Cliff Federspiel 0:00
Like the first investor, you know, I talked to him, and he's just like, Yeah, I'll put my money in. That's before the crash. That was, like, in august 2008 and nobody would put money, you know, I couldn't get anybody to invest this meeting with the, again, the director of operations for this company, for data center operations, and he described this whole thing to me, you know, every 30 minutes, or every hour, or something like that. He's getting on a meeting with over 400 people. It went all the way up to the executive vice president. We had a customer really early on. He's like, I don't think people want to have AI in their BMS, you know, that's how he described it to sort of fear, just of AI and then chat. G PT, especially when that came out, everybody got super familiar with AI, what it's good for, but they also got familiar with like hallucinations and things like this.
Eric Bell 0:56
On today's podcast, we have Cliff Federspiel With Vigilent cliff. Welcome to the podcast.
Cliff Federspiel 1:01
Thanks for having me. Eric. All right.
Eric Bell 1:02
Well, before we get in, dig in here. Let's, let's learn about you in rapid fireed mode, yes, and get to know you. So do you prefer late nights or early mornings?
Cliff Federspiel 1:14
I think probably mornings. You know, especially if I'm like out on my bike in the morning, it's really nice,
Eric Bell 1:21
yeah, yeah. And then sweet or salty,
Cliff Federspiel 1:25
definitely salty food can never be too salty for me. Wow, yeah,
Eric Bell 1:31
introvert or extrovert?
Cliff Federspiel 1:33
I'd say introvert, yeah. When I was a kid, I was pretty shy kid, yeah.
Eric Bell 1:38
So what's your favorite food, and I bet it's salty, peanut butter. Yeah, by far as well, yeah,
Cliff Federspiel 1:47
peanut butter sauce, it goes with so many things. You know, I use it as a training fuel when I'm cycling. It goes great on ice cream, all these great foods that have peanut sauce and stuff like Thai food. I mean, it's just,
Eric Bell 2:01
it's great, yeah. Anyways, favorite sport?
Cliff Federspiel 2:05
Well, so these days it's endurance cycling. I do a lot of that, but I've done a bunch of other stuff. When I was in grad school, I played rugby and did some masters water polo for a while. So yeah, but yeah,
Eric Bell 2:19
what was your first job before you got in the industry. So think, like high school time frame?
Cliff Federspiel 2:23
Yeah, I had this job in high school where I'd go door to door selling, you know, people like paint their house address number on the curb. So I sold the service of coming around later to paint their house number on the curb door to door once
Eric Bell 2:39
around this last summer. To do that you look like a high school guy. Yeah, I felt I did. I did decline him, but then afterward, I did notice that our number was really faded, so maybe I need to take UPS guy on that. Yeah, that's right. And then secret talent,
Cliff Federspiel 2:54
yeah, I guess maybe this is, this is something I kind of learned about myself from founding Vigilent, is I'm pretty good at convincing people to believe,
Eric Bell 3:03
you know, I think that's a skill set that's important for founders, is to get people on board and believe in the mission when you're just starting out.
Cliff Federspiel 3:13
Yeah, investors, new hires, customers, all of it, yeah.
Eric Bell 3:21
And then a few work questions here, do you prefer a meeting? Meeting in person or email?
Cliff Federspiel 3:28
Meetings are way better. You know, just the email bandwidth is so low, it's, you know, it's only good for, like, little little
Eric Bell 3:35
things, yeah. And then air cooling or liquid cooling, you know,
Cliff Federspiel 3:39
liquids the future. So, you know, that's, that's what's driving all the interest these days.
Eric Bell 3:46
And then who would play you in the Netflix series data center?
Cliff Federspiel 3:49
I had to think about this one a bit. I put Daniel, Craig, okay, yeah, you know, like I was thinking, you know, I can Casino Royale, right? He just, like, never gave. Oh, so first of all, he's gambling, he's taking risk. And then he just never gave up, right? And then, more recently, I don't know if you saw he's in the knives out series. He did. He plays this character, PUE Nolan, a like an investigator.
Eric Bell 4:14
So, yeah, I like it, okay. And so that, for folks that aren't familiar with Vigilent, how would you describe, I guess, number one, what Vigilent does. But also, how does, how would you describe AI cooling or kind of, you know, your technology, to a non engineer,
Cliff Federspiel 4:33
basically, the software, basically, it has a crystal ball built into it that allows it to look ahead, you know, sort of predict the future, and use that predictive ability to figure out the best thing to do now to create a good outcome later.
Eric Bell 4:50
And the crystal ball is probably based on the data you collect and
Cliff Federspiel 4:57
models that get built in the software.
Eric Bell 5:00
Yeah, and so you're identifying trends. Perhaps every Tuesday gets warmer in the morning, or whatever it might be. I'm just guessing, because this is not my expertise, but yeah,
Cliff Federspiel 5:09
what are the things that affect the things that matter? Some of that is based on the data model itself, like what's related to what in the software. And then what it does is it figures out of the things that the software is configured to do. How are those going to affect the things that it wants to, you know, manage control in the in the future?
