Podcast #34: Leveraging Aerial Intelligence – by George Mathew

Good day everyone, this is Ed Maguire, Insights Partner at Momenta partners with another episode of our Edge podcasts, and today we have George Matthew who is CEO and chairman of Kespry which is a company that focuses on Aerial Intelligence, and we’ll get into this a little bit more about what they do and the value that they deliver. It’s great to have you George, thanks for joining us.

Absolutely Ed, thanks so much for having me, looking forward to this conversation.

Let’s start off by hearing a bit about Kespry, can you tell us what does the company do, and what are some of the technologies that you guys are working with?

Kespry by background is an Aerial Intelligence product that focuses on the collection of sensor-base input, predominantly coming from drones; we manufacture our own drone here in North America. We take the sensor-base input that’s coming off the drone and process that in our cloud infrastructure, where we convert the sensor-base input into physical reality capture, and that physical reality capture can then be analyzed using machine-learning artificial intelligence, and package that up with a series of analytical applications for industrial use case, whether it be the inventory management solutions that we deliver in the mining space, topological assessment of a construction worksite, going into asset management in the energy sector, and of course being able to do claims and similar processing for the insurance roofing space.

So, we think of that end-to-end experience Ed as the way that we can create a lot more industrial capability around drone tech, as well as effectively understanding the physical condition of a lot of these key assets in the industrial world.

That’s great, I’ll come back to that and would like to drill into some of the things that you do, and the problems you guys are addressing. But first, could you provide a bit of context around your background, what has brought you to be focused on Connected Industry, and can you characterize what has shaped your particular view of IoT, as we call it?

I’ve been in the software industry mostly as a focused executive in the analytic space for quite a few years. So, in the 2010-2011 timeframe I was at SAP and I ran the business objects BI Division for SAP, and saw that proliferation of business intelligence getting more and more democratized, particularly in that period where I had an opportunity to join Ultrix very early-on as the president of the company, really to look at the democratization of the analytics in the hands of business users, specifically when it came to analytical data prop blending and modelling.

When I worked through that experience for 5½ years, and helped to take the company public, we saw a real opportunity emerge, particularly around the analytical applications, and specifically the industrial focus in the analytical applications that just were underplayed in the market, mainly because the data needed to be collected from a lot of physical sensor-based input. So, with that mindset that there was data that needed to be collected from physical sensors that are on an industrial worksite, Kespry having an opportunity to build a full autonomous solution in this space, and the opportunity to bring forward analytical applications for these industrial use cases, I thought it was just the perfect next opportunity as Alteryx got to scale, and that’s when I joined Kespry about two-years ago as the CEO of the company.

Could you talk about some of the developments that you’ve seen in the analytics space during your career. I think that’s pretty fascinating, because a lot of folks that we talk to that are in the industrial space as it were, actually come from operational technology backgrounds. So, I’d be intrigued in what you saw evolving, and what really attracted you to work with industrial data.

When we look at the industrial data that’s emerged, particularly in this last decade, there’s a need to get better digitization of these physical assets that are in the industrial work world, and to understand the condition of those assets to predict the future value of those assets. If we look at the data that’s now collected off of drones and other sensor based input, this is actually the first primary source of collecting that insight, in a way that we can drive these predictive values of how assets perform over time.

In that regard I believe that taking the internal sensor based input of things that are streaming from an IoT standpoint, with the physical data capture of sensor based capture of information, whether it be a fixed camera, whether it be a drone, whether it be IR sensors, light based input, really, you’re developing this inside-outside view of the creation of what we think of today as a digital twin. Those digital twins are really able to be harnessed to understand the future state of those assets, and the predicted maintenance value associated with those assets. This has been a fundamental lack of understanding that’s occurred in the industrial work world, because there hasn’t been a simple way to be able to capture these insights, particularly when it comes to physical reality capture, and combining that with the economic value of these assets over time.

So, Kespry feels that there’s a great opportunity to be able to deliver the analytics around the physical model, largely using the sensors that are on the drone, as well as fixed sensors that are on an industrial worksite, being processed in a solution like the Kespry Cloud, and being able to integrate with those internal data sources that are already on many of these industrial worksites. And really that’s been the focus of the company for the last three or four years.

