According to a report from Forbes, 75% of CEOs of the Fortune 500 List believe that the pandemic is going to accelerate their pace of technological innovation and adoption.
This rapid push toward digital transformation is unprecedented but is crucial for companies to consider given the effects of the virus on the workforce, process, and development time. The scalability and ease-of-use of IoT solutions provide companies a competitive edge in rapidly launching and managing their initiatives.
We partnered with Ubidot’s CEO, Agustin Pelaez and See Insights LLC’s President to discuss implementing IoT.
Click on “Ask a Question” to jump to a specific time in the conversation to hear an answer from a question you want to know!
Transcript:
Danny DeLaveaga: Alright, I think we’re live. And I’m excited to be here today. So, this is the second episode of a product development webinar series, and I’m very excited today to be talking about web application development with regards to the Internet of Things. And so, getting started, I’m here today and honored to be here today with Chip McClelland and Agustin Pelaez from two different companies. We have See Insights’ Chip McClelland and Ubidots. And so, I guess before we get started and get going, I’d love to just give you guys a brief introduction. So, I’ll introduce myself. My name is Danny deLaveaga, and I have a mechanical engineering background, but a lot of development lately and a lot of experience with IoT, with my last company, Breadware, and now this company, Ioterra. And so, brief I’ll start with you Agustin and fill in the dots from there, but in general. So, you have an electrical engineering degree and worked in the aerospace industry in Germany, correct? Before founding Ubidots in Medellin, Colombia?
Agustin Pelaez: Mm-hmm. Exactly.
Danny DeLaveaga: And you have a management degree from Grenoble as well?
Agustin Pelaez: Right. Yep. [chuckle]
Danny DeLaveaga: Alright. And you’ve been a maker at heart for quite some time now, and Ubidots is kind of like this very interesting… Or it’s positioned in the middle of the maker world as well as the professional enterprise world. I’m excited to hear a little bit more about that.
Agustin Pelaez: Of course. Thank you.
Danny DeLaveaga: And then, Chip McClelland, seasoned software development expert, over 14 years at IBM before it was acquired… Or sorry, before working at Red Hat and Red Hat was acquired by IBM.
Chip McClelland: That’s right.
Danny DeLaveaga: Okay, got it. [chuckle] And then you founded See Insights approximately three years ago or so?
Chip McClelland: Yeah. Actually, a little before.
Danny DeLaveaga: A little before. Okay, great. And a maker as well. I think you have over 80,000 views and impressions on the various projects that you posted on Hackster.io today, which is amazing. [chuckle] So, has that been a hobby or a… Or how long have you been contributing to the IoT space?
Chip McClelland: That started about five years ago. I had gotten my undergrad degree like yours, mechanical engineering, and I got a Master’s in Systems. And then I put all that on the shelf for a while, while I was working at IBM and Procter & Gamble and other places, kind of more in a business role. And, to be honest, I started to have a hankering to get back into some technology. And the great news is, it’s become much more approachable. Things like Arduino, the Maker Movement, 3D printing, there’s a lot of things that have made it easier for people to get back into those things, so I got into it. Then once I started making some stuff, I started thinking about, “Well, what could I do with this?” and that’s how See Insights came about.
Danny DeLaveaga: Alright. Well, so just to get started, I’d actually like to explore a little bit the mutual history here, because you guys have had quite a history, I believe. And can you tell me a little bit about that Agustin?
Agustin Pelaez: Sure. Building on your intro, and thanks for having us, by the way. Pretty much like the three of us and I think for the same as most of people in the audience, we like to build stuff, build the great stuff as makers. So, as a maker, I’ve always been convinced that the creative force or the talent to solve most of the world’s biggest challenges is already out there. And it’s more of a matter of coupling this talent with the right tools, the right motivations, and the right resources. So that’s pretty much what Ubidots does nowadays, so we created this platform. First of all, going back to some history, we, me and my co-founders, all had an electrical engineering background. We began doing some products together, mostly in the healthcare space in Medellin, Colombia.
Agustin Pelaez: And that’s how we began looking for options to, sort of, where can I put all these data that we’re generating and go back to focusing on what we knew best at the time, which was hardware development. And back in the days, we couldn’t find anything that answered the requirements that we had, and that’s the reason why we founded Ubidots. You fast forward to today, there’s a lot of companies, like we used to do back in the days, that are building end-to-end IoT solutions and then use us as a means to go faster to market. And a lot of these companies just happened to be founded by makers like ourselves who are, like Chip just said, they discover a passion for something and to learn and to build things using Arduinos or types of tools, and then discover they can apply it somewhere else. So I think that’s where we begin to intersect, Chip. So I’ll pass it on to you for the rest of the story.
Chip McClelland: Yeah. And kinda like you said, I was getting to this Maker Movement. I started with the little tangible device, the Arduino actually could hold your hand. And I realized you could start throwing off a fair bit of data. You connect this thing up and start taking measurements, and I was looking for a place to put that. At the time, there was a number of different options out there, but I found about Ubidots, I think primarily because I was online, they do a very good job of documenting the “How to Get Started” guides, “how do things work.” There was an excellent set of tutorials and I started doing a few of them, and I realized that I could take advantage of what they had built and focus on the things that matter to me, which was finding the right device to build and collecting the right data and then getting it back to the back-end where Ubidots would take care of it and make it easy to record and present and save the data we got. So, I started reaching out to them a long time ago, I think about four years ago.