Nolan Turner 5:31
Maybe a little silly, but between the cooling and AI, what? What came first, the passion for cooling or the passion for computer science,
Cliff Federspiel 5:40
actually, I learned about both kind of at the same time. I did both in grad school, you know? I mean, I took a whole bunch of so all my degrees are mechanical engineering, but I took a lot of courses over in electrical engineering, computer science, Brain and Cognitive Sciences, but the application that I was focused on was HVAC. So I've been doing the HVAC thing since grad school.
Eric Bell 6:02
So was there a specific point or incident or something, a problem that you might have encountered that, you know, provide a spark or an idea that you said, I need to form a company or build a solution?
Cliff Federspiel 6:14
So I left grad school, did corporate R and D for a while, and then took an academic position at UC Berkeley, worked on a bunch of projects having to do with wireless mesh networking controls for buildings, and thought the time was right to, you know, put all that together so the technology, the wireless, the smart software, and apply it to Building Energy Management, because At that time, you remember clean tech was like all the rage, right? Investors were pouring a ton of money into that. So, so the whole energy thing was really big deal. So we launched the company focused on applying the technology to old, inefficient buildings, like office type buildings, right? And it worked great. But there's two things. One, it was obvious that it was going to be a hard area to build a business. And then the all the concerns about the rapid growth of data center energy consumption happened right then it's like, oh, we should pivot to data centers. So about what year was that. So we incorporated in 2008 and I remember, in 2007 it's the guys at Lawrence Berkeley National Lab published a report on the, you know, expected exponential growth of data center energy consumption, and that went everywhere and got a lot of attention. I think there was a lot of concern in the industry that you know, is going to result in regulation if something wasn't done about it.
Eric Bell 7:44
Yes, I know that's been a concern for a long time, and it's primarily been self regulated. Yeah, interesting. And in those reports from the projections they had then are probably dwarfed by what's actually happening now in the in the expected energy usage that's gonna, well, yeah, it's,
Cliff Federspiel 8:02
it's exploding so much in the last, you know, three to five years that it's, you know, straining the grid in North America, right? Things like that. It's not about efficiency, really. It's about capacity to deliver power to loads. You know, that's really the problem now.
Eric Bell 8:19
So you're President and CTO, that's two roles, right? You know? And which curious of like, what hat do you prefer wearing? And then how do you balance, like, kind of the kind of creating new products from a CTO perspective, with kind of running that the business day to day.
Cliff Federspiel 8:37
So, so we got a team that runs the business day to day is certainly not just me and I, I convinced a guy named Mark Housley to join his CEO pretty early on. So which is a great move, you know, he, he's a, he's great at running the business, but I do, I mean, like, I'm interested in in building, really, you know, like, almost magical technology, but that solves an impactful problem, really, you know. So, so you want to have stuff that's cool and exciting and really seems like magic to some of our customers. I think, you know, it does seem like magic, but that solves real problems that customers have. And so that's that's sort of the balance between the business it's got to be driven by solving a like a real business problem
Eric Bell 9:28
made the crux is the problem that you're solving is kind of energy usage. So if you solve the cooling issue, is it is about energy, or is it really about cooling?
Cliff Federspiel 9:37
It? So it was about energy. When we started the company, we would sell and close like the whole everything about the sales process was about energy. And we learned along the way that that's hard because it's not the priority, really. So it's hard to, you know, to have that as your soul. Value proposition, but we also learned that the value of the solution goes way beyond energy, but the other, the other just to extend on that a little bit more. So you mentioned energy saving. So the same actions that the AI produces to save energy, they also, you know, help the equipment run longer, more reliably, so you get better capital efficiency. You get fewer repairs and all that stuff too. Along with it's the same thing. It just basically is turning things off that don't need to be on, and slowing motors down, and all of that which and that, you know, some of that, that value is, you know, more, quite a bit more valuable than the energy. And then finally, if you're not using up a bunch of energy for cooling, that means you have that power available for the it. So, you know, if you have the ability, and we have some customers who've actually done this, they, you know, shuffle their assets around so that they can basically get more capacity out of their datacenters.
Eric Bell 11:02
That's all that we're dealing with all the time, is finding more power. And there's concern that there's not a power in the grid, et cetera, et cetera, right? And do you find that? Because it almost surprised me at first when you said that, that maybe your customers maybe aren't, or prospects may not be as in tune with $1 saved is $1 earned, right? You know, meaning, if you're saving money, that's not compelling enough, but perhaps, maybe finding more power in that data center. You know, if it's a 50 megawatt data center, and you can be 5% I don't know what your efficiency ratio is when you deploy, if you can save 5% on that, that that's significant. You know, it's several, it is.
