From Data to Action: Optimizing Solar Operations with Raptor Maps

Thank you, Harshit. Great to be here.

Yeah. I started not in marketing. I started in a hybrid, multiple hats role at a small consulting/data analytics shop, which later got acquired by a much larger consultancy during my time there. But during that time, I wore sales, marketing, biz dev, growth, and solutions engineering, multiple hats all at once. From there, I pivoted into growth marketing afterward, which solidified my marketing pathway. After that, I joined a company called Smsara, which makes fleet management hardware and software for logistics companies. I was a growth Marketing Manager there and transitioned into being a Growth Product Manager where I was building internal tooling for our go-to-market teams. Then I missed more focused marketing, so I pivoted back into marketing where now I’m at Raptor Maps That’s brilliant.

Yeah, totally. Wraps collects, aggregates, standardizes, and analyzes data from all different sources, from sensors, from robotics, such as drones, as well as people on a solar farm. We aggregate and standardize that data into a digital twin of each solar asset. Then we deliver geospatial, geo-referenced intelligence and insights into the performance of the asset, equipment issues that might be causing power production loss, safety issues that might be causing unsafe operations for people to go so that they can fix it, site condition monitoring like there was a big storm, let’s go investigate what happened to our equipment, to the soil, to the ground so that there’s no erosion, so on and so forth. We take all of that data when we spit out easy-to-understand intelligence for people to go understand their operations, to go to fix issues, feed into their models, tell their investors how well they’re doing, et cetera. Then as a big part of that, we allow for the operations of autonomous drones on solar sites. When I say that, it’s a very big emerging technology in not only the drone space, but also in the solar space, but it’s a blank slate.

You can’t just put a unit on site. You need to, one, be able to program missions that are specific to what you’re trying to do in the solar world, but two, you have to take all of that data and make sense of what our AI and ML models can do. Our software provides the foundation for both the operations of the robotics unit on site, as well as the analytics that comes from all the data it collects. At the core, data is really what we’re trying to do here.

Five or seven years ago, we changed how solar inspections were conducted. Previous to Raptor Maps, the mainstream, so to speak, practice for inspecting solar panels was a practice called IV Curve Tracing, which was very intensely manual. So someone would go in and plug into the back and take measurements from each module. So as you might imagine, very inefficient. You’re going module by module, but also pretty dangerous, electrically. So what our founders did, they were, they pioneered and pushed for drone imagery to be used at scale to inspect solar panels. So solar panels produce a thermal signature when they have issues. Depending on the type of thermal signature that’s detected, you can identify what the issue is, and then you can scale that up to multiple hundreds of acres, farms, and millions of modules because drones collect that data efficiently and scalably. We pioneered that, became the premier player in that, and that was our starting point as a company, and that I would say, what sets us apart there is our digital twin technology. Accurate, granular data localized and geo-referenced to specific pieces of equipment on site. Not only that but it’s delivered through an interactive, really easy-to-use, map-based data model.

Building on top of that, really What sets us apart is that we create that data infrastructure for you to clean your data and have a clean system of record for your asset health so that nothing gets lost to history. Second, your technicians or people in the field, ways to collect data, and close the last mile loop. Technicians or electricians on site need to be able to navigate to work orders or tasks to finish, which our platform allows them to do because it’s massive. Solar arms, do you want to pull a paper map? Probably not. We provide technician workflows and tools for them to navigate and perform tasks safely and efficiently. Then third, I would say that because we are that data infrastructure to allow for the applications of new technology like robotics, we get to move rapidly there. Robotics is a key component, but also other AI solutions. For example, our instant inspection solution allows for people who operate commercial and industrial solar of our solar assets. So think about solar assets on the rooftops of a Walmart or a Target. We have a solution that allows for technicians or electricians to go for regular maintenance visits to just fly the drone, collect data, and then upload it to our platform, and they’ll get results back in 45 to 90 minutes on really high-priority issues.

We built that specifically for that sector because, one, it’s a burgeoning sector in the solar space. I think of how many rooftops we have, and how many parking lots we have in the United States. But two, the industry requires a good amount of turnaround time around reports and things like that because they’re going to be QAQC, et cetera. Once you receive that report, it gets added to the backlog, and then it takes a couple of weeks or months for them to prioritize it. But then during that period, issues might get worse. We’re trying to cut out all of that and give them really powerful tools to fix issues right away for that particular sector of the solar industry. I guess that’s a long-winded way of saying we’re building specific applications of AI and robotics based on direct feedback and problems we’re seeing from the solar industry. I think we’re doing so in a methodical way and one that allows for much more scalable applications of these new technologies.