I think it’s really interesting the way you’re characterizing how you need to create physical and virtual models I guess, in the form of a digital twin, but your being able to create that model is really something new, you’re not pulling from ERP systems or existing machinery for instance. I’d love to get your insights on what’s involved in being able to effectively build a model, or build a digital twin? You’re kind of hacking away at the jungle here when you’re building it! I’d love to get your thoughts on what are some of the considerations and pitfalls, and what helps you effectively generate an effective digital twin?

I’ll give you an example of what we accomplished in our first go-to market use case and the mining space which is very relevant, and you mentioned the ERP in that regard, to how ERP data is understood specifically when it comes to inventory and the mining space. So, it turns out that historically to collect the insights of what’s occurred on a mine site on a quarry, as far as the inventory of how much material you have, you would literally take a GPS backpack, or precision laser guided equipment, point it at a stockpile and be able to create a biometric model of that stockpile with 15-20 points of measurement. The points of measurement would create a biometric model, that biometric model would then be multiplied by the density, that would give you the mass of that material, that would then be inputted into your inventory management system, largely surrounding an ERP environment.

What’s been a challenge in that space is that the collection of that data was pretty infrequent, and the source of that information was pretty de minimis, because you were collecting a few points of data to be able to create a biometric model, and so you had 10, 15, 20 points of measurement around a volumetric stockpile of material where you got it from a GPS backpack or a laser pointer.

With a solution like Kespry and other drone products that have emerged in the space over the last half a decade, we’re now able to fly over an industrial site, a mining operation for instance. We take the imagery that’s coming off of a drone, and in our case its high-resolution quality imagery that we convert from 2D to 3D because it’s a series of angled images that we would use photogrammetry pipelines to create the 3-dimensional model, because the 2D can be converted to 3D using a point cloud. The point cloud then can be meshed up with the imagery that creates an accurate 3-dimensional model, that’s stockpile. That underlying stockpile could then have about 400-500 thousand points of measurement and that becomes the new biometric calculation input for then taking the density in and calculate the mass out for that stockpile material, as well as that entire mine site.

Now, why that is compelling is that when you were taking that with a small number of points of measurement previously, with the status quo in that market, you were leading into inventory overages and variances and underage’s, because the topological measurement that you were taking was around 10 to 15 points of measurement, and you could see forecasts for your answers to be as high as 15 to 20 percent per stockpile. We’re able to bring that down to 1 to 2 percent on a per-stockpile basis, so that you can get much stronger accuracy of how much material you have at any point in time. This has immediately changed the way that that industry works, because instead of hiring a surveyor to come in once a quarter, once a year, to get your accurate physical measurements in place, you’re now flying a drone once a day, once a week, to be able to understand that change in inventory measurements, and bring that into your ERP system.

So, these kinds of fundamental opportunities have emerged because you can now take a more frequent view, and a more accurate view, and frankly a safer one, because you’re not manually climbing your stockpile any longer, and get a digital twin-generated that is far more useful on a consistent operational basis day-to-day, than you will have historically done in the mining space for the last two decades. Really that’s where Kespry went from five customers to over 200 customers just in the mining aggregate space, because we’re able to deliver that level of accuracy safety, as well as reliability for the collection of that inventory management data which gets naturally sent into the ERP environments on a weekly and monthly basis. That happens over and over again in construction, in roofing, in the energy sector for a multitude of use cases, and we tend to just focus on finding those real high-value use cases, and deliver an intern solution to the respective markets that we’re in.

Could you talk about the challenges of managing these much larger amounts of data. I guess your background at Alteryx makes you very familiar with large volumes of data, but how do you go about determining the most relevant data, given that you’re collecting quite a bit, and ensuring that you can essentially zero in on the most efficient, or most efficiently what are the deltas that have the most business value? Do you work with domain experts that are able to interpret the data for you, or is this something that the customer has an instinctive feel for?

When we look at the industries that we’ve focused on to-date, we’ve taken a very customer first approach to what the value is in that market for having an innovating product like ours, to really be able to take a market by storm. In that regard we’ve worked with both customers as well as domain experts to validate that the solution is a viable alternative to the status quo. So, I’ll give you another example of this, when we went into the insurance roofing space, we were being compared against men who climb in the roof and taking a measurement of the roof, a tape measurement, and then drawing a chalk-based physical test square where there might be hail damage, and physically counting the amount of hail that’s within that test square.