Cliff Federspiel 11:41
It's real money, but it's not the only thing, and it's not the it's not the number one thing on their list. You know, there are a bunch of other things. And so the other ones that I mentioned, like labor. You know, we have a partner in Japan, and, you know, this is what they've told us right now, is like labor is their number one problem. You know, the Japanese are aging out, and they're not bringing in a lot of immigrants. So it's just getting really, really hard to find good employees in Japan. So being able to make the people that you have more effective is super valuable, right?
Eric Bell 12:19
And so do the systems go beyond just, obviously, beyond just cooling infrastructure, you know, it's almost like, is it a replacement for DCIM, or,
Cliff Federspiel 12:28
I think it's, you know, sort of works alongside DCIM. I mean, a lot of what DCM does is asset management, so keeping track of things, and especially, like, you know, some of them keep track of things as they move in and out of the data center. So, you know, that's that's not what a Vigilent system does, but there's certainly overlap, and the thing we've done in the last four or five years is extend the so. So when we started the company, it was sort of a point application for optimizing cooling in data centers, right? But to make that work. We had to, we had to build all the platform infrastructure underneath it. And so we finally got to the point where we said, you know, we need to just turn this into a platform. And so we've now extended our data model so we can natively model all the assets, so the power assets, or even the IT assets in the data center, those can all be represented in the software. She means, then you can, you know, assign data to them, have trend data, and all that kind of good stuff.
Eric Bell 13:28
Do you have some good exact examples my dog just knocked over my garbage man below my desk? Anyways, do you have any good examples of maybe building custom apps for a customer. That sounds interesting to me.
Cliff Federspiel 13:43
So we built some apps that allow our customers they wanted to so in our software, we have a temperature map, and we have another thing called an influence map. So I mentioned the crystal ball that that's driven by a thing we call an influence model, and so we can visualize that so people can kind of see the effect of cooling on different parts of the data center. But, you know, we had a customer that wanted to do more visualization than that. They wanted to be able to see like temperatures in their data center, asset by asset, but also back in time. And so we created this, this thing that the engineering team named it Neo it sort of looks like that, you know, that pattern in the matrix, you know. But then when you look at this thing, like in a single snapshot, you can see exactly like where temperatures are getting too high or too low over time, and which ones, you know, and they sort of cluster together. So it's this analytic that the customer wanted to be able to have so they could not only see what's going on right now, but look back in time and have it all in a single graphic.
Eric Bell 14:49
It sounds like that would be even as an app for you built for that customer. It was custom for them. It sounds like you could almost take that and then roll that out to the rest of the well,
Cliff Federspiel 14:58
yeah, that's the idea, right? Is, you know. Yeah, and this is, you know, like, I think an important kind of, like, part of our business philosophy is to solve customer problems, right? So build the tech and the apps for customer problems, and then, like you said, that chances are pretty good if one customer has the problem another customer has,
Eric Bell 15:16
let's see. So from a tech perspective, and talking about the sensors, you know, I'm kind of curious, of like, how these are placed throughout the facility. Do you put one in each rack, or is it one in each row? Or how, how are they placed? And what else do you measure besides temperature?
Cliff Federspiel 15:33
Yeah, well, so a lot of the time we provide wireless telemetry to our customers, and, like, our nominal default is top and bottom, or high and low on every third rack. You can go lower density than that. If, like, if all you cared about was just saving energy, you go lower density than that. But then, you know, like, all the visualization in the UI starts looking a little wonky because it's hard to get the interpolation to work right. So, so all the analytics and the graphics and stuff like that are really, you know, a bunch of those are quite a bit more valuable if you increase the sensor density.
Eric Bell 16:08
And AI, you know, you talk about using AI in the system, but AI is, you know, is a term that's used very broadly. And AI can be many different things. You know, machine learning, large language models, you know, physical, robotic models, that sort of thing. So I'm kind of curious, in the context of new Vigilent or datacenters, you know, what, what do you guys use for AI under the hood? Is it? Is it an LLM that monitors this, or is it some sort of
Cliff Federspiel 16:36
other, I mean, we don't. And, yeah, this is, so this is the thing you know, as AI, and the public's understanding of AI sort of evolves as the technology is rolled out, kind of unaware of it in the early days, and maybe a little bit afraid of it. Like I said, the we had a customer really early on, and he's like, I don't think people want to have AI in their BMS. You know, that's how he described it to me, sort of fear, right? Of just of AI. Now, people are like, and then chat GPT, especially when that came out, everybody got super, you know, familiar with AI, what it's good for, but they also got familiar with, like, hallucinations and things like this. And so we go to a customer sometimes and they hear AI, they're thinking, Oh, my God, you're going to put chat GPT in my, my, you know, cooling system. I don't want that. And it's going to hallucinate, and, you know, and, and that thing has, you know, is, you know, out at a third party cloud and, you know, all that stuff. So, so we don't, we it's not an LLM. We don't put llms in there. We have, like I said, we have a thing called an influence model. We use machine learning to train the parameters of that you know, the models are, like, millions of times smaller than, you know, like GPT five. So, so the software you trains these machine learning models automatically doesn't require any human intervention, and then it uses those models to optimize so it makes optimal control decisions using those models. It starts off with sort of a template model and some simple control to make the things safe. And we have so the software does sort of two kinds of control actions. You know, in the reinforcement learning area, you know, there's this term explore versus exploit. So some of the time to facilitate learning. You want to explore. You want to have the software explore so it can generate information.
