In this episode, we discuss the state of edge computing adoption in manufacturing. We also explore the most common edge computing use cases in OEE optimization, predictive maintenance, asset condition monitoring. 

John Younes is Co-Founder and Chief Operating Officer at Litmus Automation. He is in charge of operations and growth for the company and draws on considerable experience working with start-ups and early stage companies. Litmus enables out-of-the-box data collection, analytics, and management with an Intelligent Edge Computing Platform for IIoT. Litmus provides the solution to transform critical edge data into actionable intelligence that can power predictive maintenance, machine learning, and AI. litmus.io

Erik: Welcome to the Industrial IoT Spotlight, your number one spot for insight from industrial IoT thought leaders who are transforming businesses today with your host, Erik Walenza.

Welcome back to the Industrial IoT Spotlight podcast. I'm your host, Erik Walenza, CEO of IoT ONE. And our guest today is John Younes, CEO of Litmus Automation. Litmus provides an industrial edge computing platform that is purpose built to simplify the complex challenges of collecting, analyzing and managing data from hundreds of diverse machines. In this talk, we discuss the state of adoption for edge computing solutions in manufacturing. And we also explored the most common edge computing use cases, including OEE optimization, predictive maintenance, and asset condition monitoring.

If you find these conversations valuable, please leave us a comment and a five-star review. And if you'd like to share your company's story or recommend a speaker, please email us at team@IoTone.com. Thank you. John, thank you so much for joining us today.

John: Yeah, thanks a lot, Erik, for having me. Glad to be here.

Erik: So, John, before we get into the topic of Litmus and edge computing, for smart manufacturing, we'd love to learn a little bit more about your background, and particularly what led you to set up Litmus. So I think you have some experience studying in China, you've been a strategy consultant, you've also been a consultant for a coffee startups. What was it that led you now back to San Francisco and to set up our Litmus?

John: So myself, I am one of the cofounders of Litmus. We started about seven years ago, was really where the genesis of the idea came from. As you were describing, I was doing a master's program, which was a Global Entrepreneurship Program, where we studied around the world: four months in France, four months in China, and four months in the US. So that's where I actually met one of my cofounders there. And his background was coming from Rockwell Automation.

So really, a lot of what we were trying to address was coming from his experience there. So he was working on various oil and gas pipeline projects where they were collecting data off of pipelines sensors, and looking to store a simple database where they can then analyze the vibration of the pipeline to then predict when it might break.

So he thought that because of the development time that it took them to actually get the data off of these pipelines, various sensors and devices and make sense of that data, was quite a long development time. So he thought if he had a scalable easy-to-use implement type system that can essentially communicate across any type of device or sensor and make that data available to applications that can consume it in a way that they were looking for, so that was really the high level idea where it started. But obviously, in that time frame we've evolved into now focused within edge computing, but still solving some of those higher level problems that my cofounder was addressing at that point as well.

Erik: So you basically started ideating during this program, and then you set it up together with your cofounder. Any other cofounders or is it the two of you?

John: We have a third cofounder as well. So the link as well as with myself, so I actually have known Sasha, who is our third cofounder for about 15 years now going back to high school, and even University: we studied together. So his background is more on the financial side of things. And so he's representing the company and leading up on the finance and investment side of things for Litmus itself.

Erik: Where are you today in terms of maturity? Are you like A round, B round? What's the current status of the company?

John: Yeah. So in total, we've raised about 12.6 million in funding. Most recently, a little less than a year ago, we raised our Series A backed by Mitsubishi Corporation. So $40 billion company from Japan, they led our A round. And so in that time frame, we pretty much doubled in size for about 55 people in the company. We're headquartered in San Jose in California. We have offices in Toronto, Canada, as well as in Tokyo, Japan for our APAC operations. And now we've hired some people as well in Germany.

Erik: Before we get into the business, how is COVID-19 impacting you? I know some companies is accelerating the adoption of their solutions, others it's putting projects on hold, what does that impact look like for Litmus?