We ended up rendering a virtual model that dimensionalized the entire roof, using the same exact techniques that I mentioned earlier, and then applied an artificial intelligence set of algorithms that identified and distinguished what hail looks like, and enabled the adjuster or the claims adjudicator to enhance and tweak that model, based on their knowledge and understanding of the domain itself. So, in this case, having that level of customer intimacy in the use case really digging in to making sure there was an end-to-end offering that was delivered for our customers, and doing it in a way where the industry found it acceptable as a radical substitution of the status quo, became the way that we were able to frankly extend ourselves into the insurance roofing sector.

So, you’ve got to dig into the use case, it’s not just about the flying of the drone, the drone is a great way of collecting with sensor-based input, but when you’re flying an industrial solution like ours, which happens to autonomously fly over 30 minutes of flight on a single battery, and cover 200 acres, you’re generating plenty of data. There’s no lack of data being generated, each flight is operating around several gigabytes of information per flight, but then what do you do with the analytical modeling, the applied artificial intelligence and the exposure of an application, or an API, and in most cases both, that’s where the value gets derived. To date we’ve been really focused on delivering that end-to-end experience, so that the physical reality capture using sensor-based input from a drone is just the starting point. The real value is in the application that’s generated from this data process and pipeline that we are supporting our customers scaling-up.

Could you talk about the environment for using drones, or unmanned UAVs, where some of the technologies have come from. Clearly there have been a lot of uses in the military, very sophisticated uses in the military, but is there a technology that you’ve been able to adapt from existing use cases, and also when you hear about people that are doing do-it-yourself inspections for instance on construction sites, how has that environment evolved since you guys have come into the market?

I think I’ve highlighted this before, if we think about the market for industrial use cases, there’s a natural tendency to initially just take a drone out of the box, consumer or prosumer-oriented solution, fly it manually, collect a few images and call it a day. There’s plenty of value in just getting physical imagery of an existing asset and being able to understand just the high-resolution detail of that imagery, and manually making assessments of the condition of the assets, or an anomaly that you might find.

What we’re tending to see is, the market is beginning to create these real industrial applications, not just from manual inspection, but having a fully autonomous process that collects as much as 200 acres over site, processes that data so that you get a real digital twin of that asset that’s down to 3 centimeters of accuracy, x, y, z, in real space because we’ve applied precision GPS on top of the imagery, that creates it from 2D to 3D model a level of accuracy that’s needed, layering the whole series of machine-learning algorithms to identify those anomalies, and expose those as applications.

I think there will always be an opportunity to fly a drone to take the corporate selfies, as I like to call it, for imagery of wherever the state of an asset is, and understand the high-resolution imagery of that asset. But I think the real value becomes taking that imagery, having it consistent and reliable to be able to create a three-dimensional model, layer in the machine-learning necessary to understand the automation of where those anomalies might exist, and then expose that as high value applications.

So, we think the market is going to continue to evolve, probably bifurcate where these high-value application used cases is where Kespry’s focus, and there’ll be plenty of opportunity to take the corporate selfie from manually flown drones in the space as well.

You made an interesting comment earlier that you guys manufacture your own drones, rather than bring them in from China. What are the reasons behind that, or advantage of that?

Historically we’ve had more of a sensor base control, and in the drone that we’ve manufactured we can actually bring in a precision GPS capability to the product, we can go ahead and introduce new technologies as it emerges, like three-dimensional light air into the solution. So, it’s about having flight control autonomy, and the payload that we’re introducing, in a way that we have truly a full stack enablement in place.

In the last few years we’ve also seen the market evolve where the prosumer market around drones has gotten better and better, and there are other fixed sensors that are coming in from cameras that might be already located on an industrial worksite, that we need to process that sensor base input. Our vision at Kespry is we’re notjustgoing to process Kespry drone data only, we should be able to naturally process data coming from third-party drones, as well as fixed sensors that are on industrial worksites, so that we can really understand the condition and state of all the physical assets on an industrial site. That’s where we think the expansion opportunity will continue to emerge, where Kespry drones, other drones that are plugging into the Kespry architecture as well as other sensor-based input can all be processed inside of our Kespry Cloud. That we’re already seeing with some of our biggest customers in the market today, Shell was an investor in Kespry in the Series C, as well as the customer now, where we’re processing data that’s coming off of refinery sites where there might not even be a drone flying at all. It’s fixed sensor based input they were processing in our cloud for the exact purpose of digitizing the stream of IoT information that’s already present in these physical locations.