Nolan Turner 18:35
Can your model hallucinate? And if so, how do you prevent from it doing so
Cliff Federspiel 18:40
it can and when it does like, what happens is, you'll see, so I mentioned we have temperature maps, right? And we also have influence maps. And if it does hallucinate, what you'll see is the influence maps look kind of ridiculous, like it'll tell you that if you, you know, change the state of a cooling unit, that it'll get, you know, 40 degrees hotter, or some you know, or even you know, something is just obviously wrong. And so what we've done is we've designed the machine learning algorithm so that it makes it quite a bit more difficult for the machine learning to hallucinate.
Nolan Turner 19:21
So how exactly does it learn the cooling behavior? Like, can you walk us through, kind of what happens from when you install your sensors to the point where the system starts generating recommendations? Yeah.
Cliff Federspiel 19:32
So it we, you know, configure everything it would put in the sensors, connect the software to whatever, whatever thing it's going to send a signal to to do control. Sometimes that's itself, is a wireless device that's on that wireless network we were talking about, so we can do control over the air. But sometimes it's across a network that's already there, some ot network, you know, it could be using back. NET or Modbus or SNMP, something like that. And that could be either directly to the asset itself, or it could be routed through their BMS some customers like that, because then they can write some code in their BMS that basically allows them to control conditions when they want to, you know, either allow or disable Vigilent control through their BMS, so it gives them sort of an automated way to have control of their Vigilent system outside their vigilance system.
Eric Bell 20:30
And then, how do you leverage your full deployment across different customers to kind of build these ml models? Or is it all does all the learning happen at an individual site?
Cliff Federspiel 20:45
Or maybe most of it is site based, right? Because, you know, the design is different from site to site. So, so the I mentioned, the one of the ways we deal with the hallucinations, we have a prior. So the prior is based on kind of a an average of what we've seen over, you know, a lot of installations that we've done after the models been trained. But you know, there are a lot of differences from data center to data center that need to get factored in. So the majority of the information that ultimately affects what the influence model knows about the site is based on the data from that site itself.
Eric Bell 21:26
But certainly, if there's a certain manufacturer, crack manufacturer, in a certain model, right? You know you might learn something across multiple customers, and then you could apply that even in a new installation, knowing that that crack unit, or the expected behavior of that crack unit,
Cliff Federspiel 21:43
yeah, yeah, you could, although, you know, I mean, the design of a particular of air conditioners is kind of has a sort of de facto standard to it. And the other thing that we see is that, you know, even though you might have, like, a whole lineup of chillers, right, that are nominally identical, and they were just commissioned three months ago. And when you start looking at their behavior, they differ by a material amount. Yeah, yeah,
Eric Bell 22:18
unique things when you put them in, you know, into different environments.
Cliff Federspiel 22:22
I mean, it's just like a fleet of cars or anything else, right? I mean, you know, they look the same, right? But they, when you actually, like, really carefully measure their behavior, they're not the same. They don't all that your fleet of cars, every, every single one of them isn't going to get exactly the same gas mileage, and it's going to change over time. Yeah. What's the like? Why? Why put AI in a data center and have it optimize the cooling? Right? And the reason is, all the value that it delivers, and the value is, is a lot more than just energy efficiency. It's it's making the people who run the data center more effective because they don't have to do the low level stuff anymore. The software takes care of that. It's the improvement in the liability and capital efficiencies, because equipment lasts longer, doesn't have as many failures. And it's the ability, I mean, these days, like we were talking about, it's the ability to have the AI help the data center not use so much power for cooling so that it can use more for it
Eric Bell 23:25
makes sense. And I've been meaning to make a slide in our, one of our customer decks that talks about the rack densities and how they're increasing. I think it would just be really cool to show it, you know, say, for example, you know, 10 years ago, the average was maybe four kilowatts a rack, or maybe even, that's being generous. And then not too long ago is 10, right? And then, you know, some of the AI systems with the h1 hundreds were 30 or 40 kilowatts. And so GPUs are pushing that right? You know, the current deployment and video is like 130 and with a with a roadmap to, you know, in the two hundreds and seven hundreds and a gigawatt
Cliff Federspiel 24:04
perhaps the announcement just this week, right? Vera Rubin is now gonna, like double that,
Eric Bell 24:09
yeah, which is insane. You know, in terms of this density that, you know, as data center operators, we're expected to to be able to support this rapid growth in power densities. So I'm curious of maybe what, what you're seeing from that perspective, you know, maybe how Vigilent is working through those, those densities, or any thoughts in terms of,
Cliff Federspiel 24:31
well, so we've had customers. Customers are certainly worried about it, like our colo customers are coming to us and asking for help, because they it's, it's a daunting thing for them, right? They're going to have to build this really expensive asset, a data center, and they want to still be having this thing, you know, useful and generating revenue for them years into the future, and yet, you know, the rate of change of the IT assets and the power. Uh, you know, it consumes, and the heat that it generates is, you know, escalating exponentially. So they're trying to figure out what to do about that, you know, they're trying to design them so that they're as future proof as possible, right, so that they can, you know, connect the next generation of liquid cooling to their cooling plant, and still, you know, be able to put those high density racks in. Yeah, it's a, it's a challenge for the designers, for sure, yeah.