John: Yeah, so April and May, obviously, with manufacturing shut down for the most part, that did slow down some of our business itself. But now since things have opened up more, it seems like the business is actually accelerating quite a lot. So we have customers that are approaching us every day now, companies that are quite bullish on their industry 4.0 initiatives and looking to get started quite rapidly.

So I think COVID has really opened up a lot of people's eyes within manufacturing, that they realized that they had some of these digital technologies, then they may have been able to continue their operations a bit more fluidly. As well as on the other side with potentially some, some of their sales being affected, they're looking at ways that they can become more efficient, or drive down costs, which start with really their manufacturing operations. So things have really been picking up quite heavily since I'd say about the end of May or June or so.

Erik: Maybe we can start with who you're focusing on. So I understand is manufacturing, do you have a particular focus on discrete or process or any particular industries, or is this fairly horizontal?

John: Mainly our customers are more discrete manufacturers. We got started a lot working with automotive companies, OEMs, as well as some tier ones. So that's really where we cut our teeth, I would say within manufacturing, but now it's quite across the board in terms of our customer base. So we have customers within food and beverage, oil and gas, CPG, healthcare, electronics manufacturing.

Since COVID hit in March, we've actually made a conscious effort to focus more within essential services, and essential type manufacturing, so food and beverage, CPG, healthcare, life sciences. And that's where we're seeing actually a big wave coming from those kinds of companies for us currently. So a lot of our focus is within those areas currently. Automotive is a bit slow at the moment as well as oil and gas.

So this focus has actually helped us out quite a lot. And we're seeing those companies actually being quite interested in using these kind of technologies as their businesses are obviously changing on a daily basis now.

Erik: Because if you look at this first target segment, automotive OEMs, I mean, those are primarily Fortune 500 companies. So, they have a pretty mature technology vendor base, their procurement processes are probably fairly challenging, as far as a first target segment that's not maybe the easiest one. What was it around your value proposition that allowed you to get into these large companies where their existing technology vendors weren't able to meet some requirements?

John: Yeah. So, one of the main reasons why automotive was quite a good fit for us, because there is a high level of automation as well as a lot of disparate type of equipment on the plant floor, and high volumes of data that are coming off. So edge computing fit was a pretty natural fit there. Since they were looking at a way that they can connect across all of their different assets, they can process more data locally at the factory itself.

Also, a lot of them are a bit more advanced when it comes to some of their in house capabilities with data science or some of their cloud computing strategies. So they were more advanced in terms of being able to leverage a platform like us that can produce a lot of value in terms of data that it can get off of the factory floor and make it available to those cloud-based systems that can really start powering their cloud or machine learning strategies. So they were very data hungry, and they had a lot of sources of data, so that made it a pretty natural fit for us in terms of getting started first.

So obviously, yes, going into these larger target companies, it is obviously more difficult for startups to get into. But also, our products really solve a lot of those complexities, more for larger scale type deployments. So it really speaks a lot more to the value where as you're trying to scale these deployments, you need some sort of product or platform in place that can really connect across everything that you have with a high degree of manageability so that their IT teams can really manage all of these deployments, be able to orchestrate applications push machine learning models to the edge and create easy templates that they can configure systems tens of thousands of systems at once and instead of just sort of one by one configuring or building out custom protocols or custom drivers every single time. So, for us it actually, the larger the customer, the more pain points that they have, which our products are a perfect fit for.

Erik: So I see here collecting, analyzing and managing data, so you're able to ingest data from a wide variety of PLCs, DCSs, CNC machines, sensors, etc. And then you put that into a platform where they can then build the vertical applications for specific solutions. Is that the case? Or do you also have standard vertical machine learning solutions for predictive maintenance or for something else that you would sell in as a solution package?

John: So right now it's mainly the horizontal platform that they can easily build these kind of vertically integrated or vertically specific application. So it really enables them to either use their existing in-house applications within the platform, or really quickly develop new applications. Within our Litmus edge product, we have a marketplace concept. So essentially, we have various partners or even ourselves, we're developing different apps that can reside within this marketplace. Because we have this normalized data coming from all these assets, we can readily deploy these applications and make the data available to them.