I think the convergence of the data that’s coming from drones, as well as the data that’s coming from fixed on the ground sensors, will again enable a more complete view of these digital twins that are emerging in these industrial work environments, and that’s where we see a greater amount of value that we can generate in the market.

Are there any special considerations that come into play regarding the governance around the data, particularly if you have clients that may be outside the US, it’s hard to think of image data as proprietary, but in many respects it kind of is, particularly if you’re in agriculture or materials. What are some of the security considerations or concerns that the customers may express around the areas of data security and governance?

When we fly today, we ensure that we’re really flying, as far as data collection goes, directly above the assets that a customer’s given explicit permission for us to collect that information around. That’s a precautionary measure on our part, to ensure that the privacy of whether it be a residential home, as well as an industrial worksite, is maintained despite the fact that we might be collecting high resolution imagery in a way that you can make all kinds of analytical decisions from it.

So, we do explicitly create a geo-fence around the area that we’re flying, and that the drone will never fly beyond the geo-fence when its collecting information. Then we ensure that when we’re collecting information within the geo-fence, we have explicit permission of where we fly. That has been the way that we’ve been able to be successful, not only in North America but also extending into Europe, because we abide by the privacy regulations that are already in place, so that there’s never a moment where there’s a concern that privacy privileges were broken by a drone collecting a lot of high-resolution sensor-base input, from not only industrial asset, but that we also just fly over residential roofs to understand the condition of those roofs, but where there’s property surrounding that roof you could make a lot of inferences on the insights of what we're collecting from that data source.

We’re super-careful about this exact topic to make sure those privacy regulations are met, particularly when we’re flying over both residential, as well as industrial assets.

I know there’s some satellite-based companies including a company called Orbital Insights, that traffic in data that can be tied back to investors, so they’ll look at the level of fuel tanks and try to track ships in the global oil supply chain. They’ll look at the number of cars that are parked in shopping malls to get a sense of what the foot traffic is, both for retailers and for real estate, so it is pretty interesting that we’re introducing this completely new dimension of intelligence and data from the world.

Well when you think about these sources of data Ed, what’s interesting to know is that these are layers of resolution that are going to co-exist with each other, no surprise there’s satellite imagery that folks like Orbital Insight are generating on a very large scale broad basis, of where the condition of large scale assets look like. Then you get a little further down in the layer of where you’re collecting this insight, where airplanes and helicopters have been historically flying over assets to get another layer of resolution in detail. Drones fit in at that next layer where you want to be able to detect call it centimeter and millimeter level detail, which no satellite nor airplane would be able to do naturally, at a cost basis that makes sense for most data collection needs.

Then frankly on the ground there’s fixed sensors as well as smart phones, that are able to collect all kinds of insights just from the sensors that are existing onto smartphones. All of these compound and play nicely together, because its creating a level of detail that goes all the way from satellite to smart phone, where we are able to generate the physical reality capture in a meaningful way, to get insights delivered to an organization. And a drone is just a big piece of that puzzle, it's not the only piece of the puzzle, it is a part of the overall need of collecting these insights in a more reliable way, and we’re just glad to not only be able to provide an industrial-strength solution for not only the capture of that insight using drones, but also taking the other sensor base input and converging that in our cloud experience for our customer base.

I think it’s really interesting, what you guys are effectively doing is, you’ve introduced this new type of data, in effect this master sensor, this meta-sensor that is providing a look at a completely new dimension of existing operations. So, in that respect its quite distinct from wiring up, or sensoring up say an existing plant, or retrofitting a plant with sensors. The point I’m making is the value prop is pretty straightforward, right, because this is an overlay, this is data you’ve never had before, and I’d be interested to get a sense from the experience of you and the team, you’ve gone after some heavy industry in mining and construction, pulp and insurance, I’d love to get your insights on what some of the maybe initial challenges were in evangelizing the data, what were customers looking to be proven before they bid? Do you think it’s an easy sell, or are there some certain types of proof points that early customers look for from you? I’d be interested in whether that differs across industries.