Eric Bell 25:33
So I guess what would be the, you know, potential solutions that they're looking at, and I know you're a little bit outside that, that realm, but are, can you speak to that piece?
Cliff Federspiel 25:43
Well, they're, they're trying to design so that the design is flexible, right, so that they can, you know, connect in new liquid cooling assets in the in the future. I mean, I think, like some of them, fully expect that they're going to have, you know, they're going to start off with large rooms that are pretty full and end up with large rooms that are not very full because the density is so high. And they, you know, they only have so much capacity.
Eric Bell 26:10
I think I only watched a couple episodes of what is it called severance, I think on his HBO or Apple, where it seems like it's weird because I never really got the show, but I know it's very popular, and it's, maybe I watched the whole thing, yeah.
Cliff Federspiel 26:24
Was it good? Better? Yeah, okay, I liked it, yeah.
Eric Bell 26:28
You know, they have these just four desks in the middle of a large room. So I can, it's almost like that's going to be our colocation environment in the future, you know, right? Yeah. And I, which is, which bizarre. So, how does Vigilent work in a liquid cooling environment? Do you know, it's easy for me to visualize, okay, you know, in an air cooled environment, where you measure, you know, outlet, you know, or the inlet from cold air coming into the server.
Cliff Federspiel 26:57
So, so a to a Vigilent system, a CDU looks an awful lot like a crawl unit, right? It's another thing that the cooling plant, the chiller plant, is serving. So, you know, it's got a it's got a chill water valve where the chillers are delivering water to it. So our software optimizes the air cooling in the data halls, and then we have software that optimizes the set points for the chiller plant. And if you have a CD use, then we it's going to do the same thing in order to produce the best outcome in terms of power efficiency for the whole plant.
Eric Bell 27:35
Yeah, that makes sense. Then, yeah, you can integrate into that. No, no problem, right?
Cliff Federspiel 27:40
Yeah. And integrating is easy. I mean the controls that are on CD us. I mean, there it's all the same commercial and industrial controls that you know you can integrate with, with protocols and of various sorts. So you know, getting data from them is not an issue.
Eric Bell 27:57
So cliff, can you describe how how a customer might deploy, and I'm sure there might be a couple different ways you can deploy. For example, you know, do you have to have sensors or not? You know, do you send everything up to the cloud, or is it all there on the customer's prem or within the customer's environment? Right?
Cliff Federspiel 28:16
So the software, of course, needs sensor data, but where that comes from, it just sort of depends. You know, a lot of the time, Vigilent provides sensors to our customers, but more and more so these data centers are all fully instrumented before. You know, if it's an existing data center or even one that's that's new a Greenfield, they're going to design it with sensors, all in the places where Vigilent software needs sensor data. So it's becoming more and more common that we a Vigilent deployment is only software now where that software resides is nominally on prem. We find that our customers don't want to have a, you know, third party or public Cloud connection to their infrastructure, and then
Eric Bell 29:07
maybe moving on to more, you know, personal things. So we want to ask a couple personal questions, and then maybe get into a data center war story and your top picks. So what is the the best advice you've you've received, you know, if you can think, I'm sure there's lots of advice, but is there any, yeah,
Cliff Federspiel 29:25
you know, when I was a grad student, there was this, you know, famous, beloved professor who his advice to the students was, tell everyone you know, everything you know. And you can't really do that in business, because you got, you know, things that you can't say, right, but, but to the extent that you can, I think that's great advice, because, you know, we were talking about believability and convincing people to believe, and so being able to tell them in detail about things that are important to them, you know, really helped. Yes, it gives you credibility.
Eric Bell 30:02
Does that mean, like, being more transparent? Is that what you're is that the concept there is being transparent and open, and maybe I don't want to say over sharing, but transparently showing?