Think of it as sort of an edge computing app store, where out of the box you can leverage whether it's predictive maintenance for a certain type of machine or certain type of process or production line, or have algorithms available for different types of systems or processes. So in that way, customers will be able to leverage them more out of the box. And those are developed either by us or from partners as well. So that's a big part of our longer term strategy is developing out this app store for edge computing.

Erik: I assume that the customers are probably not so eager to put their apps on the App Store. But is it then mostly like other solution software vendors or ISVs that want to use this as another channel for their software? Or is it system integrators are consultancies? Who would be the primary partners for developing these critical apps?

John: Exactly the ones who mentioned so either ISVs, or system integrators or consultants data science companies that have built out these kind of applications based on their domain expertise. But customers themselves, they can actually create their own private marketplaces. So if they develop something they can host a private marketplace, and then make that app available to all their factories, for example, from the same marketplace. So there's also the private marketplace concept where, let's say you're an automotive manufacturer, and one factory has developed a specific type of application, they can then put it in the marketplace and it can be available to all the other factories as well.

Erik: We were working with a tier one automotive components manufacturer right now, and there's so limited visibility in terms of what software is being deployed in different factories. So it's really just people saying, hey, did you know about this? You're thinking about building that, you should talk to some factory in Germany because I think they have a similar solution, which it’s a little crazy because it means that there's a tremendous amount of redundancy in terms of the software. So this private marketplace approach I can see would be quite valuable for a company that has a big global footprint when it's hard to coordinate between people.

So you're based primarily right now in North America, you mentioned a few of the markets that you're in, what is the state of, let's say, edge computing for manufacturing? Is it that there's 10% of companies that are ready to do this that have the data science capability and the rest are followers? Or what percentage of the market do you think is ready to begin adopting these solutions today?

John: Yeah, I would say it's still very much early adopters at this phase. So it's probably the 10 to 20% or so that are actually looking at edge computing or looking for an edge computing platform specifically or have their own in-house data science teams that can actually leverage the data or have developed machine learning models. So even from last year, maybe we will get a handful of customers that are saying I have this machine learning model and I'm looking for a way to deploy it at the edge and provide the data to it as well as the closed loop back to my control systems.

But even this year, I'd say that the demand for that or customers coming with that specific sort of problem or inquiry is probably gone up three or four times. So definitely the market is picking up and becoming more sophisticated quite rapidly. But again, I'd say it's still very much the early adopters. Most customers, they're just looking to get data and send it somewhere, or just set up some sort of simple KPI dashboards to have some analytics and some condition monitoring, or remote monitoring of their assets. But more and more the requirements are coming up for edge computing, I would say.

Erik: And how do you see this? Is it still the majority of projects that are kind of in a lighthouse or pilot phase, and they're really exploring how this solution might work for them? Or what percentage of customers would now be ready to really ramp this up to the operational scale across multiple factories and really integrating this into core operations?

John: Yes. Actually, this year, we've seen much larger deals, customers that are looking to buy licenses up front for 30, 40, 50, factories even. Whereas last year, I think most of it was doing some pilots or one factory, two, three factories, and they weren't necessarily talking about a full scale rollout across the enterprise. But this year, it's become a much more common theme where customers are looking at deploying this across the whole organization.

And at the end of the day, I think that's becoming a bigger thing for these manufacturing companies is they do want to create some sort of standard and have a standard platform that they can really scale it across the whole enterprise, because that's obviously where you get the most value of IoT, when you have everything connected across your whole production globally and centralize everything. And that gives a lot more control and value back to the corporate IT teams that can really actually then do something with that data, whether it's presenting it up to the top floor or providing better ways for the shop floor to be able to actually get more real time insights into how these machines are performing in real time.

Erik: That's been a problem with the pilot phase over the past few years, is that somebody would deploy a solution on a few machines, and the way it operates on a few machines is quite different from the value that would be generated across an entire factory or larger scale operations. There's not just in a linear path. It's quite different value proposition.