Danny DeLaveaga: Four years.
Chip McClelland: Yeah, and we were… Very quickly, I decided, “Hey, this part of my solution is done, I’m just gonna… I’m gonna take advantage of what they’ve built and I’m gonna focus on building the other things.” And I think, to a lot of things… One of the things about Maker Movement is there’s so many things that you can go down a rabbit hole on. You really have to pick where are you gonna focus your time and where you’re gonna put your expertise.
Danny DeLaveaga: Yeah, and I think the Maker Movement is rather interesting. It’s a maker kind of, early starting points, and a lot of these projects get turned into businesses, and actually, I’d love to… Is that along the lines of what happened with See Insights? Just as a brief overview from what I’ve seen, you’re taking all kinds of applications for sensor to Cloud integration and providing value to organizations and businesses and cities.
Chip McClelland: Yeah, so I think a lot of it was just born out of some passion of mine. I love to get out in the woods and ride my mountain bike. I think it’s one of those great releases and a chance to get away from the office, get out from the screen and enjoy the fresh air. And what I realized is in a city like Raleigh where I live, there’s tremendous pressure around open space. There’s a lot of development, a lot of building, a lot of families are moving to Raleigh, ’cause it’s a great place, and we have to balance the needs of that development with the need to give people a chance to get out in the woods. So I took a chance and reached out to some of the park managers that work near my house, and found out that one of the things that they needed was help in getting a handle around their visitation, in part because that helps them better manage the park, in part because it helps them justify the value they deliver to the community and to…
[pause]
Danny DeLaveaga: Hello? Not sure…
Agustin Pelaez: Yeah, I think it’s on Chip’s side.
Danny DeLaveaga: Oh, wait Chip, apologies, I think your microphone got messed up. So we’ll move on. We’ll move on. I think this is a good segue actually, just getting going Agustin to talk about web application development and how that’s related to IoT in general. Every single product that’s collecting sensor data and sending up, sending the information to some dashboard or some place where analytics can be done or good business decisions can be made, has a lot of different parts and… Yeah, looks… Chip, looks like your mic is muted, so if there’s some kinda… Yeah, we’ll go ahead and move forward, but I think it would be useful, Agustin, to talk a little bit about just the general architecture of an IoT solution and then we can zoom in on how web application becomes a large… And transfer of information from hardware to the cloud becomes a big part.
Agustin Pelaez: Sure.
Danny DeLaveaga: So in general, any IoT product is broken down by it’s sensors or actuators that are collecting information and controlling environments, and then these platforms that can help move data from the actual hardware to a cloud environment, the communication layer and connectivity, and then some kind of custom web application development on top of that. From the Ubidots’ perspective, can you give me a little sense on what you’re seeing on the hardware side first, how are people collecting data, how are people sending data, what are you seeing right now? And then we’ll zoom in on the software side.
Agustin Pelaez: Sure. So starting with the hardware, and just to recap what you just said, the way we break it down is, as you said, actuators and sensors, then you need something to do the telemetry up to the cloud, and then that communication might happen over a Wi-Fi or ethernet, but it could also happen through a more specialized IoT connectivity provider, such as Sigfox, LPWAN and now Cellular IoT. And then once it gets to the cloud, it’s where we can zoom in later. So, going back to the hardware part, what we see being used the most it’s still Wi-Fi. So Wi-Fi is still king. I know there’s a lot of new ways of connecting things to the cloud, but if you look at a factory, maybe even ethernet, it still reigns in terms of the market share that we see for devices. And in terms of the brands or the enablement, chipsets that are being used in hardware in general, I would say it’s a mix between self-built and off-the-shelf solutions.
Agustin Pelaez: So the self-built ones are pretty much companies that design their own PCBs, they incorporate chips from Expressive, Particle or maybe just some generic Wi-Fi chip, and then use this to go to the Cloud. These are the ones that are growing more because they have these custom designs because they’re ready to scale. And then on the other side, we see still a lot of devices that are already working under a specific brand like a weather station from weather.com or an industrial sensor from ncd.io which is a great company that already built all… It took care of all the encapsulation, the brand, the connectivity, and then you can buy a device and just put a brand to it and then start selling from day one.
Agustin Pelaez: You got also some reference designs from the enablement companies like Particle, they just released their Tracker One, which is a smart device that already has encapsulation and protection, and it’s easy to start doing something like on a production level with their chip inside. So, yeah, I would say that the biggest takeaway at the hardware portion is to ask yourselves, do you want to build something or buy something, depending on your expertise and the time that you want to wait until you go to market.
Danny DeLaveaga: Okay. And Chip mentioned Arduino as a quick and early platform for collecting data. And it looks like you have over 40,000 active applications right now, different devices sending data up through your platform. Are you seeing a lot more of these connected products being kind of development kits and are there things that you’re seeing launched fully developed custom PCBs?