Cliff Federspiel 30:14
Well, I think so. So this professor, I think what he was trying to counter is the sort of the academic tendency to, like, keep everything, like, safe and secret until you can publish it and, like, release it to the world, right? And, you know, you see that in startups, right? They go into stealth mode and they don't want to tell anybody anything, but, but getting people to believe that you have something means that you got to tell them. And so you got to, you know, you got to, you know, you got to, you got to talk to people, yeah, you got to be somewhat transparent. Otherwise they won't believe.
Eric Bell 30:46
And this might be related to the days our war story. So, if so, we can move into that. But what is the toughest moment of your career, and how did you get through it?
Cliff Federspiel 30:55
Yeah, so one really tough moment was really early on. We incorporated in 2008 and I started raising money, you know, from
Eric Bell 31:05
2008 that was the financial crisis. Yes, it was great recession. Okay, that you started. So you and Elon Musk, he raised money in 2008 for Tesla, I don't know. In other words, it's a tremendously hard time to do business, particularly as a startup. So go ahead, I just it was super
Cliff Federspiel 31:25
hard, right? So, like, the first investor, you know, I talked to him, and he's just like, Yeah, I'll put my money. And that's before the crash. That was, like, in August, 2008 and the crash happened, what? In October, September, October, I forget which now, and nobody would put money in? You know, I couldn't get anybody to invest,
Eric Bell 31:44
but you still had that first investor, investors, the first
Cliff Federspiel 31:47
investor, maybe two. Had two at that point, yeah, you know, put a little bit of money in. So, right? Not, this is not VC. This is like, you know, angel investors, putting money in
Eric Bell 31:58
from that you had enough, and you just had to scrimp a little bit more than you would have.
Cliff Federspiel 32:06
Really hard. Was right there, there wasn't 500 people that I had to make payroll for. So it was that part of it was easier, but it was hard. It was really hard. You know, I mentioned that I convinced mark to come on board early, that that was part of that right, nobody's going to invest in me. I'd never made anybody a gazillion dollars before, but Mark had a lot of business experience, and so he added to the credibility of the team. So basically, anyway, one of the things I did was like, strengthen the team so we could get through that. And one of the things I did, in fact, to help win that grant, is that Steve Chu who was then the Secretary of Energy, he was given a talk over at the Harvard Club in San Francisco. So I went over there with, like, a two page printed out thing about Vigilent, my business card all stapled together, sat in this room, had dinner, and then Steve Chu shows up, and also in the crowd, just like immediately crowds around him. So I wormed my way into the crowd, got up there, was able to, like, talk to him for 30 seconds about what we were doing, handed him the two pager and the business card, and what I heard later through my contacts at lb and L is that he reviewed every single one of those proposals, and so maybe that helped, I don't know
Eric Bell 33:28
they might have, right? I mean, I think that's a way to kind of get ahead in career. Is not just sending in a resume. You send in a resume in a unique way, or you show up in the door, or you reach out to them over LinkedIn, right? It's about really interjecting yourself and putting yourself out there, because otherwise you're not known, if you're just an application without that personal introduction, and putting that name with the face. And he's like, Oh, he's eager, he seems smart. He wrote something that was good, and it resonated with me. You know, who knows what it is, right, that actually got you that.
Cliff Federspiel 34:01
But well, the other thing that happened when I was talking to Steve too, is he, you know, I explained him what we do, and he started telling me how the system ought to work. And so I told him, No, no, you're wrong. And here's why
Eric Bell 34:15
was he in retrospect, was he wrong and you're right? Yeah. I mean, he's probably smart enough to kind of see your perspective on that then,
Cliff Federspiel 34:23
no, he kind of paused for a minute thought about he's like, Yeah, okay, yeah,
Eric Bell 34:28
okay, cool, yeah, that's always risky, that to kind of call someone out, you don't know how they're going
Cliff Federspiel 34:33
to react. No, you don't. But, you know, seems like maybe it worked. It worked.
Eric Bell 34:37
Yeah, so let's move on to the data center war story, although that sounds like the war story right there. But any, any tell us about time, either yourself or protect the innocent, if it's if it's a customer, you know, that sort of thing. But you know, what is there an outage that happened or something that and how did you recover? A couple of. Them.
Cliff Federspiel 35:00
So this, I'll give you two, and you can decide which one you want to go with. But yeah, so tech company got, you know, our system installed, getting ready to turn it on, and they had a, you know, a BMS vendor, you know, guy that would help them, you know, run their BMS from the BMS company, and he just did not want to have this Vigilent thing turned on in this data center that he, I think, felt responsible for, you know, helping this company operate. And so, so he was, like, pushing back really hard about turning it on. And he escalated this all the way up to the director of operations for this tech company and and I thought it was just going to blow the whole deal up, you know. And this is early, early days Vigilent, and so we really needed this to happen. So I got on a call with the director of operations and the guy from the BMS company, and just like went through every objection that the guy had, you know, and you know, said this is what will happen if that happened, you know, just all of that. And got the director of operations to agree to let us turn it on. So, you know, we had a, you know, we always do. We had a, you know, somebody from Vigilent there, carefully monitor it, you know, I assured him that, you know, we'll have a person there. If anything happens, we'll just roll it back. It'll be good and safe. And of course, it it was. It was a great outcome. And, and the interesting thing here, and you know, credit to the guy from the BMS company, he sent us a note afterwards, and he said, I would have never believed this if I hadn't seen it with my own two eyes. It was amazing,
Eric Bell 36:41
for sure. Yeah, and what was the second one?
