Blaine Mathieu, CMO at VANTIQ, discusses digital transformation using Internet of Things and real-time enterprise applications. This interview took place on March 2018 with Adam Stein at the nGage IoT6 Exchange at Ponte Vedra Resort, Florida.
Download the audio-only version here.
Adam: Welcome Blaine!
Blaine: Thank you! Great to be here.
Adam: Yeah! Glad you can join us. Thank you, everyone, for coming back from lunch so promptly. Hopefully, we’ll keep you awake and entertained and educated – or as my kids would say “edjumacated” for this half an hour. We’ve got a lot to talk about.
Adam: We definitely want to talk about the event-driven, as you called it, the event-driven application, and I think of it as an event driven automation application. EVA squared.
Blaine: I like that
Adam: Yeah, you talked about that in our one-on-one. I think a lot of people want to understand what they can do in their environment with applications, with processes, without getting overly complex. Yesterday on stage, I talked a little bit about citizen developers, and that whole citizen developer phenomenon is – everybody can get involved inside the corporation, and I think VANTIQ definitely embodies that philosophy: that people want to use data and interact with the data. As you and I have talked before this conference, it’s not human-machine interface, it’s human-machine communication
Blaine: And collaboration
Adam: And collaboration. That’s right. You have that right. So, why don’t we start off on that? I’ll ask a couple questions, and please expound on that as you see fit. What are a few IoT use cases and examples from the VANTIQ point of view? Talk about some of those use cases and why they are relevant to the audience here, please.
Blaine: Yeah, a lot of amazing use cases that the folks in this audience are working on, and I’ve been talking to some of you over dinner and even at past nGage shows. So, I guess one way I would start is by saying: I almost push back on the notion of an “IoT use case” only because I don’t think what we are trying to build here are “IoT applications” any more than what you are trying to build are AI applications or block chain applications or big data applications. IoT is a technology, and it’s an important one. It’s a driver of a lot of what we are here to talk about today. But fundamentally, you are trying to build – I imagine in many cases – truly transformative, enterprise applications that are helping your business do something in a way that’s not getting done today, right? And IoT is one of the technologies, a very important one, which is a driver of those applications or a contributor to it.
When I think of a standard sort of canonical IoT application – some of you heard some of this before during the breakout sessions. But what this use case really involved was, I will say, one of the two largest telecommunications infrastructure providers in the world selling their equipment across the United States. And they need that equipment to be serviced and maintained in real-time literally because this equipment being down is unacceptable to their clients. And the company they hired to do that service built, with some help, in rapid time, an application that senses the state of those machines – sensing the state and reporting and sending off an alert that a machine went down, that’s sort of phase 1. That’s the easy, early, low hanging fruit of IoT. The next phase is: now you start to predict when that machine might be ready to go down. Maybe that’s phase two. But what these guys are doing is, and what I would call phase 3, is taking a real action based on whether it actually goes down or hopefully they’ve predicted that it’s going down.
To do this you have to kick off a very complex system, a very complex application, that looks at who could potentially provide service for this equipment, and by the way, these service providers aren’t employees of the company. These are outside contractors, not employees. And you have to pick one based on their location, their past rating, their experience and training level, the equipment they have in their machine based on RFID sensors, maybe their security clearance level, and while all those factors are being taken into account, you select one. Maybe the person decides they don’t want the job, so then you have to go to another one. Finally, a person accepts the job. You track and move them Uber-like toward where the situation of interest, the machine, is broken down. Of course, while that’s happening, maybe the machine went from a problem state to a broken state, and the technician you picked is able to prevent, to do preventative maintenance, but not to actually fix the broken machine. (laughter) Now you have to stop that person, pick a different person, all of this is being orchestrated and happening in real time while the data from these machines are continuing to flow in, including during the fix process itself.
Adam: And collaborating with the people that are involved in using the data, fixing the machine, etc.
Adam: And I think that really points at your discussion that human-machine not interface but collaboration.
Blaine: Right, right.
Adam: And let’s expand that, unpack that a little bit. Talk about why that’s important in an Internet of Things environment.
Blaine: Yeah, well HMI has been big for decades now, and it’s about how do you effectively enable people and machines or interfaces to work together? What we’re seeing now is probably two things: These machines, so called “machines” are mostly software systems not so much robots, although they could be.
Adam: They could be.
