Bringing AI to Energy Management with Sam Cox of Mountain Vector Energy

Ryo Koyama Hi, I'm Ryo Koyama. Welcome to Remotely Local, where we talk about all things that are happening in the device world and physical AI. I'm very happy today to have Sam Cox on who's CTO of Mountain Vector. Welcome Sam.

Sam Cox Good to be here. Thanks for having me.

Ryo Koyama So, for those who don't know what Mountain Vector does, uh why don't you give us a little bit overview what the company does and um what's happening in your particular space of the market.

Sam Cox Uh so Mountain Vector we do all things energy management from you know basic utility bill management type stuff all the way down to high level advanced analytics to breaking down you know solar battery setups getting specific uh usage and savings numbers usage and spend savings numbers against that. So lots of high level analytics in regards to energy management a lot of AI different pro uh AI processing different types of things for uh just making people's lives easier when it comes to that environment.

Ryo Koyama Now we we've talked a little bit in the past and one of the things that I found really interesting is that you know energy has been around for a long time and you know adoption isn't always as quickly done as people think and so you know you you talked about dot matrix printers but it sounds like you know part of the challenge isn't as simple as hey let's grab all this data let's analyze it you know the way that it's portrayed in the media talked about the dot matrix type of experience because it goes all the way down to that.

Sam Cox Sure. Definitely. So, utilities are a very I mean, utilities have been around since electricity's been around and getting them to do or to adopt any new technologies is a very difficult feat because there's all these integrated processes and things like that that have been, you know, people have been using for years. So, they don't want to break things. So, they just kind of keep going with the same setup over time. And what you'll find is a lot of smaller local municipalities, they'll have these uh utility bill formats that are just, you know, paper documents that you have to scan in. Um, but the format is something akin to a dot matrix printer, right? Like you just have very poor quality account numbers, very poor quality all of the data. It's very difficult to read for a human. So trying to get all that stuff together such that a computer can do anything with it. It takes a decent amount of effort. And thankfully we have, you know, AI tools and different things like that to help us these days. Uh, but it's just trying to help move an industry that's very much stuck in the past into any type of really to the 2010s even is just it's a feat within itself.

Ryo Koyama You guys are in the entire domestic US or how do you typically work?

Sam Cox Yeah, so we're in throughout the entire United States right now. I think we're coast to coast. So we've got some spots in California all the way out to uh Detroit, I think even Massachusetts area. uh we work with lots and lots of different school districts, a few different industrial clients and a few other utilities that we'll actually kind of work with to do specific uh things that they need.

Ryo Koyama Well, that was kind of a range because it sounds like you work, you know, specifically with a lot of public schools all the way to industrial companies. And so I would imagine that the whole environment looks very very different and it's probably not shocking anymore, but there must have been some cases where you walked in and you just thought, "Wow, this is not what we expected to do." Any any fun stories in that arena?

Sam Cox Sure. I think um one of them we had an industrial customer that one of their primary problems was getting the information out to their facilities managers to be able to make energy decisions based on locational LMP pricing, locational marginal pricing. Um so you're tracking that you know 15-minute, five-minute price signal uh live throughout the day and they want to know are we going beyond a certain threshold so you know should we curtail operations should we keep going how should we plan for this in the future so we're doing a lot of work with them to like set up some things so they have more information in terms of figuring out what to do next and we're also working on some more predictive type things so that they can you know instead of reacting to the present maybe start to plan out for the future a little

Ryo Koyama What I love about what you guys are doing as a company is that the media loves to talk about physical AI and I think to the point where there's a c certain amount of uh probably exhaustion about the topic but you know to me it's AI is obviously very sophisticated everybody reads about it in the paper every day in terms of how powerful it is in some ways very dangerous but at the same time what you have to do and physical AI is really about taking the existing world and making it be able to take advantage of AI and obviously you know at Remote.It that's one of the things we pride ourselves on is we kind of think of ourselves as kind of that that infrastructure there that allows that connectivity to happen so maybe you know dig into you know sort of what are the problems that you were facing and why you guys enjoy using Remote.It or maybe what the use cases might look like