Agustin Pelaez: Yeah. It’s a very, I would say, asymmetric comparison because on one side, yes, we do have 40,000 and more IoT applications that are sending data as we speak, and yet the majority of that is we work under the Freemium model. So, majority of these are still prototypes for students, teachers that are using Ubidots to provide a specific training, nowadays more with the virtual world. So in these cases, of course, there’s a lot of prototyping platform usage, such as Node-RED, Arduinos, or just a simple Python script that’s running on a Raspberry Pi. And that’s, I would say, a lot of application of a small count per device per application. Or if you go to the industrial side, where we now have more than 400 users or customers, then a particular customers might have 2000 devices that are using a customized PCB. So yeah, I would say that depending on the stage of your project, you’re likely to use either a prototyping or at the other end, an industrial-grade technology.
Danny DeLaveaga: Okay, awesome. And so, moving from hardware to software, you’ve got all kinds of devices sending data up. From a perspective of an IoT entrepreneur or a startup, what is entailed in all the steps that go beyond getting the data to the Cloud? What’s entailed in web application development in general? And maybe we can start with just some of the things that you went through. Chip, let’s see if you’re microphone is back on.
Chip McClelland: I hope so.
Danny DeLaveaga: Yes, alright. We’re good. [chuckle]
Chip McClelland: I’m very sorry about that.
Danny DeLaveaga: Oh, don’t worry…
Chip McClelland: Something happened, I had to reboot my machine. I apologize.
Agustin Pelaez: It’s all good. So we kind of broke down the layers of an IoT solution, so from hardware to cloud. And now, we’re kind of focused more on what that software part entails, and then… Is that correct, Danny?
Danny DeLaveaga: Yes, exactly. There’s all these… All kinds of different ways you can collect data with both development kits and custom PCBs, and we’re seeing it in all kinds of environments. And if you’re a startup or if you’re a company looking to grow and help a park for instance like See Insights is or a city collect data on usage, what’s entailed after the actual hardware sensor part? How difficult is it to give the city something that’s useful? And can you just walk us through some of the process that you went through as you got this going on the software side?
Chip McClelland: Yeah, I think the big thing for me was that I wanted to rely on some platform partners to do a lot of that backend work for me. As a small business with a certain focus, I had to be choiceful about where I focus my time and effort. So, the big thing is when the data comes off that device, the first thing is you have to have a reliable way of collecting that data and managing the fleet of devices. That’s the first piece of software in the web, it’s fleet management software, things that allow you to monitor the health, to provide updates, to be able to kinda track utilization. And I rely on Particle for that. So, Particle is the fleet management software interface for me. I can see all my devices which are online, how much data they’re consuming. I can monitor the data flows from the device, and then send that data flow on to endpoints. And then for me, I send them on to Ubidots. And Particle even ensures that that data is encrypted end-to-end, from the device all the way until it’s received by Ubidots.
Danny DeLaveaga: And I guess, kinda breaking back a little bit, I’m just curious, how did you already decide that you wanted to use a platform? ‘Cause there are other options, correct? You could’ve built up an ecosystem to collect this data, a Cloud environment to move the information, and then build out a user interface that can be displayed to your customer base.
Chip McClelland: That’s right. The way I did it is by, I think Agustin said it, breaking a lot of things. I went through all the stages. At first, I built the devices that stored the data on memory cards, and I’d go and collect the memory cards every couple of weeks. Then I went to a big step forward and I got Bluetooth, so I didn’t have to get out of my car and open up the box, I could drive by the device and get the data. Then I added cellular modems and I was able to send the data back, in this case, to Ubidots. But it was very much me managing those devices, and all of them came with significant pain.
Chip McClelland: If a device, for example, is not constantly connected, and let’s say it goes down, you don’t collect your data about every two weeks, you could lose two weeks of data. So, by having the devices connected, and by using a platform that manage things like cellular data plans and manage the quality of the data flows like Particle does, just made the thing possible for me. I was never gonna be able to scale my business beyond just a couple local parks without some sort of a global partner. So now, if I turn on one of my devices in New Zealand, where we have some today, or in Rwanda or in California, or in the North Carolina, it connects and it sends data, and I don’t have to worry about that. So, in the long run, it was because I tried all the other things and realized that I wasn’t gonna be able to scale without some sort of a platform as a base.
Danny DeLaveaga: Really, really interesting. Thanks for kinda walking through the evolution of where you started with SD cards moving all the way up to where you’re at now. [chuckle] And that’s for four years, you mentioned?
Chip McClelland: Yeah. Well, I spent a good two years of that going through those painful iterations and… Especially if it’s on the woods. So you go out there and you discover the box, it’s buried under the ground and water got in and it’s flooded and like, oh my gosh. It wasn’t pretty. And it’s the same thing on the backend. Sending the stuff to Ubidots, for example, one of the things my customers wanted is a portal so they could log on using a username and password and they could see the data specific to the devices in their park. I could have built a custom portal and dashboard for them, and I could’ve worked out the user authentication and log in and transferred the data from Ubidots to, let’s say, a Flask or some other kind of portal kind of product. But again, if you’re trying to set up dozens of parks and maintain these dashboards and user logins across a number of different clients, you need a platform or a scale, at least while you’re small. And at some point you get big enough, you start to take on some of these things, but where I’m at, I like the fact that I can rely on these partners to provide those capabilities.