Yeah, totally. We think drones are a fantastic way of collecting data, whether you have a pilot doing a yourself or this autonomous unit. But for sites that are out of the way or solar projects that are out of the way, out there, sending someone to fly drones regularly can get costly. There are labor constraints, there’s labor costs, there’s people’s time, et cetera. It could get costly quickly. What the autonomous drone allows you to do is collect more data more frequently and consistently across a really wide variety of use cases. Because the drone lives on site and doesn’t need to have someone drive out, there’s the speed of response. But two, it’s no longer that costly to collect that much data. That means that more regular flights and more regular data collection means you can do so many different things beyond just dermal PV inspections. We think the use case of PV inspections that I was describing earlier in our conversation that we pioneered, that, I think, is almost a cherry on top of all the immense amounts of use cases that you can unlock with a drone box or the autonomous drone.

In terms of the safety piece, as you might imagine, let’s say there’s a storm or a security breach that the customer wants to investigate. I don’t know. I I feel uncomfortable sending out my people into a potentially flooded area where there’s electrical equipment. I would rather have a robotics unit, like an autonomous drone, fly and investigate and do some initial data gathering before sending anyone out on-site. That’s one of the safety elements there. We had a big DAO leadership executive event a couple of months ago, and something that resonated with me was, and this wasn’t about the autonomous drone itself, but just even flying drones, was one of our customers said, Hey, at the end of the day, it’s a $5,000 piece of equipment. If that goes down or gets damaged, that is infinitely better than having someone get hurt or putting them in danger. I think that the power of AI and robotics is to empower the people who do good work. That’s the mantra of how we’re thinking about building this autonomous drone solution. To clarify, we’re just a software layer we don’t provide the hardware.

So that is the PV inspection piece I was telling you about where you collect that data at scale, the thermal image at scale, and then using the thermal signatures that you see within each module or rows blocks of modules on the solar farm from an aerial view, you can start identifying, Oh, this is this issue, or this is an issue with this piece of equipment because you can see thermal signatures across this entire row of modules. We do it at scale because not only of the data that’s being collected by drones, but also our analytics engine that processes the data and identifies the issues, and then our localization engine that puts that data into a geo-referenced layer on top of our digital twin. The digital twin is essentially just a virtual replica of the solar farm with each piece down to the individual equipment level. What that means is that once the aerial thermography inspection is done, that is then just the data layer on top of that. Then every insight from this inspection layer gets localized from the specific pieces of equipment. I know that this piece of equipment at this GPS coordinate has this particular issue.

Then if I was either an asset manager back in the office or an electrician or technician out on site, I could pull up the entire history of that equipment. Oh, this equipment seems pretty problematic because we’ve identified multiple issues and have fixed multiple issues. Maybe it’s time for us to look into replacing that piece of equipment or looking into whether it’s warrantable, et cetera.

Do you mean prioritization of how to fix things?

Got you. The TLDR of how it works is that we calculate an impact based on how much of the module or modules are offline or active by this thermal signature. As an example, at a really small scale, an individual module might produce 300 watts or 400 watts. Or maybe 650 watts. Then if there’s an issue for a certain type of module, there’s something called a bypass dio. I’m an electrical expert, so I don’t know what that does. But typically when there’s an issue with the radio, there’s a third of the module is going to show a thermal signature. We’ll just say two-thirds of the module is producing power, and one-third of that module is not producing power. Then you can scale up from things impacting just 1% of the module to all the modules to rows of modules, all the way to blocks of modules on a platform. We provide prioritization based on power impact and dollar impact so that people know, Okay, I’m going to put this on the watch list. We’re going to go fix this, and we’re going to go replace this piece of equipment or something like that so that it gets smarter recommendations for people to make more informed decisions about where and when to send out people and what they should do.

Yeah, that’s a great question. The digital twin that I just mentioned exists both on desktop and on mobile. The mobile app users, they’re typically the folks who are out in the solar farm doing the electrical or technical work, where they’re either going for what we call corrective maintenance, which is like, We found something, let’s go fix it. Prevent maintenance is, We think this might cause issues later on, let’s go give it some love, or just some other work that they do. Oh, I need to, I don’t know, tie a zip tie that’s loose here or go fix some wiring issues. The mobile app provides really precise geo-locations. When you open it up, it has a little blue dot and it’s like Google Maps. You can just Google Maps your way to the exact work order task that needs to get done. We have technician workflow so that someone in the office can create, Oh, I got the latest inspection results from Mapthrough Maps. I have this insight site from recent data that we’ve captured from a technician being on-site and finding issues, I’m going to create a bunch of tasks and assign them to my team to go fix.