Sam Cox definitely I I think it's a great point you said like setting up the infrastructure because everybody and their mom is like we have AI-driven insights we have AI driven this AI driven that somebody's got to build the road if you're going to drive on it right So setting up that infrastructure, getting that data from point A to point B, that's super important for us. That's like our primary pain point is just getting the data to the right place. So from a Remote.It perspective, what we're doing is we have, you know, hundreds of devices deployed all across the country and we're we're a small company based in New Mexico, you know, one state, I think 10 or 12 people. we don't have the resources to be deploying people all over the country to go on-site if there's issues with our devices or issues that we have to work with somebody to figure out. So what we're doing with Remote.It is we're setting up all of our devices so that they can be remotely accessed you know anywhere in the country uh so that we can get on change configurations for things upload data figure out you know what's going on if we're having problems and we're taking all of that and then it's feeding right into our backend so we can do different analytics uh live real-time spend calculations against um utility rates things like that so it really just supports our ability to have that uh that live metering infrastructure.

Ryo Koyama Yeah, I mean that's the one thing I mean it's interesting because you know as a networking guy I always think about connection but so much of it is now just about communication like you talked about it's like you know at any point in time you know that's sort of what inspired the name of the podcast you know remotely local and so you know you want to be in New Mexico and like you know and I think your case Montana and you want to be able to see your whole environment as if it's there and it sounds like that's a that's kind of it's not only important it's kind of become fundamental because if you can't see the world today you sort of have a problem and it sounds like that's part of what you're able to take advantage using as a service.

Sam Cox Definitely. I mean if we didn't have Remote.It available I was going to have to build it myself and this was this is four years ago that I was looking at having to do this and I was you know thinking about reverse SSH tunnels and setting up all this infrastructure and I was like this is going to be a lot of work for something that we're not going to have to rely on every single day. But I found Remote.It and I was like, this is exactly what I need. It the price is, you know, I'm I'm saving money by not doing this on my own. The customer support has been phenomenal. Anytime I've had an issue, even if it wasn't with the product, I was having some uh internal issues on the device that I was having trouble diagnosing. And I was, you know, emailing back and forth with with rep and they tracked it down to an MTU level problem and got it fixed on my end and all of a sudden we're back up and running. So Remote.It I've had nothing but spectacular experiences with Remote.It.

Ryo Koyama Well, that's awesome. We we obviously love to hear that and we're not trying to make a commercial here, but uh it never hurts when use. Well, that that's awesome. Like I said, I I think the big challenge is that we're starting to see and again what I appreciate for what you're saying is that, you know, most of the internet was invented when it was a much simpler problem, you know, and eventually got to everybody just needs to get to a website. But now that you're facing not only you know internet of things or industrial internet of things along with AI all of a sudden now you need a real-time communications with everything you have cellular networks you have people with Wi-Fi you have all kinds of different network configurations and one of the things we pride ourselves on is kind of making all that invisible so people don't have to deal with that and it sounds like maybe that's the key benefit for you is to not have to think about that. So taking it further then I mean talk a little bit about how you see physical AI developing now because you know the way I look at it now is going back to what I said no one who is in the smartphone generation they they fully expect that they can get to anything anywhere you know they don't understand the complexities of it because that's sort of the expectation today and so as the world of energy and energy management starts to develop what what do you think is going to be in a crazy amount of time like 3 years from today what is going to be the expectation that isn't the expectation today.

Sam Cox I think I guess a more declarative world meaning that we know what we want in our heads. You know, for example, like at a CEO level, I know what product I want, but I don't know how to get to that implementation level. But if I can just say to an AI, I want this and it can do it, then all of a sudden the world opens up in ter in terms of possibilities. So like you can go from having kind of loosely structured data or like very like in the case of utility bills for example varying different types of data with like loose relationships. I can turn an AI at it and say go retrieve this information, go do this, go do that, you know, put it all in this report, put it in this database, and it can just figure it out and do it. And so it's it's kind of just it's going to augment our reality, I think, in terms of human capability. It's just another tool really that we're all kind of learning how to use and things are getting better and better as we go. But yeah, I think if I just had to look three years out, it's just going from that more imperative setup to that more declarative. Just say what you want, you know, get what you need.