Danny DeLaveaga: Yeah, just thinking about it, if you had to do that yourself or hire… ‘Cause you’re a software developer as well, but if you had to hire someone to do that, what is actually entailed, I guess, in the backend of a platform like Ubidots? I’ll direct this one at you, Agustin.
Chip McClelland: Sure.
Agustin Pelaez: Sure. So, I’m gonna speak in more generic terms, not just Ubidots but in general, I think what you wanna look for when deciding to outsource this portion of your IoT business is a partner that can provide at least five components. So, the first one is the data management. So, are you able to stream data to a particular protocol, MQTTs kind of getting the most difficult one in IoT, but you also have HTTP, TCP, which is also useful for similar applications, it’s more lightweight. First, is this data ingesting. Second… And by the way, I can give you example of each layer of providers that do it themselves, so that you get a sense. So in that first layer of data ingestion and storage could be someone like a Mongo Atlas, for example. You could literally go to MongoDB, create an account and start sending data to Mongo and then it’s gonna be stored, and that’s a first layer. But then, what do you do with that data? So second layer, it’s more about device management, and this is where an IoT platform typically intersects with some of the partners that are in the IoT ecosystem. Because depending on the type of connectivity you choose, you’re gonna have a device management platform, like Chip said, a fleet management platform. So, it’s a place where you can know…
Agustin Pelaez: It’s not so much about monitoring the business process or assets that your solution is about, it’s about monitoring the devices themselves. So that’s what we mean to as fleet management. Can I reset my device? Can I reboot it? Can I turn it off? Can I do a mode firmware upgrade? What are the [24:15] ____ levels? And examples of this are Particle, which already includes not only the connectivity but this device management layer. Balena.io, it is great for embedded Linux devices, for example. So, yeah, you wanna make sure that you have this device management layer. In the case where you’re going directly to the cloud, in this case Ubidots, then there should also be a layer of, “Can I see a list of all of my devices and see if they are online?” at the least. So we got these two, and then you got the rules engine. You wanna be able to act upon your data based on some specific thresholds or some logic that you define. You wanna be able to either trigger an alert at your customer site, be it a business process or simple as an SMS, but you wanna make sure that you have this rules engine.
Agustin Pelaez: And then, last but not least, is the whole making-sense-of-the-data portion, which I would split into visualization, real-time dashboards as Chip said, which is something more about what happened in the past, how we analyze trends, and what’s gonna happen in the future. I would say visualization is more about the present, what’s going on with my fleet, with data of my devices. Analytics is more about what happened at a particular time in the year, and can I predict what’s gonna happen in the future. So data management, device management, rules engine, visualization and analytics. I think that entails most of it.
Danny DeLaveaga: Awesome. Thank you for breaking that down. ‘Cause I think the way that I’ve kind of envisioned mobile apps before, it’s like, “Oh, I’m gonna build an app for my product.” It’s a lot simpler than it actually kind of maybe. That being said, when… I’m gonna make you answer this one which… [laughter] What is an example of when it’s not a good idea potentially to use a platform? Why would someone just actually decide to build something from scratch? I know for at least consumer products versus more industrial use cases like this one right here, there might be a need for more custom visualization, custom user interface. How does that interact with a platform like yours, like Ubidots?
Agustin Pelaez: Okay, yeah. The build versus buy question I think it’s an ongoing one across many different type of enablement platforms, not just IoT. I guess the answer is mainly, first, are your requirements so specific enough that you couldn’t build something with the tools that are out there, and then do you have the budget for it? If you’re looking at, at least, 6-12 months of engineering, hundreds if not thousands of hours that will get into a solution to get it up and running, and that’s perfectly okay. We’ve had customers that at some point say, “My application is just too specific. I’m a large company. I’m gonna go with just piecing together all the small components, and then Ubidots is gonna be just an API to get the data and store it. But then I gotta build an entire thing on my own that might be ending up costing more than what I pay Ubidots subscription.” And that’s okay because they have the budget to do it. I think that’s one of the components of build versus buy. And I know maybe Chip, I know this is your day job as well. You can [28:10] ____ some lights to it.
Danny DeLaveaga: Thanks. Yeah, this is great. You’re talking a little bit about pricing here, and I’d be very interested because we got from everywhere from ingesting the data, doing the fleet management, doing applications and making rules based off of the data and alerts and things that happen, visualizing it, and you already mentioned Chip that you offer this as a service ’cause your customers asked for it. They’d like to see the data at each specific park, and then the analytics predictive side. Can you talk a little bit about how you price through this thing, and then also how much does Ubidots cost.
[laughter]
Chip McClelland: Well, I think the first piece is, and just to add to the building versus buy thing, a couple of quick things. First is you gotta think like what’s your competitive differentiation. Where’s the unique value that you’re delivering, and in some cases, that unique value is in a very highly customized interface or a very personalized experience, and maybe those are things that draw you to build your own. But I think that for a lot of things around industrial IoT in particular, the value is in your experience in where that IoT sensor meets the real world. You’ve learned how to operate a device, in my case, sense a hiker walking down a trail in the woods when it’s raining, when it’s cloudy, when it’s cold, when it might be a deer.