Now they have precise geo locations or GPS locations for each task so that they don’t need to double guess, or second guess where they need to be going. They can just use the app to navigate to it. They can pull up all the details they need. They’ll know, Oh, this is a bypassed IoT issue based on this imagery that I can just pull my phone. I know exactly what to do. And then they can just leave notes behind or they find another issue that wasn’t another issue that’s occurred since the last inspection. They can record that, and then all of that gets uploaded to the cloud to close the loop. People can track, people can track, Oh, I finished this task. Oh, I found some more information, and then boom, and then uploaded it to the cloud. And it’s available offline mode, in offline mode, because most of these are in, as you might imagine, not a lot of houses and buildings around. So they might not be Signal. We built it so that it can be used offline, and then all of the data gets uploaded once they have Signal again.

Totally. Transparently, that’s more driven by our product team than from the marketing side. Ho, however, we work closely with the product team to understand how our customers are using the mobile app. It’s built with technicians and actual users in mind. But then we always find that they’re doing creative stuff with the app and how they’re using it. I think in that situation, my job is how to educate our customers about the power of our mobile app and what others are doing using our mobile app. The second is, that I truly think that we need to make sure all of our solutions, again, go back to our mantra of all these emerging solutions around AI and robotics need to help people do good work and make their lives easier. Part of my job is also bringing that technician to focus to the forefront and saying, We’re building solutions to make the people who go and fix the issues and build these farms for them to be able to do it more efficiently and more safely. How do I tell the market that this is a big focus of ours? We’re building robotics-related solutions. We’re building a whole suite of solutions that make it easier for people to scale up their operations.

Inefficiencies scale just as much as efficiency scale. How do we take all of that inefficiency out? That’s what we’re trying to do. And that’s what the marketing is trying to tell that story and bring that narrative to the market.

Yeah, we take a content/leadership-heavy approach, as you might have realized from my long rants so far. We take a leadership-based approach. As such, email and LinkedIn are really big channels for us. It allows us to disseminate this content and knowledge in a targeted way to the people who might be looking for it. But two, it also gives a bit of credibility, I think, when you’re seeing thought leadership on LinkedIn or a nicely formatted email versus, I don’t know, a Facebook ad. I know paid LinkedIn or getting LinkedIn emails. Then we have a look. Organic traffic is huge for us. That is a function of the dark social that a lot of people call and a lot of other brand plays that we’re doing. Website is a big focus of ours. How do we optimize customer journeys on the website and things like that?

Yeah, this is certainly a work in progress, given how we are still an early-stage startup.

Yeah, it’s a never-ending process. But I think what we’re trying to do is do a balance between A/B testing, and incremental changes, but also being bolder about big changes. What I mean by that is if we see something isn’t working and we have a hypothesis on why it’s not working, we rather start new and then do incremental A/B tests from that rather than do incremental A/B tests that may or may not move the needle. If I’m not going to test against the control that I already know is not working. Maybe that’s not the best way to think about A/B tests, but when we’re running multiple A/B tests as well as trying to make some big changes and keep up or stay ahead of our messaging and our product development. Sometimes you can’t just do slow incremental A/B tests. That’s part one. Part two is very analytics-driven. Looking at GA4, looking at our traffic analytics. Where are the drop-offs happening? What are the engagement rates? What are the bounce rates? Are people even clicking on our stuff? Are people going to the pages we want them to go to? If not, how do we introduce more pathways to those pages?

Those are two really big ways we’re thinking about it. But A/B testing at its core and first principles-based thinking are two building blocks for that.

Yeah. Our target market does not consist of a ton of small businesses that we can acquire at a lower CAC. What that results in is that some of our sample sizes are not statistically significant all the time. What I use instead is, that I study statistics in college, so this rurhurts to say this, but sometimes I can’t wait for a statistic, and I can’t wait for the experiment to run that long. I see that there’s a pretty big divergence in the testing control, and even if it’s statistically significant, I don’t know, an alpha of 0.05, I would feel comfortable calling it if I see, I don’t know, two weeks of data, whereas consistently there’s a divergence, even if that divergence isn’t statistically significant. I think it’s a balance of understanding when to call things, but also having a little bit of context to know whether this is an assessment of the experiment or not. We try to make it as much of a science as possible, but there is a little bit of there because, again, you can’t wait for three months for you to reach statistical significance if you’re already seeing that divergence.