This has not been an easy sell, by any stretch of the imagination Ed, particularly when there is a fair amount of incumbency in the status quo of how mostly survey grey data was collected historically, and how it was used. I think the moment that the light bulb really turned in many of our industrial customers minds, was when they realized that this was more consistent, more reliable, frankly cheaper, and generating a level of insights that’s more frequent than you would typically do, with a once a quarter, once a year survey. That’s where we built a pricing model that supports it, our customers have the ability to fly the Kespry products, and generate the data products that we’re mentioning on an unlimited basis, because it's an all-in subscription, a singular subscription fee and you can fly and generate the amount of data that you need on an unlimited basis, for the unit of the drone effectively operating on the industrial worksite that we’re providing services for our customers on.

So, that behavioral shift occurred when it was understood that this was something viable, it could be safely accomplished, that there was an ROI that was demonstrable, and you’re doing it in a way that it changes the way that the nature of the work was historically accomplished, where you were previously doing a spot-check on this industrial asset that you might be trying to understand the asset value of, or predicting what the future maintenance value is on an annualized basis, and suddenly now you’re flying over a mine site and having real-time input on the inventory changes that are occurring, as much as on a daily level, on a stockpile by stockpile basis. That changes the way that work gets accomplished, in the construction section when earth works projects get accomplished and we’re doing topological assessments, it’s no longer what the plan was, and then how we bid it as the construction firm, and where the project ended at the end of the construction; literally you can see the progress and the changes over time, in a time-series model that’s being generated off of the physical condition of that construction site, using the data that’s generated off the drone flying more frequently on that site.

So, the type of work and the nature of how you can respond to it, has been very much an eye-opener for some larger industrial customers. The full side of it is, it’s all very early, it's all very much the way that the future of this work will be accomplished in the next decade or two, and we’re as I like to say in US baseball parlance, the earliest innings of the ballgame, but so far so good in terms of how much progress we’ve made.

This is really interesting. Are there jobs for instance that are potentially replaced by your technology, and is that a factor in some of the inertia that you faced early on? I’d love to get your thoughts on the impact of technology, and the expanding capabilities of solutions like you guys offer, where you're essentially automating this data collection, which may in the past have been somebody going out in the truck, or doing a visual inspection, what are the implications of this technology?

We’ve thought about this quite a bit, and the natural question here which always emerges, does this replace an Automator weigher job, or does this create new opportunities? There’s two angles where I believe it actually is creating new opportunities, the first is the productivity of the labor force goes up exponentially, and I’ll give you a good example of this. When a hurricane hits anywhere in the hurricane belt today, particularly these last two hurricanes seasons it was pretty interesting to see the level and extent of the damage, as most storms are now hitting the Gulf Coast for instance at category 3, 4, and 5 levels, these are catastrophic events that are occurring. When hail hits the hail belt in the mid-west, it’s pretty catastrophic with what hail is doing as far as damage goes. In both of these scenarios what we’ve realized is that the catastrophic response from an adjudicator adjustor, a CAT response team, you can only cover so much without enabling technology like this.

So, a good example, last year the CEO of Farmers who is a customer of ours, said that without drone tech they were able to cover three homes per day with this traditional tool of manually climbing the roof, and getting the assessment of what the damage was for hail-related damage on a residential home. That same adjudicator that’s doing industrial work in this case can now be scaled in terms of covering with a drone, about three homes per hour versus three homes per day. So, now you’re seeing an 8 to 10x productivity improvement in terms of what an adjudicator can accomplish on a day-to-day worksite, or a residential home or residential area that they’d want to cover on a single day, than was previously possible without drone-tech, like the ones they’re using from Kespry.