Blaine: Software machines are becoming increasingly intelligent themselves. Enabled via AI, machine learning algorithms, etc. And so what you’ve got are two sets of expertise: the expertise of somebody with wisdom and experience, and then you’ve got the machine level of expertise. And we’ve literally seen them in the use case we’ve talked about collaborating while the real-time state of the machine is continually being reported to figure out what’s wrong and to fix it and – oh you tried something? That caused the machine state to change to something else. Now I know from my experience that that something else means you should now follow this step. The machine and the person working together.
In this particular use case, they’re not using augmented reality yet. We are using it in the lab, starting to play with heads-up displays and indoor location tracking. So, partly it’s about the intelligence of the system. And partly it’s about the devices, NLP. Some of you saw the demo with Brett telling the pump to shut down or take me to the pump. So basic NLP interaction with machines is already there obviously. The augmented reality stuff will come on very strong.
Adam: Yeah. I think that it is totally important that they have come a long way from the augmented reality, early days of people who used to talk about it. I know with SAP we talked about somebody going out to a stadium not knowing what a lighting box looks like and finding a broken lighting box and using augmented reality to fix that. That’s all well and good, but what if there are application that manage some of that lighting that can tell the repair person, “Oh, before you come, ______.” You know, give me context. You brought this up earlier, and this came up yesterday in one of our discussions too. Give me context to understand what it actually is happening so I know what I am going into, so I have the right materials and also the right data.
Blaine: That’s really important. That’s why again I resist the notion of calling something an IoT application because IoT devices and sensors are only one set of inputs you’re going to need to build an effective application that takes action in real-time. You need to add context to that. External data sources, inputs from legacy data systems, from systems like weather.com. That’s why they sold for so much, because that data is so valuable to many, or should be used by many enterprise applications in many use cases. And so that contextual data, along with the rules, the constraints, the other things you need to put into the system, it’s much more than just an IoT application.
I do want to touch on one other thing you brought up and I brought up the notion of augmented reality. Now I almost want to retract it. This makes it sound like I am talking about some future scenario of some glorious hologram appearing in front of us.
Adam: Yeah, well it doesn’t have to be.
Blaine: Everything except the augmented reality stuff is real now. This is all happening. And even augmented reality stuff is being tested in some places. But you can do this kind of collaborative IoT enabled, powerful enterprise application today without augmented reality. You don’t need it. But it will be cool when it gets there.
Adam: A good example of that is the other example we talked about before this session in healthcare and hospital settings. So let’s talk about that application a little bit to help the people understand context because of the people, context-based information and how that would flow into an application that helps the people work at the hospital, helps people. God forbid you’re in the hospital – so let’s talk about the IoT chain there
Blaine: Really interesting example: so I actually met this Canadian healthcare provider of a large, major hospital operator at the nGage event in Canada last fall. Thank you nGage for connecting us with them. And they are doing something and are continuing to develop a use case that’s really interesting and entirely different to the telecommunications use case I was talking about, entirely different and yet somewhat similar. So we all know most hospitals today, we have IoT sensors already put on patients: pulse-ox, and respiration rate and heart rate. And if there’s a problem, the machine that the person is connected to starts beeping. Maybe, a nurse gets a red light in his station. But what they’ve done is take that to an entirely new level by writing intelligent software that orchestrates, again, the interaction of people and systems and these sensors all together.
So, in this case, a patient goes code red or has a particular alarm situation, yes the light goes on in the nurse’s station and the machine starts beeping. What also happens is, the relevant doctors, depending on the condition, are located either in the hospital or maybe not in the hospital. The system intelligently triages which doctor to bring to that patient’s room depending on their location and their level of expertise. Maybe the real expert, the patient’s actual heart surgeon, is five miles away. So, forget about that person. Instead, maybe take somebody who is not actually that patient’s doctor who is in the next room and bring them over. [The system] automatically flattens out the beds so they can easily put the person onto the gurney . Brings the elevator to the right level of the hospital and locks the elevator open so nobody else can use it, then unlocks the elevator when the gurney goes into the elevator. So they’ve written software to orchestrate an entire complex collaboration between people and machines and devices and sensors to dramatically improve the potential outcome for the patient.
Adam: I think the keyword to bring up there is orchestration, right? Because we have this sea of data – or lake of data, I like sea of data better – a sea of data out there and some of it, as we talked about yesterday, that’s in there. It can be monetized, but some is life critical. So how do you orchestrate that information in such a way that will – all of it may not be there at one time – but when it is there, you have to build in that those chords, you have to build in – pardon the analogy, musical – put enough to build those instruments to make sure that are all included. And by including that data set into an application or process, which VANTIQ is doing, you’re able to do that. So what’s the thought process behind that?