You mentioned, buying licenses for maybe dozen factories, is your business model licensed by factory or is it per machine that's integrated or data flow? What would it look like when you're figuring out the right scope for a particular customer?

John: Yes, just this year alone, actually, at the beginning of the year, we launched quite an innovative business model, which is based on number of manufacturing sites. And essentially, we created three different functionality packages for customers. So based on where they are in their digital journey, whether it's the first package, which is essentially they creating this common data layer, where we include all the drivers to connect to all their machines, and then the integrations where they can push data out, as well as some processing and store and forward type capability.

And then the next package, which includes the marketplace, our management console, where they can manage all these edge nodes as well as our analytics engine. And then the third phase, which includes more about orchestrating applications, running machine learning models, big data integrations: so it's really meant for the most sophisticated customers, I would say.

So customers can essentially choose one of these three packages. And it's based on number of sites. There's a certain amount of data points, but the data points are quite high. So it's meant as more of just a protection for us. But mostly the customers are able to utilize the full scope of the license for their factories without having to go over that. And then this really makes the conversation easy, where they just choose a package, and then based on number of sites, this is what you're going to pay. So it's not complicated at all for the customer. And then we can really discuss more of an enterprise wide pricing and incentivize them to do it at more sites by giving them much more economies of scale, obviously, as they purchase more sites at once.

Erik: Yeah, I've seen a big trend towards trying to simplify business models. I think that's very necessary for driving adoption. Initially, it was quite complicated.

John: Yeah, definitely. So, before we used to go on more of an instance based license, so you would base on number of gateways and then within that gateway, you can connect a certain amount of devices or data points. But that became quite difficult to actually to even have that discussion with customers. Because at the end of the day, it's very difficult for them to figure out upfront what their architecture looks like, how many gateways do I need, how many data points am I really going to be connecting across everything? So it became more of an architectural discussion than an actual commercial or pricing discussion at that point.

And we even had customers that were going around the business model to figure out ways to making poor architectural decisions by even using gateways where they shouldn't be, or trying to use too low cost of hardware. And it was really actually affecting the architecture at a higher level. So now this way they don't have to worry about number of instances. We don't care go even under install 100 gateways or 50 gateways at a factory, and go ahead and make whatever architectural decisions that you want, which are going to be the best for you without having to worry about number of software licenses.

Erik: Do you pay the cost of data capture? Or is this anyways so moving and storing the data? I mean, is that something that like the customers are paying anyways, or do you just have to, include that cost in your in your business do and your cost structure, regardless of whether it's a small volume or large volume for particular factory?

John: So we're an edge first company. An edge, we deploy at the edge. So everything we do is running on premise within the factory, or running within their own local private data center. Our management piece can run in their own cloud, so that would just be running in their own cloud, or it can also run on premise as well. So for us, there's not really any additional costs in terms of storing additional data on our servers, for example, because we run as an edge first product.

Erik: Okay, so that makes the decision simple, then yeah, you don't have any additional cost of allowing them to steal. Who would be the primary person that you will be speaking with? You have, on the one hand the factory GM, who has very powerful voice about what goes into a particular factory, but maybe not a very powerful voice around aligning technologies across a network of factories, and then you have the conflict between IT and OT, and maybe different priorities between those functions and your spending those quite strongly. Who would be the primary contact point at a customer? Is it corporate, or are you going directly to the factories, and then is it more on the OT or the IT side?

John: So we target more of the corporate side of things. And it's generally these larger companies have built out industry 4.0 teams or advanced manufacturing teams. Or a lot of times, even, we're seeing it being a big driver for this because they want access to this data and they also want to be able to manage everything centrally. So IT is actually becoming a much more common buyer for us, even if it's a director VP of IT, or anybody who's reporting up to the CIO.