We’re seeing that productivity rise, we’re also seeing the skillset change the labor participation rate in a way, where some skills of someone who is now doing their job with the drone, versus doing their job without a drone; the ones that are doing it with drones, and others similar sensor-based technology that I’ve highlighted earlier, their labor participation, as well as their wage-rate is increasing, because now the skillset is more needed in the market and people are able to use this as a levelling up in the workforce itself. So, today we haven’t seen the moment where this kind of automation is reducing the labor force, or reducing the wage rate, in fact its increasing the participation, and increasing the value of that work to-date, and driving greater productivity. So, generally the use cases that we’ve been involved in are driving an upward mobility in the industrial workforce, versus creating downward pressure in the market.

Could you talk about where you see the greatest value coming from use of drone-based data collection in the future? It looks like you guys have established a solid beachhead in several industries, but I’d love to get a sense of how some of your customers are thinking, or how you and your team are thinking about where can this go over time?

I think the biggest area that we see opportunity is in the industrial inspection scenarios that have now emerged, particularly as thermal-based input is now very much a real offering. So, historically if you look at commercial property, as well as light industrial manufacturing assets, to know the condition of that asset it would be optimal not only to get high-resolution imagery, but to overlay a thermographic view of where that asset stands. So, if you were say for instance in Illinois today and you wanted to do a commercial transaction on commercial properties, in a state law I believe that you have a thermographic view of that asset, specifically the roof itself. If you’re collecting that on a manual basis, that’s pretty laborious work to detect if there’s a membrane leak of the roof, to check that there’s water-pooling, to detect if the industrial refinery there might be structural corrosion versus cosmetic corrosion. In all of those use cases, you can now have a thermographic view being generated by having a thermographic sensor complement the EO sensor that you might be flying on a high-resolution imagery basis, and mash those two together to create a digital twin that also layers in thermographic insights.

So, when we’ve done that it opens up all kinds of interesting used cases for industrial inspections, as well as going into a fair degree of solar use cases, asset condition use cases, refinery use cases where thermography gets layered into the physical inspection scenario. We’re doing that work right now in both industrial worksites, as well as commercial property, because we know that there’s enormous opportunity in those areas which were frankly the work that we’ve historically done, even in the mining space or the residential roofing area. So, that’s kind of where we’re focused next, as we get into the entire value industrial use cases that are more and more prevalent and accepted, with not only drone tech which is a big part of it, but even taking existing fixed sensors and processing that data in the Kespry Cloud.

That’s really interesting. So, ultimately you are looking at not just visual, but you’ve got thermographic and potentially other types of sensors, and all of this data of course can be correlated with the ERP systems, or existing analytic systems that the enterprise may have. Who do you see as potential stakeholders in developing new solutions? I know you’re working with a number of insurance companies and I think that’s super-interesting, is there a way that the use of this technology, and the increasing data extends more deeply for instance into the financial sector?

Right now, a lot of our focus is in these industrial sectors where the leaders that have responsibility for the line of business function, for getting operational efficiency on an industrial set of assets that they might have, coupled with regardless of what the title is, there’s someone that has responsibility for digital transformation inside the organization, and that might be not necessarily in the same role as the line of business leader. But the convergence of those two folks inside the organization tend to be our buying center, we tend to focus on the line of business leader, as well as the head of digital transformation for the customers that we serve. We can create a lot of market opportunity by just delivering these high value capability to those leaders in the organization, and they’re very much seeing that value quickly being generated, like we see customers expand from initial one or two unit opportunities, to multiple units, multi-year deals in the first year of commercial operation with Kespry.

Now, what’s becoming interesting is, the data that we’re generating is pretty much now helping us to understand the current state of that industry in a meaningful way. The first time we’re able to now do this is in the mining aggregate space, and specifically aggregates we’re now measuring 25 percent of all of the production volume of aggregates material in North America. Having that level of insight starts to create all kinds of opportunity downstream, we have not considered opportunities of how that gets packaged as a product in financial services, mainly because our primary focus is on serving our customers directly, and being able to generate the value that I ascribed in the earlier portion of this discussion, by the industry use cases that we’re targeting. But these data products have derivative capability to them where you can understand the physical change at a broadscale basis that we couldn’t do before, at a level of detail we certainly hadn’t had available to us. That will start to redefine what the products are that are relevant for these industries.

It’s a future that’s further out for Kespry, right now we’re just targeting serving our customers with what their needs are directly, versus what these derivative data products could look like over time.