Cliff Federspiel 36:44
So the other one is, we have a customer. It's a, you know, one of the gigantic financial companies and so, you know, they're super conservative and kind of similar sort of attitude amongst some of the people in the organization. You know, they're not really comfortable with having a vigilance system in there operating because, you know, I mean, something goes wrong in one of these data centers, it's, it's a real disaster, right? I mean, they, they run trading floors and all this kind of stuff. I mean, it's just terrible if something goes wrong. Now, we designed the Vigilant alarming system so it's got, like, has these roll up alarms so you don't, you don't have anywhere near the level of granularity like you can alarm on every single little thing in your BMS. And I think because of that, people often will and so what happens is, we have an event like that. That's what can happen. You get this alarm flood, and it's sort of like a denial of service attack on your BMS. And that's what they had. So the only thing, and so I had this meeting with the, again, the director of operations for this company, for data center operations, and, and he described this whole thing to me, you know, and, and he said, I forget how often it was every every 30 minutes, or every hour, or something like that. He's getting on a meeting with over 400 people. It went all the way up to the executive vice president, explaining what's going on. He said the only eyes that he had was his vigilance system. So after that, they love Vigilent,
Eric Bell 38:13
I bet, yeah, and you just filter that. You just didn't take on every alert you filtered.
Cliff Federspiel 38:18
Yeah, that's right. I mean the software. I mean it mostly, you know, our kind of thinking is, first of all, like, Look, you got to be a mess. So it's got a bunch of alarms. You got an EPM s, it's got a bunch of alarms. You got maybe a decent, it's got a bunch of alarms. So, so, you know, we're kind of frugal, and, you know what we alarm on? You know, we do have, and we've actually, recently, you know, built in more flexibility to do alarming but, and this is kind of a thing that, you know, a lot of our customer facing people are almost, you know, sometimes complaining about, oh, we need, we need more flexibility in our alarms until we have added that. But when this happened, we didn't have that yet. And frankly, it was a good thing. No, you know, you're like, alarm on the things you need to alarm on, and no more. Because here's an example of what can happen if you don't.
Eric Bell 39:06
Yeah, it makes sense. Let's move on to the top picks section. So this is a time where, you know, we have a guest provide and and I can provide a topic as well, and no one is your choice, too. But you know, effectively, you anything in your life that you find useful, that may people may or may not know about, and you can go first, or I'm happy to go first, your your choice
Cliff Federspiel 39:33
was looking at your list. And so the travel tip, my travel tip, I would give people. Now there's sort of two of two of them, but the very specific one is, if you're traveling in London, hold on to your phone. I was traveling with Mark in London, and we're heading to a meeting, and he's holding his phone out in front of him like this. You know, we're navigating while walking to a destination and motorcycle. So I'm on Mark's right hand side of motor. And the streets to our right. Motorcycle pulls up right alongside me, right and so I just kind of like move over closer to Mark, just naturally, it's like, what's going on here? The guy reaches, uses that to lean in, snatches his phone, guns it, and heads off. And then he stops about 100 yards ahead, turns, looks back at us and waves and then rides off.
Eric Bell 40:22
Is that, like, a little, like, creative flare that the little wave left? I mean, like, but,
Cliff Federspiel 40:27
but this is a in London, they have a major problem with phone theft. Yeah, it's a, it's really bad. So you gotta, you gotta hold on to your phone if you're traveling.
Eric Bell 40:37
I thought, you know, I don't know. I've been in the Apple ecosystem for a while now, and I so, I don't know about Androids, but I thought, like, it's, it's becoming, I thought it was becoming harder to steal, because I have heard that this, there's this problem in Europe there where people come and snatch phones, but for me, it's like, you know, can't you just lock up the phone and put it in Lost Mode? Oh, yeah, he did.
Cliff Federspiel 40:55
He did all that. But, you know, it took, it was like a lost half day the next day, you know, like dealing with the fact that you didn't have a phone,
Eric Bell 41:02
yeah, but I almost, I feel like, you know, if I leave my phone out on the table at a restaurant or whatever it might be, right? I feel like, safer than having my wallet on the counter of something like, the wallet can't be tracked or it can't be shut down or off, or whatever, or disabled.
Cliff Federspiel 41:20
But yeah, if you're traveling, you lose your phone. It's, it's a problem.