Brett has done some great demos. I saw the demo. What’s the process in understanding the business? How do you – VANTIQ engage a customer, and again this doesn’t need to be a sales pitch at all and you know that. But basically, how do they engage the customer, in order to understand the business, to understand what those instruments are in the orchestration? Because these folks are all from different industries.
Blaine: So this notion of building these “real-time” so-called “reactive event-driven applications” that can deal with asynchronous data flows, it’s not a new thing that’s come about in the last few years. Gartners and IDCs and Stephanie’s group and others have been talking about this increasing over the last few years – I think because IoT has become a big driver of the market. But this is not a new thing. CICS systems on mainframes have been running real-time banking applications for decades.
I know we’ve got somebody here from American Airlines – Is American Airlines in the room? In the front. Orchestrating complex real-time applications for passenger flows and airline scheduling. So, this notion is not new, but I might ask, I’ll put you on the spot: [motions to American Airlines} how easy it is to build, deploy, maintain, and enhance these enterprise applications that American Airlines is using?
American Airlines: Super easy!
Blaine: Super easy! [laughter] You’re a liar, but we’ll talk later!
Adam: He took the easy way out!
Blaine: Yeah! We know these things are not easy. And if you have a large organization with lots of resources and lots of time, you can build these systems, right? The challenge today is, even if you’re a large organization with lots of resources and a super large IT department and big budgets, you no longer have the time.
I know UPS is in the room. UPS? I don’t often talk to UPS about their use cases and what they’re doing yet, but I know Amazon has already spent the last few years building drone control systems to enable the delivery of packages across the US and across the world. This is another example of a real-time event driven application: monitoring the state of all these drones and have they picked up or dropped off? Are they interfered with? Are they running out of battery? All of these kind of things. Now, either UPS has started building their answer to that two or three years ago like Amazon did, or they have to think of another way to do it. Because building
low-level, reactive Java applications and stringing together all these tools and message queues and Kafka and Cassandra and all this stuff to make it work [is hard]. And you guys know, a lot of these you are mostly technical folks in the audience. The business doesn’t have the time to do that anymore because by the time you even spec the idea out, the business need has changed, or the disruptive competitor has come up.
So to take the long way around to answer your question. Now that some tools have started to appear that do allow you to build and also deploy and maintain these kind of applications much more rapidly, we see the challenges shifting more to conceptualizing the problem. It’s actually more on the business side now, in effect. We end up dealing more on the business side then we do on the IT side, from a VANTIQ perspective.
Adam: Yeah, I think the business side is, in some ways, more important in that you understand all the business processes and all business data points that come in. Again, this is a technical audience, but it isn’t, as you said earlier, not just the technology but about the business use of the data, technologies, enabling engines for that.
Blaine: We spend more time with customers and prospects, helping them conceptualize what’s possible. What could these applications do? Because, to some degree, think of the healthcare example I gave, the field service example I gave, some of these examples. Many of our clients aren’t even considering these as possibilities, right? Either because it’s just outside the box, or because they think, “Well yeah we could do that if I had two years and two teams of 20 developers and unlimited budget, but I don’t. So, I’m not even gonna consider that I’m gonna do something – let’s just get a dashboard. Can I just get a dashboard showing the pressure and temperature of my IoT devices? That’s good enough!” And no that’s not good enough.
Adam: It actually brings up a question for the audience: How many people, show of hands, have heard of the term design thinking? … That’s about half, a little more than half of you. So, for the other half, and correct me if I am wrong, raise your hand [motions to Blaine] and you as well, obviously,
Blaine: I should raise my hand, sorry!
Adam: I saw the virtual hand! [laughter] Design thinking is basically the art of being able to take a business requirements team and put them in a room with the application team or development team over the course of like a week. And basically capture all the requirements and get a SWAT team in there and say, “Here’s a sketch or a blueprint of what an application or process would look like based on the feedback that you’ve given me.” [motions to Blaine] That’s what we did. That’s what we did at SAP. So, before I went on my own, we worked at SAP, and I worked at SAP. And we would bring in companies to build these applications and processes over the course of a week.
But we didn’t have something like VANTIQ. We didn’t partner with VANTIQ. We had large systems of integrators that would come in with us, and they’ve have a SWAT team, and they would use a native development kit, and they would have developers on the fly and say, “Okay here’s what it looks like.” And they would take in all the different business processes from the team that was there. Well, that’s well and good, but that not only takes a week, it’s really expensive, it doesn’t allow you to capture everything other than what people are in the room. Whereas, if you have something like a VANTIQ or another solution that allows you to capture all the different data points – data lake or data ocean that’s out there – you can’t capture that infrastructure. And I think that really has changed, so why don’t you comment on that a little?