Ryo Koyama Well, I love that. Like I said, I think we're gonna we're gonna definitely title this podcast declarative, not imperative because uh I love that expression, but I think to your point, the big challenge of shifting to AI is that everything isn't just plain language. And it becomes very very declarative because you're not thinking about how do I make this happen. You're thinking about the what do I want to have happen? And it can be very very frustrating if it's not possible because again the expectation in the the world of smartphones and next with AI is going to be why can't it happen and obviously ChatGPT and all the other LLM are very good about saying well of course you're right Sam I should be able to do that for you and so that's the world that we're heading into. All right as we wrap up then what's your hot take? So what's the thing that you think that's going to happen in the world of physical AI that's maybe not conventional wisdom that is maybe a little bit orthogonal to what everybody else is thinking?

Sam Cox I think especially around the beginning of uh generative AI becoming more and more popular. There was this general kind of I guess concern or thought that like AI is going to take your job. AI is going to you know do this AI is going to take over all this stuff and then people are going to kind of be left on the sidewalk with nothing to do. And in my experience, having AI available to me, it's sped me up in a lot of ways, but I am more busy than ever because now instead of doing one feature a month or one feature a week, I can do six. I can do 10. And so it becomes like I was always going to have to solve those six or 10 other problems, but now it just the queue empties and fills up more quickly. So, I think AI is really going to go from I'm going to take your job to you're going to have to have me to use your like to do your job. For example, like a computer or Excel. Like there's certain accounting things that you would never want to do by hand, but that's how they used to do it. Then you get Excel and now you can do 50 different accounting things, but you've now you just got 100 more customers. So, it's all just a matter of volume. And I think uh we're just going to see more and more volume pour into people's work queues, I guess, as a result of AI, rather than seeing people's jobs be taken away.

Ryo Koyama No, I I I think that's a very good take. I mean, one of my views is I always say every industrial revolution was marked by indentured servitude. And if you look at the economics of AI, in a lot of ways, this is the cheapest indentured servitude we've ever had. And I think you explained it well. It's like, well, I would have had I could have thought about these six things in serial, but now I do them all in parallel. And and I love your Excel analogy because I think that's the big thing. It's like, I could do this, but it'd be terrible, and now it's like, oh, I want to get this done, and it's a great thing to be able to do. So, we talked a little bit about how AI is a little bit too positive. Do do you see that as a problematic thing where they it is so reinforcing? I mean I I joke that it's sort of the creation of religion because it is the source of truth for so many people. What what do you think especially like in the energy sector what's the risk of AI being so declarative about your right and how do you think that will affect your industry as it goes?

Sam Cox It's an interesting question. I think the risk you kind of run into there is like a very simple one I can think of is like you ask somebody to generate a report that would historically require you to go and you know gather all this information kind of do all this manual analysis things like that when now you can tell the AI to go do it maybe it has to go to the internet fetch things maybe it has to read some things from documents and maybe you've described exactly what you want maybe that's what the AI produced but what you said and what you needed are not the same thing. And that's kind of what I'm sometimes finding when I'm working with AI is that I'll say that I want one thing and then it'll go to do it and I'm like, "This doesn't work in a production environment. Why would you even attempt to do this?" And so I think it's it's something that's got to be closely monitored. Not sure how well that answers your question, but I guess that's kind of how I'm thinking.

Ryo Koyama No, I I think that's right because, you know, again, to reinforce your point about it's going to take jobs in a lot of ways, it makes human competence more important because you're going to have to evolve what those things are. So, well Sam, it's been great to have you and anybody who's looking for physical AI in the energy space, Mountain Vector is the place to go and u they're definitely leading from our perspective a lot of the activities that are out there. So, it was great having you on and I'm sure we'll have you on again as we introduce some of our own AI tools because I definitely would look forward to getting kind of your feedback on that.

Sam Cox Amazing. Sounds great. Thank you so much for having me.

Ryo Koyama Thanks, Sam.