Yeah, I think it’s a combination of a lot of things. One is that we try to keep up the momentum of our content output. That’s not just SEO blog content, but meaningful value white papers and ebooks. I know people are like, Oh, ebooks are dead in 2024 or whatever, but we find that that’s BS. People genuinely want to read. It’s only dead if it’s just marketing fluff. What we try to do is push out content that people want to read based on the data that we’re collecting. A really big report we release every year is the Global Solar Report, which is us taking the immense amount of data that we aggregate across all of our customers and then extracting interesting trends for the industry to be like, Hey, It must be really important for us to be honest with what the problems of the industry are because we’re solving for a very real problem of climate change. This is a really important, urgent mission for us. Solar is emerging as the biggest source of new clean energy deployed rapidly and cheaply. We’re a young industry. We’re growing fast, and we’re critical for the world.

What our thought leadership wants to do is we want to help the industry grow efficiently, and we want them to grow sustainably and healthily. And sustainable being like, we can let inefficiency scale not sustainability in a green way, but also that. I think keeping up the pace on that thought leadership and differentiating thought leadership helps with our organic traffic. I think there’s a halo effect from our paid ads, our around from LinkedIn, and probably from Google as well. And we also have a big field marketing presence. The conference circuit is big in the solar world, so we do a lot of marketing around that as well. And then we, again, thought leadership via our people as well.

Are you talking about Wrapped in math specifically or just solar in general?

Yeah, you might have guessed this would be my answer, but I’m excited about what the future of robotics looks like in solar. I think that autonomous drones are a really big portion of where we are Our energy is focused right now at Raptor Maps. But just generally, I think there are tons of cool companies doing full robotics stuff and building cool robotics hardware for the solar industry that I think I’m just generally really excited about. I’m excited about Raptor Maps being part of that ecosystem, providing that software layer to operate ingest, and analyze the data for autonomous drones. We’re working with leading OEMs to do that. That’s genuinely exciting to me as a sci-fi nerd and as someone passionate about the solar industry. I guess it’s every little kid’s dream to have a little transformer, and we’re getting closer to that, at least from a solar perspective. Excited about that.

Yeah, I think so.

That’s a very good question. This might be a controversial answer, but honestly, I think TikTok, it’s just for all of these distractions, I I think that intentional use of it has helped me learn a lot. There are a lot of experts on there sharing knowledge that once you start looking, and obviously, there’s a level of media literacy and double-checking information and trust but verifying. To get done. But I don’t know, there’s a plethora of really knowledgeable people on really niche topics that I’m curious about that I learn a lot from. That would be my controversial answer, but my other answer is Strava. Not that I’m like an exercise fardeau, but I love seeing everyone validating each other for their workouts. As someone who needs a lot of motivation to go exercise, it’s nice.

Yeah.

Oh, wow. Oh, man. I wouldn’t say this is necessarily bizarre, but it’s something that I was shocked by how well it worked when I was at Samsara, Direct Mail was a really big channel for us. There was a hot sauce campaign that they ran where they directed to me with branded hot sauce saying, Are you curious about Samsara’s secret sauce? I don’t know, I thought it was a little bit gimmicky, but it worked well. People love gifts. That was earlier on in my time at Samsara where that opened my eyes. Marketing is not just digital. There’s some old-school stuff. It does work if you do it right. For sure.

Wow. The weirdest place where I’ve come up with a brilliant idea. I would say a lot of it. A little bit more, a higher percentage of my good ideas come from shower thoughts than I I’d like to admit to. It’s not the best place for brainstorming, but I feel like a lot of interesting ideas come to mind when I’m just meditating in the shower or something.

Oh, wow. That would be nice. I think I would invest in probably more I’ve headcount plus probably some AI solution that works for us. I think that we’re in the early stages of how Gen AI can be applied, and there are a lot of companies out there that, to my point about AI should help people do good work, I think that I would probably put at least some of that budget into ways that we can make my team’s life easier because we’re a small mighty the am. do really good work, and I want to empower them to do great work. That’s where I feel like I would spend most of my budget around helping my team and elevating my team.

A lot of roles. I think it’d be a product. Our product team does a lot of interesting work, obviously, but they do field testing. They’re out in the solar farms talking to customers and testing our solutions and making sure it works. It’s so hands-on in a way that I love. From time to time, I miss being a product manager.

Last Google search. I was looking up restaurants in Edinburgh because my partner and I are going in August for her birthday, so I am trying to make some reservations so that we don’t have to scramble and find a dinner place.

Thank you so much for having me. And looking forward to seeing this go live. Thank you. Bye-bye.