Customers in similar industries, I assume they benefit from your experience working with multiple participants of the industries, but are they engaged or interested in sharing best practices, or insights across the industry? I guess the question here is, are you seeing the type of data sharing and cooperation that you’ve seen say in the insurance industry, or financial services around credit scores, with some of the data that you guys are collecting?

I think that will happen, and as I mentioned earlier it’s still early innings in the ballgame to date, and so that’s not quite where the markets that we serve are today, but we’re increasingly being asked to create more in-depth benchmarks of the data that we’re collecting, so that people can see even on an anonymized basis how industry peers are operating in the sector that we serve. That’s still a few years out as well because again, you’ve got to have enough data generated, it's much more possible today to create a data product like that in the mining aggregate space where we have some level of ecomancy at this moment, but I think the other industries will soon start to follow, particularly for these benchmark insights that can be potentially generated.

It’s again not the area that we’re primarily serving the market in today, mainly because we’ve got a lot of direct work to do with our customer base, and that’s where Kespry’s focused.

Absolutely. I’ve got a question about interesting technologies or start-ups that you may be looking at, are there any that jump out at you?

Sure, there’s definitely companies that have been doing incredible work in the industrial sector, that we have been working with, potentially even partnering with, and see a lot of similarities in terms of what the market opportunity looks like. A good example is uptake, there’s a lot that they’ve done in terms of data science, particularly when it comes to creating the economic model that drives a lot of the key organizations that they serve today. If you think about what Future IoT and Uptake are doing, a lot of the physical reality capture and the analytics that surround the physical asset themselves, Kespry can naturally provide that to complement what a solution like Uptake looks like today. That’s a very natural extension of what Kespry is doing, but we would not go into the detail of the economic model on the data science workloads, that they have been delivering into the market in a lot of ways. So, that’s a complementary example that we see.

You mentioned satellites, in the case of Cape Analytics in the insurance space, Cape is doing large scale broad based analysis of before and after satellite imagery, on what the physical condition of an entire region could be, based on what happened before and after the hurricane hit. They would never have the optical detail to dive into any specific asset itself, but you’d have broadscale coverage used mostly for underwriting and risk purposes, that could be complemented with the data that Kespry generate for claims management. So, we see plenty of opportunities continuing to emerge, particularly as various layers of these digital twins are emerging in the industry, where we don’t necessarily have to be the end-all provider of all of these capabilities, we have to plug into where the industry needs other data sources, and frankly other application experiences that complement Kespry, and those two are just good examples of things we’ve already seen in the markets that we serve today.

That’s a really helpful insight. We had Ganesh Bell from Uptake on the podcast a couple of weeks ago, they’re doing some great stuff over there. Well George, this has been super-interesting and informative. I always like to close out our podcast with a recommendation on a good book, or a resource that you can share with our listeners?

I’m still in the early stages of start-up mode at Kespry, so if you think about where things stand where Kespry is, and how we’re building a solution to market, it's still the business of creating a market and driving the opportunity for how that market can scale successfully. So, of course one of my favorite books in the topic is, ‘The Hard Thing About Hard Things, by Ben Horowitz, I would say it’s essential reading for what it takes to build a business, when there isn’t necessarily a pre-existing market that is immediately available to you. I’ve been very much inspired by the journey that Ben and Mark took long before they became venture capitalists, and that was one of my favorite books in terms of what the personal ups and downs are, in terms of building a business when particularly there’s no pre-existing guide book, or market analysis that gives you enough confidence to say that this is something that’s significant and that can be real.

We’re in a very similar place right now in a lot of ways, because we know that that market is possible, we have to execute and create it. In fact, I look at this as having to lay down some of the tracks, as we’re going over those tracks and creating new territory for ourselves. It’s been quite an experience for me in the last two years to be able to accomplish as much as we have in those two years, but I always go back to Ben Horowitz’s book, ‘The Hard Thing About Hard Things’, when the going gets both easy and tough along the journey.

That’s a great recommendation, and with that we’re concluding our conversation. George it’s been a pleasure speaking with you. This is Ed Maguire, Insights Partner at Momenta, once again with another of our Edge Podcasts, and our guest this afternoon has been George Matthew who is the Chairman and CEO of Kespry Technologies. Thank you again for joining us.

Thank you, Ed. Really appreciate it.