the information, but we’re going to allow it to learn off of all each other’s experiences and every inbound call. So that three week old technician, he’s getting data that goes into the model because he’s hearing something different than when you go get the guy who’s been doing it for 30 years and you send him out on the really complex problem. And we’re going to aggregate all of that experience into one solution and ingest it through this voice recognition software that goes, oh, this is what the guy’s asking about. Let me go ask everybody all at once. I mean, I know I’m being crude when I say, let me go ask. Let me go query the system about all the experience I’ve collected and provide an answer. Virtual institutional knowledge.

Bill Kleyman: Exactly. It’s ingesting that institutional knowledge. And it’s not just about everyone. This engine is smart enough to go into specific pieces and places of documentation to ask contextual questions that apply specifically to the conversation that’s being had. It’s a similar thing that we’re doing with an intellectual property law firm, ingest all their IP, all their laws, rulings, all of that stuff. I’m not a lawyer.

And then ingest all this other information that’s publicly available to them. And then the request is like, can you write a legal document or a legal brief for example? We’re like, don’t think of it so linearly. Tell us the judge that you’re presenting to. We’ll ingest all of his or her rulings and actually create a brief for you that is going to give you the highest possible probability of winning the case because we now understand how this judge rules. And so it’s a deep conscious, I use that word carefully, approach to leveraging data to solve these really, really complex problems. And again, in building these models, supporting people and bringing value. We’ve seen all sorts of use cases, everything from healthcare, increasing the number of people that go into beds, supporting more healthcare services and functions using large language model training to on the fly data changes, for example, to a virtual AI data centre inspector.

A company wants to build in a certain region. You as an organisation can submit and input any document in any type of form it takes, it quantifies and qualifies it for you and acts as an actual inspector and says, you’re going to have problems with this regulation. You’re going to have problems getting this part through this actual road because there’s going to be forecasted closure, so you’re not going to be able to use this highway. So it gives you this detailed perspective of not just the issues you’re going to run into, but how you can fundamentally bring capacity online faster. That’s another concept that we’re working on right now with the company. It’s absolutely fascinating. Again, if you can hear us talking about this, this is way beyond there being a stream of data, us finding a pattern, reporting on that pattern, and then going to business as usual. This is a truly added deeper layer of context.

Raymond Hawkins: Fascinating stuff and an exciting time to be alive. I think change gears, people, I hear people go, oh, we’re going to have, the laws aren’t going to be able to keep up. Well, whenever have the laws been able to keep up? I mean, that’s not a reason not to be charging into the future and engaging with this technology and figuring out how it can change your business and your life and your company and your customer’s experience. I mean, it’s just fascinating stuff and exciting stuff. And the rapid adoption, I think speaks to how quickly it transforms things, right? I mean, that’s the reason. The adoption’s through the roof.

All right, well Bill, Vitali, nothing has happened in this 40 minutes other than it convinced me that we have to do more than one episode with you. So we are definitely going to have to ask you back.

We will do… I’ve got 19 different ideas with things we’re going to have to talk about. So we’re going to have to see if we can get a contract in front of you and convince you to do multiple episodes. But man, it’s been awesome to have you. We’re super grateful to have you with us on the podcast, but to have you in the industry and have you as a friend of Compass. So we really appreciate everything you do to promote our industry and what we do as a business and grateful to have you on with me today.

Bill Kleyman: I appreciate you, Raymond. I appreciate all of your wonderful listeners. Thank you for tuning in. By all means, don’t be shy. Find me on LinkedIn or whatever it is. Let’s continue these conversations. I understand that some of these new technologies you’re hearing about can be scary or confusing or concerning. Let me kind of point at you this way. At the latest AFCOM data centre world event, two of the three keynotes focused on AI, specifically generative AI. And at a data centre world conference, a generative AI company, software company won an innovation award. This is not a fad. This is not something that’s going to pass us by. It is a fundamental part of what we’re going to be doing every single day. Your goal isn’t to just jump on the horse or both feet in the pool. Your goal is to ask questions. Be curious, not judgmental.

Raymond Hawkins: Yeah. Hear, hear. I love that. Be curious. All right, well Bill, awesome stuff. Thank you for having time to join us and grateful that folks get to hear you and hear you on our little bitty platform. Mark our little corner of the world. Awesome stuff.