Whereas it makes it a little more difficult if we go one by one to individual plants. We do have those scenarios where it's a factory, or maybe a set of factories that are looking to solve a specific challenge. And we can provide a solution to that challenge where it becomes more than based on the specific ROI that they're going to get based on that use case. Rather than trying to sell a holistic platform to IT or to this industry 4.0 teams where they really see the value to get started quickly and have quick time to value where they can as well be able to scale this across the enterprise very easily. That's where we're seeing a lot more success, I would say.

Erik: So you mentioned that you're your edge first, you're primarily deployed on the edge, but then you can use the customer cloud if necessary. I guess in some of these situations, we're probably dealing with fairly high volumes of data, and potentially fairly high computing requirements. Is it quite typical to need to move some of the data to the cloud for storage or for processing? Or in most cases, are you able to do everything on the edge?

John: So in some cases, we are able to do everything on the edge. It really depends on the use case. I think eventually those use cases as well, they'll expand upon it and there'll be looking to push data to the cloud where they might get into some more centralized edge and start doing things around machine learning. But if a customer is just looking to collect data and get some real time KPIs to the operators or to the plant manager, then they're able to be able to fulfill those requirements at the edge or set up just sort of some rules and alerts to notify based on the data in real time as it's coming through.

But then when customers are really looking to enhance or combine with their cloud strategies, and I think that comes more to the sophistication of the customer, and where they're at in their digital journey. So in that way, we can really then start providing more of those integrations to the cloud, where they can then do even more with the data. And then once they've processed or analyze this long term data that bring those findings or algorithms back to the edge where you can then optimize it in real time so you don't have to deal with latency and sending data back up to the cloud and back so really taking more of the decisions at the edge on the real time data as it's coming through.

Erik: And then your primary use cases, I know your team has mentioned a few. So we have OEE, predictive maintenance, asset condition monitoring, machine learning, which I guess could be used for a lot of different situations. Are these the primary four or is there then like a long tail of quite niche use cases that your customers will also be deploying?

John: I'd say those are the primary four. A lot of customers are looking for OEE. So, being able to have accurate OEE, reporting and metrics across all their machines, so we're seeing that is pretty common, and almost most customers are looking for that, or just asset condition monitoring where they can monitor the machines and have some better visibility, or even over the processes as you mentioned.

And then there's some other use cases that are more around, I'd say traceability of the products as they go through the manufacturing process, or even having to do with quality related to even vision processing systems on the plant floor. So analyzing, let's say, for a food and beverage company, for a beer producer analyzing a good bottle versus bad bottle on a high speed bottling line so they don't have to manually inspect each one after the fact. So using vision processing, they can easily identify these with an algorithm running at the edge, and then take that bottle off the line based on if it's a bad bottle. So we are also seeing these use cases around quality or traceability as well, I would say, are some of the other ones that we're seeing as well.

Erik: And when you're engaging with the customer, had they pretty much already figured out the business case and the justification for a particular use case and then they're searching for the right technology enabler? Or do you also have to walk them through the thought process of what would be the potential impact of machine vision, for example, and support that justification for the investment?

John: Yeah, so for the most part, we're trying to find customers, obviously, that have a specific problem that we can help them solve with the technology. Maybe they don't know what the technology is, but they at least have a business problem because that's really when these projects will get legs and will take off and scale when there's a real ROI, or business challenge that they're trying to solve or a goal they're trying to get to in terms of a certain efficiency or being able to track specific metrics.

So really, we're looking for customers like that. We don't necessarily have a large consulting team in house, obviously, that we can consult customers on. But we have a good breadth of use cases and industry experts as well that have been within the manufacturing world for 30 years, 40 years, and they can really help guide customers in the right directions or give them suggestions based on what we've seen other customers doing as well.

Erik: And then on the deployment side, there's going to be quite a range of, especially when you get down to sensors, for example, really quite a broad range of different data sources that you might want to connect to. What would the deployment or integration look like? I mean, is this a fairly time consumptive process where you'll have a team on site for a couple of weeks doing an assessment of all the devices that need to be integrated and is there a fair amount of kind of customization to integrate with different new sensor classes? Or what would this look like for a medium scale deployment?