Chip McClelland: There are all sorts of things that you gotta think through and it’s… That’s the expertise that differentiates it. That’s why the other parts I’m happy to consume rather than trying to build, but consumer device or something like that might be exactly the other way around. As far as the build versus buy piece on the economics, I have this customer conversation in my day job too, I work for Red Hat. We sell free software and sometimes we have conversations with people about, “Why don’t I just download your software from the Internet? It’s open source and off I go.” Sometimes if you wanna get… I live in Raleigh. If I wanna get to Los Angeles, the first thing I do is I buy an airline ticket. I don’t build an airplane because it’s more important for me to get to Los Angeles than it is for me to have the satisfaction and experience of building an airplane.
Chip McClelland: Some people like building stuff, that’s fine too. But I think there’s real value in keeping your focus on what the objective is. The way I sell my devices, I basically sell the device so the parks buy the physical hardware. And part of that is because I don’t have any control over that device once it leaves. It’s gonna be off in the woods somewhere and in a park. I might have limited access to it. I certainly don’t have the ability to ensure it’s physical safety or proper installation and maintenance because I just can’t physically be there. They own it and if something happens, then either it gets fixed or replaced. As far as the service part goes, there’s two parts. There’s one part is the connectivity plan, the cellular data plan for each device, and the second part is the cost of storing that data and doing all the things that Agustin just mentioned about the back-end value that Ubidots provides.
Chip McClelland: And round numbers, it’s about 10 bucks a month per device for both. And I think a lot of my customers the conversation is around, “How do I justify that cost especially to a park, which they don’t have huge budgets there. They are a public organizations, they have to be frugal and prudent in the way they manage the public’s funds.” But I give you an example. At Umstead State Park, a 5000-acre park down the street from me, everyday they sent a ranger on a 35-mile drive around the perimeter of this park to manually collect the counts from the counters that they had at each of the entrances. And the justification for placing sensors there and charging them 10 bucks a month per sensor and the cost to install it was more than offset by the time and the gas and the wear and tear in vehicles to obviate that need for that 35-mile drive. And not only did we save them labor and time, we gave them better data. Before they’d get one data point a day, and whether it was 22 hours or 26 hours would depend on traffic and when they collected the data before, and when they collected the data on the day of the… Let’s say the next day. We now give an hourly resolution of data, and we give them more reliable data because we can tell them if something…
Chip McClelland: If one of those devices doesn’t report in a two-hour period, there’s an event that’s triggered by Ubidots that says, “Hey, I haven’t heard from this device, it’s supposed to check in every hour, I haven’t heard it for two hours,” and it proactively reaches out to me and lets me know so I can fix it. So I think there’s real value in this, and it really is about the justification, the time savings, the better quality and the better justification of the budget, the better management of the park.
Agustin Pelaez: That’s very cool.
Danny DeLaveaga: Yeah, thank you for explaining that use case scenario, it really does paint in a picture that’s really easy to consume and I understand and see the value here. I’m gonna jump back and… So we have some questions from the audience, and by the way, anyone who’s watching, feel free to ask anything on the side here, and we’ll address it, but we have a question from Anise. Does Ubidots offer a remote firmware upgrade OTA? To you Agustin?
Agustin Pelaez: Sure. So to that, we… Because of the diversity of the type of devices that our customers are using, we typically leave this firmware OTA feature to the device management platform that they’re utilizing. If it’s Balena then you push a Dockerfile down to the device, if it’s Particle, then you push an Arduino file to the device. So it really depends on what you’re using. So, we don’t support it off the shelf. We do have cases where customer… We have an API Gateway builder, which is… We call it Ubi functions, so it’s just a way for you to put your own Node.js or python code, a server-less environment running through an API. So we see some customers actually using that to create an API that their devices then hold to download the latest firmware. So that’s something that’s more built on the side of the customers using our tool because of the diversity of the devices that we have to support.
Chip McClelland: And now, let me just offer one other thing on that, just to give you an idea of the complexity that comes with that OTA capability. So you think my devices are solar-powered, they’re cellular-connected and they’re off in the woods, all over the… Basically around the world. They sleep 99.9% of the time. They wake for about 90 seconds every hour, and a lot of that is negotiating the webhook that sends the data off the device. The device management software not only has to be able to accept, let’s say, a firmware update and hold it in the platform ready, but it also has to coordinate how to deliver that at the appropriate time when the device wakes up, when it connects, without disrupting the flow of data that’s essential for that connection session. And it took me a while and…
Chip McClelland: We had talked about earlier I went through all sorts of pain before I finally got to it, but when you finally get to the point where you can push an update and then you see that device, that update go out to a fleet of sleepy devices over a matter of a couple hours and you see them reporting back as being updated, it’s a magical thing. [chuckle] And I… The first time it actually worked as advertised, I was blown away by it. So it is something that it’s an important thing to look for from your platform provider. You’d have to have some serious technical chops to build that yourself, some people do, particularly if their device is maybe more of a Linux device, more of a constantly connected, the way something has reliable or broadband internet connection but for these edge devices that are low power and low bandwidth, it’s really something.