Eric Bell 41:23
Sol, yeah, yeah, no doubt, at least for a day, until you get to get to a store, mobile phone store, yeah. So don't, don't keep your your phone. You kind
Cliff Federspiel 41:33
of don't, don't wave that thing around.
Nolan Turner 41:37
I have for travel. It's, it's not like a fanny pack, but it's, it's much thinner, and it's to be worn kind of under your shirt, and that's where I keep all my travel documents. Yeah, that's a good idea, right? It's like, super thin passport, a little bit of cash, and then my phone is the only thing that fits in there.
Eric Bell 41:55
And for me, I would say I had held off on buying a Tesla for the longest time, and I bought one in mid this year. And it was partly because I, you know, outside all the politics that seemed, it's crazy that it's gotten to that point, but outside of that, like, I don't pay attention to politics much, but for me, it was about the technology and it getting closer, right? And so full, you know, followed it, you know, read about it for a while, and it felt like, you know, full self driving had gotten to the point that it was would be worthwhile, right to, you know. And so got it, and had been using it, and it's been, I don't know if you have had an experience like this, or a Tesla so well.
Cliff Federspiel 42:42
So, you know, I live across the bay from San Francisco, so I've been in waymos, and those things are awesome. It's, it's amazing, yeah, that technology is just great. I love it.
Eric Bell 42:55
But Waymo is, like, it's great, but it's, it's, it's something that you have to procure, and I'm in Denver, and that's not available here this I can have my own garage and pull out, you know, I can just program in where I want to go, hit a button, and it drives me there and now parks and I barely have to, you know, correct it at all. That sort of thing has gotten to that point, and I, you know, my wife isn't as she's okay with it, but she doesn't use it that often when she drives the car. But for me, in my mindset, I'm like, you know, I have a dishwasher to wash dishes, right? So I don't have to sit there and wash the dishes. Same thing with a washing machine and all that. Machines are meant to offload our you know, just like you talked about with Vigilent, right? You know, you have a machine to do work for you and replace labor and all that stuff I don't need to pay. I still pay attention to the road, but I don't have to, you know, it takes a cognitive load off and relaxes you a little bit when you're driving or riding along and having the car drive.
Cliff Federspiel 43:59
So, yeah, Subaru. It's got this thing called eyesight, you know, the lane following and car following, and it gives you some of that, right? You get in stop and go traffic, especially if it's on the freeway, just turn that thing on, and all you have to do is point the car. You don't have to really do anything else. It's much better. But I may have to look into self because this self driving stuff was great. I mentioned I do a lot of cycling, right? So I'm, you know, interacting with cars all the time, and it's not fun, let me tell you, sometimes interacting with cars. But I was over in San Francisco, I'm coming along the Embarcadero in the bike lane, and it's a really narrow bike lane that they provisioned for the cyclists on the Embarcadero, and there's a slow cyclist in front of me. So I go to pass, and I kind of move over a little bit, but I want to go all the way out into the lane, and I kind of can see in my peripheral vision a car is there, right? And it pauses. And that's sort of unusual, like cars don't give cyclists that kind of room, usually, right, especially in San Francisco. So I look back, and it's a. Amo. So it saw me move over a little bit. And, you know, the law is you're supposed to give three feet of clearance if you're going more than 25 miles an hour. So it did, and it waited for me to move back into the middle of lane, and then it carried on. And, you know, so the safety benefit of self driving, and this is, I use this as an example, when I'm talking to people trying to get them to believe about AI, you know, there's, like, really well documented evidence about the safety of driverless cars compared to human driven cars. It's like five to one safety.
Eric Bell 45:30
Oh yeah, yeah, for sure. There's no doubt. And yeah, I would bet, if you look at it, you probably know better than I but safety on roads for either pedestrians or cyclists must have had been declining for even now, because I think that there's a lot of distracted driving with phones. Yeah, people are on their phones, and that's our thing, and distracted, you know, updating whatever they're updating on their phone or reading. And, you know, with self driving, it's become safer, just for the example that you gave with cyclists, right? So I think, yeah, 10 years out, I think it's the roads are going to be a lot safer.
Cliff Federspiel 46:09
Yeah, I can't wait. Yeah,
Eric Bell 46:13
all right, well, where can thanks for jumping on. Learned a lot about this. This is a topic I didn't know a lot about ahead of time, and we got pretty technical, but where can we find out more about you or reach out to you? If anyone has has questions or want to, want to reach out to you?
Cliff Federspiel 46:30
Yeah, so I'm on LinkedIn, and fetters feel is a pretty easy Google able name, so I'm pretty easy to find, but certainly find me on LinkedIn, for anybody that cares about cycling, you can also find me on Strava, and I let anybody connect who wants to
Eric Bell 46:46
so cool they can see your routes and speed and such. That's cool. Awesome. All right. Well, thank you, Cliff, yeah. Thank you, Eric. You.