John: So everything that we have is productized. So we even had customers that have done 40 factory deployments on their own in the year and a half, two years, which you won't really find in a lot of the other larger products, which are more SDK driven, or you have to pull pieces together multiple products to get what we have. So everything is ready out of the box from all the drivers we make them available to the customers. We have the widest library of protocol drivers for PLC, CNC machines, robotic systems, different sensor types, so they can immediately connect to all their assets and then they can easily set up their data workflows for processing.

We have integrations pre built, where they can send data to either cloud based systems, big data platforms, enterprise systems, with no coding required. And then our analytics engine, again, it's all workflow based driven. So everything is all user interface driven. So it's a very modern user interface, very intuitive, that even an automation engineer, we can turn them into a sophisticated business analyst with this kind of product, it's that easy to use.

So sometimes customers will even do deploy it on their own, or if they want us to support them, we can go on site and deploy their whole factories in even a couple of weeks or less, where they can have all their machines connected and some dashboard set up within the factory, which really gives them a very quick time to value. Or if it's for very large complex type deployments, or if they need some applications built out or some data science assistance, then we have some integrator partners or data science partners that we can bring in to help fill the voids that they might have within their own resources.

Erik: I see you have two primary products. So the one is the Litmus Edge, which is the edge computing platform, and then you Litmus Edge Manager, and I suppose that's the front end management tool. I guess you'd have quite a wide range of users so it would want to have access to this. Is this also something that the operational users, so would there be shop floor manager, a maintenance manager that would want to have access to this on their mobile device for real spotting issues in real time is this? Or is it more of a higher level management tool that would be used by the factory GM, and maybe IT?

John: So there's really multiple users or multiple type of consumers of our product that can get benefit from it. So as you mentioned, it can be the maintenance manager that's getting real time KPIs, or even alerts based on how some of the machines are performing, so they can do their jobs better. Or it can be a plant manager that's looking at OEE, or how much product that they're producing on an hour hourly or daily basis, for example, and give them the sort of KPIs that they're looking for so that they can do better reporting, or operate their factories better.

And then it can also be data scientists that are just looking to get access to this data so they can build out models or machine learning models on top of the data. And then it can be as well, even from the corporate level where you have executives that can look at these kind of interfaces, where they can see how all the assets are performing and all the factories from one central interface. And then IT can have one layer where they can manage all the devices as well across all the platforms.

So it really brings together a lot of different groups all the way from OT to IT event to the executive level, which I think when you have a product that can tie together those many different types of key stakeholders, that becomes extremely powerful. And that's really where they look at taking that at a much larger scale.

Erik: Data security at factories is obviously taken quite seriously. Are there particular headaches that maybe not that you need to address, that do your customers need to address in order to make the data kind of fully available to the stakeholders that need it? Or do you find that most of the customers that you're working with already have the required policies in place and are basically able to just deploy the technology? I mean, do you have a lot of conversations with legal or with any other stakeholders around how to secure data when they're making this available to such a wide group of stakeholders?

John: Yeah, so that's always a conversation that we have to have at some point with the security teams within IT, where they take us through all the necessary requirements that they're looking for. So we've had to have our product audited even by third parties or customers will even audit themselves. So we've handled security, something we've handled very well within our products, and also the flexibility that they can integrate it with a lot of their IT policies, whether it's their user management systems, or LDAP authorization mechanisms.

So it's very open from that perspective where they can easily utilize some of their existing security tools and policies as well. So that gives the customer a lot more flexibility. And they obviously really like that point of view as well. Data security for us, especially because we're running as an edge first product that also gives a lot more assurance to the customer, that it's running within the four walls of their factory so that data is doesn't have to necessarily leave the factory itself, or whatever we're running is pushing to their existing cloud based systems which are already secured as well.

And they have their own transport layers for that, which we're leveraging. Even if it's a cloud first product, then I think those become a lot more of a difficult decision, especially if you're trying to push your own cloud on the customer. The fact that we can run everything within their own infrastructure, that gives us a lot much of a larger benefit in terms of quicker deployments or quicker conversations that we're having with customers around security.