Danny DeLaveaga: Thank you. Thank you very much. So, I’m gonna step to another question by Taylor, so… And I’m not sure if you’ve had experience Chip, with AWS IoT or Azure but, what is the difference between their default device management dashboards and the dashboards provided by a platform like Particle?
Chip McClelland: Yeah. I’ve played with it some, and I do… Using… The nice thing about the webhook kind of approach or using a platform like Particle is that you can very easily send data off to two or three places at once, some of my clients actually require me to send data, particularly, some of our international clients, require that we keep a copy of the data in country. And so, we’ll send it to Elastic or something like that or the AWS IoT engine. The big difference is that Particle’s platform for device management is very opinionated.
Chip McClelland: It’s specific to the requirements of that Particle device, and particularly the cellular or WiFi connected. Whereas AWS, IBM, Watson IoT or Google is gonna have to be more general, more generic, because it has to accommodate devices from all number of different manufacturers. So… I’ve used a couple of them, I’ve particularly used the Watson IoT one but I… In the end, since I’ve standardized on Particle for the device itself, I went with the platform that… Yeah, it’s very tailored and focused their device management on that platform. So that… I don’t know if that answers the question but that’s my experience at least.
Danny DeLaveaga: Definitely. Definitely. I think and actually… Oh, great Taylor. Taylor says it does. [chuckle] So I have a question… I have a question again, about your business model. And I guess we could kinda wrap this up with a couple of them here for you Chip, but for the park managers and the instructions that you provide on how to install the device and how to provision the device. Do you provide these? And how does that affect your business? Is that a big part of the business? Or is that a problem?
Chip McClelland: No, it’s a great question ’cause it kinda gets back to, at one point, this is kind of a hobby I was doing with local parks, and at some point I said, if I wanna make a business of this, I’ve gotta have the ability to scale. I’ve gotta be able to put these things in places I’ll actually never physically see. And so the good news is that if you build a device in a thoughtful way, it can be installed by the park managers and done in a very professional way. Most parks have a dedicated facilities management staff. They are used to installing and managing the physical infrastructure of a park. I provide detailed documentation of the trail counter, or bike counter or a card counter, here’s how you install it. And I have to admit, I’ve been very much impressed by the quality of these installations. Some of them, they came up with ways of preventing the snow plow from pulling the tube and ripping the device off the stand by all sorts of really kind of ingenious ways, so, I’ve been very happy with that. The nice thing about a connected device is basically you just follow the instructions and when you get to this point, you turn it on, and as soon as it connects, just send me an email.
Chip McClelland: And then basically the fleet management software takes over from that. As soon as that device comes online, it’s identified as part of a product group, which pushes a specific firmware update, which makes the connection. It immediately sends data to Ubidots. The Ubidots platform has a way that automatically creates a new device once it receives that first data packet, and then based on the data that’s provided, can put that into an organization which adds it to a dashboard and off you go. So the backend piece of it, if you do it correctly, is pretty straightforward, they basically, literally turn it on. The only thing I’ll tell you about IoT in the outdoors is, this is cellular. We take for granted the amazing $1000 phones that we carry in our hands, like the latest Apple. These are not $1000 devices, and they’re trying to connect and communicate in places that even your phone struggles to get a connection to, like out in a park, out in the woods. So getting… And actually getting these park managers to actually install high-gain antennas correctly, again, didn’t turn out to be as big a deal as I thought it was gonna be. So I give them instructions and they physically implement the infrastructure consistent with how they build the park, and we’ve had some good luck.
42:08 Danny DeLaveaga: Way Cool. [chuckle] Okay, so we’ve got a question from Alexandra here. We all know the true value isn’t the data, but the insights, are you thinking about improving your analytics engine or maybe add some, ML machine learning or artificial intelligence to the platform for you Agustin?
Agustin Pelaez: Sure, well, I love that question. We’re working on something very exciting, which is a new analytics model. So I’ve said, we already have some layer of descriptive analytics, which is, give you the ability to correlate data, what happened at a specific time, so not only real-time data, but also going back and looking closely at the data. But with that, a couple of requirements have arose from the current beta testers. So one is, they want to be able to load datasets. So if you think about IoT, IoT has been around for 10 or 20 years, if you bake in the end to end/GPS times. But what if you have a dataset of, say the earth’s temperature for the past 100 years, or an estimation of it. So we want our customers to be able to load these data sets that they have, maybe in a CSV file or an Excel file, and just use it to correlate it with the IoT data that they already have in their accounts. So it could be things as exciting as maybe you wanna track the dollar price for the past 20 years and see how it correlated with a specific share price that depended on weather variable, for example.