Erik: Well, let's walk through one or two deployments. Could you maybe pick one or two representative ones and try to walk us from start to finish, who were you first speaking with, and what were the problems being solved, and then walk us through the deployment process?

John: So [inaudible 36:45] secure it. It's a division within Sango bath, which is a Fortune 500 manufacturing company. They have about 1,000 factories worldwide. Secured is one of their divisions, which has 45 factories, they make glass for automotive vehicles. So they're a tier one supplier. And for them, it's such a large company who's grown through acquisitions and is very geographically dispersed. The equipment that they have in their factories is very different. And the systems that they have in each factory is very different.

So for them, the first thing they were looking for was a platform that they can easily connect across all of their different types of assets. That the requirement and challenge, number one. But then they also, were looking for a flexible product that they can easily make some quick dashboards in the factories without any heavy coding which can be leveraged by their local engineers in the factory itself.

But at the same time, that same data can be usable for their internal data science teams that they can work on within their big data or machine learning strategies. But at the same time, corporate IT can also manage everything. And then as they build these models, they wanted something that they can easily deploy their own models or applications on a single edge platform.

So this was driven from an industry 4.0 team. We were working with their industry 4.0 engineering leader, who was really the champion for us, because this could really provide him with a product that can fulfill a number of different requirements for a number of different stakeholders. So those are always the most ideal teams for us when we can work with these kinds of industry 4.0 teams that have a strong vision, and they definitely do. And so they have already deployed in over 20 factories, and in just over a year, year and a half or so, with no system integrator, everything has been deployed and developed on their own with just some assistance from us.

Erik: And you said how long did that take to deploy in 20 factors?

John: In just over a year, year and a half or so.

Erik: Do you happen to know how many resources from their IT team are required in those deployments? Are we talking about three people full time or just to get an indication of the effort involved from the customer side of deploying?

John: Yeah, so in in either a specific region or even in some factories, they have more advanced industry 4.0 engineers, they call them. So either we would train that person who could then deploy it in their factory or group of set of factories, so they really had some dedicated resources that were a bit more advanced from an industry 4.0 perspective that they knew how to leverage these products.

So in almost every factory, there was one of these kinds of engineers which made it much easier for us obviously, so we could train them. But then if they had some other factories, which maybe did not have the right skill sets, then they would have us come on site for maybe a week or two weeks, and really implement everything for then.

Erik: So on a faculty basis, one or two weeks is insufficient provided you have the right skill set there?

John: Yes.

Erik: But what would be like a typical cost range if you're looking around? I guess you have probably quite different pricing structure for one factory versus 40 factors. Can give you kind of a range for what deployment might look like from a cost perspective?

John: Yeah, so most customers go for that middle package I described, which is our growth package. And that is about just under $50,000 per factory. So it makes the conversation very easy when we can say we can solve this problem for you for $50,000, where they don't have to worry about anything else. So that's generally for list price one site, but obviously, we get larger discounts when the customer brings, whether it's starting at 5 factories to 10, 20, 30 and upwards of 50, then there's discounts that will apply to that so that they can get some economies of scale from using the software.

Erik: So being able to connect all the devices on the edge across 20 factories for $1.5 million, that's almost a no brainer for a large enterprise.

John: And saying that they can do it within 6 months, 12 months, that's something that they're very much up for.

Erik: I mean, would it make sense to go to another deployment? Or are they fairly similar in terms of the processes followed?

John: So I'd say that's fairly similar for our larger customers. We're working with the one largest health care providers where they build different types of healthcare supplies or equipment for the healthcare industry. And for them, they were after a specific application where they could more monitor their products and inventory. So we worked with them to actually help them develop this specific application, which took, I think, 2-3 weeks or so. And then now that application was then put in their private marketplace so it could be leveraged by the first factory.

And then now they have five other factories that have the same problem and they want to use the same application. So they're looking to then just make this product available to end application available to the five factories. So they started with more a specific application or use case that they were looking and scaled that to the exact number of factories that had the same problem. But then obviously, now that they've seen the value of what the platform can do, they're really starting to think about a number of different use cases that can not only be used in these factories, but far more of their, they have over 50 different factories around the world.

So it's very much a sticky product. And once you get your hands on the product, the possibilities are endless. And they just keep coming up with more use cases that they can develop on this on top of the same platform, which is always great for us.

Erik: I suppose, on the one hand, it makes sense to continue focusing on discrete manufacturing. But are any of your customers pushing for deployments in warehousing, yard management, fleet management? I suppose then you're getting into situations where cloud is more necessary if you have a more dispersed environment? But are there any other environments that you're considering covering or that your customers are pushing you into?

John: Yeah, I would agree with that statement. They're more either cloud based, or there's already vertically specific applications available for them or solutions, which are much more tailored to those industries, so it makes it more difficult for us. And the assets are much less disparate, which is one of the main headaches that we help them solve. So that's why manufacturing is an ideal fit.

But we've seen some success within oil and gas, for example, where they're looking at an edge computing first mindset also because they have a lot of remotely located assets or remotely located areas with poor network coverage, for example, where they need to have edge computing. So that's oil and gas, where we're doing upstream, either oil rig monitoring, or midstream pipeline monitoring.

Other areas we're dealing with, again, remote assets, like energy or wind farms, windmills, as well. So those are some other areas that we started to work on with some customers as well, where they need a much more edge first strategy.

Erik: Last topic would be technology trends. So what are the key technology domains that you are working on or monitoring as far as moving the core technology forward?

John: So what we're really working on is making the products very seamless to deploy in any way that the customer is looking to. So whether it's as an appliance based model or operating system on a gateway device, or a virtual machine that they can run it within a server or existing hardware. And then the newest one, which is a lot more customers are asking about, which is having support for Docker; so running our system as a series of Docker applications that they can easily deploy it within their own environment.

So Docker as well as Kubernetes are becoming topics for our customers, which they're bringing up as, I'd say, newer technology trends. It's still early, I'd say, customers understand the idea, but sometimes there's a lot of times where they don't actually have the capabilities to use it. But they're definitely intrigued by the idea of Docker containerization technology.

Obviously, we're continuing to build out drivers. So we already have over 80% coverage of the market in terms of different types of PLCs, or industrial systems that are out there. So we're continuing to build out our coverage of different drivers as well as integrations on the other side.

And then expanding much more on the analytic side, so really trying to provide as much more value as we can around analytics or machine learning at the edge. So those are some of the areas and providing the ability to run in a flexible way any type of machine learning models that customers end up building on the cloud, whether it's an Azure machine learning or Google TensorFlow, or can be MATLAB runtimes, providing them with a number of different flexible ways that they can deploy these models at the edge.

Erik: Well, John, it looks like you've built a great business here. And I think you have the timing just right. I think we're really just at the cusp of hopefully, the hockey stick slope here. Any other topics that we haven't touched on that you think we should cover?

John: Nothing that comes immediately to mind. I think we've gone through quite a lot.

Erik: Well, John, I really appreciate you taking the time to walk us through the business. Like I said, I mean, really it's a great model and a great value proposition. So wish you all the success in the world and just thanks for taking the time today.

John: Great, thanks a lot. Appreciate you having me on here and looking forward to further conversations. Thanks for the time, Erik.

Thanks for tuning in to another edition of the industrial IoT spotlight. Don't forget to follow us on Twitter at IoT one HQ and to check out our database of case studies on IoT one.com. If you have unique insight or a project deployment story to share, we'd love to feature you on a future edition. Write us at Eric dot Relenza at IoT one.com

Rate transcript quality

Erik: Thanks for tuning in to another edition of the industrial IoT spotlight. Don't forget to follow us on Twitter at IotoneHQ, and to check out our database of case studies on IoTONE.com. If you have unique insight or a project deployment story to share, we'd love to feature you on a future edition. Write us at erik.walenza@IoTone.com.