Agustin Pelaez: So these kind of correlations are still descriptive, I would say, but already adding some analytics power that, it’s exciting because of the surprise or the surprise factor of what our customer is gonna find out that they don’t know about because of these new tools. So we’re working on that, and then that’s kind of looking at the past and then looking at the future, are we able to predict based on that data? If I have 10 or 20 years of data, then suddenly I open up a door to train a model, be it an ML model or a traditional model, statistical model, why not, to sort of predict an anomaly or predict what’s gonna happen with a specific variable. So yes, we’re working on these parts. It’s a challenge, usually with the way we’ve approached machine learning, it’s on a per project basis, so if a customer wants to predict a machine breakdown or do some risks rating based on specific variables, then we’ve done it more like looking at a specific use case, but prioritizing it so that anyone can do an ML project has been a challenge, but we’re working on that and it’s an exciting part of our today’s jobs.
Danny DeLaveaga: Thank you, and I think you kind of answered another question from Jay. Any chance to build integrations to other data sources, specifically weather data? And you just mentioned that.
[chuckle]
Agustin Pelaez: Well, yeah, that would be through a dataset, but there is a way to do it already, it’s just… It needs some heavy lifting, it’s called Ubi functions. So you could, in theory, set up an Ubi function to run every minute and then pull data from IBM’s weather-channel API, so that you have the weather data, not from a particular sensor that you installed yourself, but from a live weather station that is offered to an API. So that’s already possible but it requires some heavy lifting. We’re wrapping it around something called plugins, so we will allow the developer community… And going back to our initial conversation, this creative force that we want to tap into is we want to allow these 40-60,000 plus developers out there to create plugins on their own. So for example, the USGS water service has an API to report on the water quality across major US rivers, that’s already there, but nobody is using it. I mean, the general public is not using it because it’s a very technical thing to do. So we are coming out with this plugin tool, so that anyone can be the plugin and then offer it to anyone, any other Ubidots customer, so that they can also harvest the power of such information that’s out there.
Danny DeLaveaga: Very cool. That’s actually… It’s really interesting, and it kind of goes to a quick question from Christina. Chip, you’re one of the most active users in the Ubidots community, which is… Well, the Ubidots community in general, and you have a whole community that’s interlinked with hackster.io, I’ve seen a lot of projects there. What’s your motivation to help others with the projects that you’re setting up, that aren’t part of See Insights?
Chip McClelland: Sure. The big thing is I… A couple things, first, I am very grateful to all the people that helped me as I got back into building IoT and connecting them to devices and actually delivering home projects and work projects and everything else, so I feel like when you receive that kind of free advice and the help that people have given, you wanna give some back. The second thing is, I do all my stuff open source. So every piece of hardware, every piece of software that I do is open source and because of that, I feel like documenting it and sharing it with people is part of the value, and a lot of times I get a lot back because people will say, “Oh, well, why do you do it this way? You could have done that.” and it’s caused me to redesign a piece of hardware or redo a piece of software.
Chip McClelland: But I never would have gotten that feedback if I didn’t share it through a platform like hackster or if I put a snippet of code up on the Particle site one, the Ubidots site. I have, for example, when I was designing the carrier boards to put these devices out in the parks, I have a thread that’s 18 months long with 2000 posts back and forth on Particle where we developed this board and at the end of it, I can’t claim that it’s my board, right. It’s something that we collaboratively did, and there’s a lot of good ideas in there, probably a minority of them are mine, but when I produced that board, I say, “Hey, anybody wants to jump on the manufacturing run, let me know. But here’s the CAD files, here’s everything and if you wanna take this and run with it, please do.” So, I think that there’s a lot of value that comes with that, and it’s something that I wanna make sure I’m contributing to, not just benefiting from it.
Danny DeLaveaga: Thank you for breaking that down. That’s a really, really amazing story and makes me believe in karma.
Chip McClelland: Yeah, well open source is all about karma. Let’s be honest. Yeah.
Danny DeLaveaga: Okay, so a question from Rain for Chip, do you find any challenges managing multiple platforms for a single product or do they work pretty well together, I.e. Particle plus Ubidots. We’ve talked about quite a bit today and would you prefer an all-in-one platform? I guess in a sense you’re kind of offering the all-in-one platform.
Chip McClelland: Yeah, I think that you have to… To some extent, you have to standardize because life is too short, there is… I listen to some of these podcasts, Amp Hour and Embedded and some of these others, and they talk about the Whizbang do… They have, I think called Tip of the Week from The Amp Hour, and you could go a mile deep on a new thing every week and it’s actually part of the fun of being in this space. There’s so much, so many new things coming. It’s so exciting. But I actually like the fact, for example, I have a few different generations of devices now in the field, and one of the things I like about the Particle platform, and then obviously in the backend as well, is that it’s totally backward-forward compliant.
Chip McClelland: So the software that I’m writing to run on that device, be it the four-year-old device or the ones I just put out last week, which is a third generation from Particle, it’s sync-code, and I think that’s something you should really look for. I think if you’re on a platform that is going through a lot of major changes and causing you to constantly rewrite, and especially as you start to deploy these things, and now you have to fork your development effort because you have, “Oh, these devices can’t upgrade to the latest version because they’re… ” Fill in the blank. That’s a problem. So I think that having long-term support, having a road map from a vendor that lets you have a four, five-year life cycle with one of these little devices is really something that you should be looking at when you decide which platform to go with.
Danny DeLaveaga: Oh yeah, that sounds distinct advantage, ’cause technology does move so fast and if you can get buy in. And what used to take care, for instance one of the last companies I worked with. What used to take $20,000 in development, just two years later it takes $2,000 and it takes half the time.
Chip McClelland: Yep, yeah, exactly. But these things are out in the woods, and in New Zealand, I can’t go get it to replace it with the newest version, so it’s gotta work, it’s got to work for some period of time.
Danny DeLaveaga: Alright. We’re coming up on the end here of our time, but I got another quick question from Alessandro, “I’m from Latin America and here, the IoT adoption is very low, mainly because of culture and investment. Can you share any business ideas to sell IoT solutions in this context?” I think that we’ll start with you, Chip, here on this as well. And maybe if you could tell us a little bit about how you got started selling in America and it sounds like it’s a global operation now.
Chip McClelland: Yeah, I think the first thing is IoT actually has some advantages in terms of it doesn’t necessarily require a large capital outlay. You can build these devices with off-the-shelf components that are relatively inexpensive and so some of the investments, like an industrial IoT or an IT infrastructure, like a data center, can be quite expensive, but this… I think you can get started and make that entry point a lot less painful. I think, for me, I always try to justify the cost based on the benefit that the customer is getting. They have to see it as a good economic decision. I gave you the example of the guy with the park, but we’re also doing, let’s say, an irrigation project in Rwanda where farmers are just now getting access to, basically, irrigation. I guess you could say pumped water as opposed to relying on rainfall, relying on runoff and they’re over-watering their crops.
Chip McClelland: And when they over-water the crops because they think, “Hey, water is great, more water is better,” but what ends up happening is they’re eroding their top soil, there’s a lot of fertilizer in the runoff, which they have to then replace and it’s obviously not good for what’s downstream. So you make an economic, or in this case a societal argument for giving people access to that technology and if you can make that argument well, they’ll invest in it. And the nice thing is that the cost of the devices and the cost of managing, like you say, come one way down. And for example, it doesn’t cost me any more to turn on the device and manage it in Rwanda than it does in North Carolina. So I can offer these things at a cost that it would just have been incomprehensible four or five years ago.
Danny DeLaveaga: I imagine the…
Chip McClelland: I think… Oh, go ahead.
Danny DeLaveaga: Rather experimentative in the way that you sell these initially as you’re learning your target market?
Chip McClelland: Yup. Yeah, I think there’s no short… There’s no way of shortchanging. Just going on asking… I actually go talk with the park superintendent or park facilities managers and ask them, “What are your pain points?” Like for example… I’ll give you one more quick one. A lot of parks have water infrastructure because they are outside the city, they don’t have city water, they don’t have city sewer, they gotta pump the water out of the ground. They have to hold it, clean it and provide it to the patrons of the park and the technology that they have on those things is 30 years old. And so one of the things that they’re looking at is a significant cost to take out the control systems for their water management, and replace it with modern industrial systems which are quite expensive. We’re able to add a device to an existing system that works perfectly well and bring the connectivity and reporting piece, so that they didn’t have to send people out to manually inspect the wellheads that are sprinkled all over the park and generally way out of the way. So I wouldn’t have known that that was a pain point if I hadn’t just gone out and asked for an interview and had a chance to ask people, “What are the things that you’d like some help on?” and then you generally can come up with a solution that they’ll find attractive.
Danny DeLaveaga: Alright. Well, thank you very much and I think that about wraps it up. I think that this has been a really interesting conversation and I’ve been fascinated. There’s a lot of different ways that IoT can be instantiated and it’s one of the things I’m most passionate about, both on the consumer side and the industrial side and the business applications, and as we started the conversation, the barrier to entry seems to be just continually decreasing. So thank you very much for the time today, Agustin and Chip, and I will… For the audience, we’ll be sending the recording to all of you and it’ll be posted on our site, at Ioterra. Before we go, Agustin, can you tell… Just give a quick plug and then let everyone know how to find you guys, and if there’s any cool opportunities to work with you in the near term.
Agustin Pelaez: Sure. So building on what you just said, entry barriers are just getting lower and lower, just as a quick plug for the last question as well. The best thing you can do is get out there, talk to people, see what their needs are, and start prototyping. Don’t wait until you got a problem to solve to actually solve it, just start doing some things on your own. The technical tools are already out there, Ubidots is just one of them. So we think of Ubidots a little bit as a superpower to IoT entrepreneurs. We wanted to build things that they didn’t think they could build before and once they log in and signed out and create their first project, they will be like, “Wow, I can actually create a portal for my customers and start selling from day one.” So my invitation will be to just go to I’m on hackster.io. So if… Also at GitHub, I’ve got a bunch… All my software is up on GitHub under chipmc. So feel free to reach out to me in any of those models as well. Or Twitter.
Danny DeLaveaga: Of course. [chuckle]
Chip McClelland: Yeah!
Danny DeLaveaga: Alright. Well, hey, thank you very much. I’m gonna go ahead and closeout and we’ll see you next time.
Agustin Pelaez: Thank you, guys. Bye-bye.
Chip McClelland: Thank you, everybody. Thanks for your time.