Blaine: Two things come to mind. That’s a great, great thought you’re bringing up. You said prototyping. And I know I’ve heard the word “POC” being used as a swear word I think last night, and today a few times along with “science fair experiments” and these kind of things. And I think the challenge with prototyping and POCs and science fair experiments is that’s sort of what they are. It’s conceptualizing and it could help getting the ideas out there, but you haven’t actually built anything real. You’ve fleshed out a concept, but then you’ve given that concept to a team of developers who start over from scratch using the right technology and the right backend and the right scalable system and a secure system because your prototype certainly wasn’t secure, and all of that. And they sort of start off from scratch and hopefully the prototype and the real application have any resemblance together. They probably don’t, right?
Adam: Probably not.
Blaine: And so, again, the new generation of technologies that are coming on are not prototype generators or POC generators or for “visualizing” real applications. You’re creating real applications.
Adam: And you’re getting everybody, as I said yesterday, the whole citizen developer phenomenon is really important because you’re getting everybody involved. Even if you’re not a developer expert, you’re getting involved.
Blaine: Every POC we’ve ever built becomes part of the final application. These aren’t throwaway ideas. The other thing that come to mind was the corollary in the event-driven real-time application world to design thinking is called event storming, and it could be part of a design thinking session where basically you take a bunch of business – and it could be IoT folks, executives, you name it – get them in a room. And you could do it literally with sticky notes. You can do it in software tools as well, now. But if you do it with sticky notes, try to conceptualize, what are all the event inputs that are flowing into your organization no matter what the sources; could be IoT sources, enterprise systems, people on mobile devices, you name it. And then what could you potentially do with those flows? Event storming is being used by organizations like yourselves to help with the conceptualization process, and really, part of the design thinking. In fact, those that have a one-on-one event with us later, we’ve got a little tool that we’ve developed that you can pick up so don’t let us forget to give you one. But you can also look up event storming online. It’s a standard technology.
Adam: I’ve got a couple more questions but I also want this to be participatory for the audience. Are there questions that our conversation hasn’t brought up so far that you wanted to ask? Any questions? Alright we’ll keep on going. So I talked earlier about security, and for me, security in this networked world that we live in, especially with IoT now, is kind of like an Escher drawing, right? Everything kind of starts – where does it start and where does it end? And I was talking to one of the gentlemen that was up here earlier about biology and how human biology is very much security oriented: whether you get inoculations, genetics, modified human biology, and I think network behavior is somewhat like that. You stop a security risk in one avenue, and it morphs and migrates and changes. How do you address some of those issues from a business application point of view, in your point of view?
Blaine: Interesting, I haven’t thought of that analogy before with the human body. I think it is a powerful analogy because of the reason why the human body or the human system is resistant to bacteria and viruses and all the things that can potentially kill us.
Adam: Threats! Threat detectors.
Blaine: Right. It’s because it was designed from the start with that layer of security, of protection, of redundancy in mind. It was, and I think that’s the same for anybody’s’ enterprise applications we’re trying to build, and especially IoT enable, the applications which by their nature, are highly distributed. Edge computing we were talking about this morning: lots of different protocols, sending messages, controlling, not only receiving info from devices but controlling devices. You can imagine how concerned that hospital was about security when so much was being orchestrated.
Another client of ours, Israeli electric company, running the high power Israeli electric grid. That’s a very secure system, and so this is again why prototypes and the old kind of POC can be so problematic because by definition, probably security was the last thing you were thinking about when you were building this thing. You’re going for speed. You’re going for agility. You want to be able to change it. You weren’t thinking about authentication either between people into the system or between the devices talking to each other in the system. And so, it’s the trite answer but it’s true: security has to be the foundation that everything is based on right from day one. It can’t be something you put in at day 60 or month 6. It has to be –
Adam: A band-aid won’t do.
Blaine: Yeah, no. It won’t do anymore, especially in the world of IoT. And a lot of studies and surveys I have seen show that security, if not the number 1 concern, it’s right up there in terms of IT practitioners that are looking at IoT projects right now for the reasons you said.
Adam: Yeah, because there’s all these different vectors that people – especially with IoT being distributed like you were saying – that the threats can come in on that they hadn’t really thought about before and how do you address all of those? It’s almost like there’s an infinite number of vectors. You can never stay ahead of it per-se, but at least you can stay in pace with it and block the obvious issues.
Blaine: Well that is the beauty of the cloud or the edge meeting the cloud to some degree because the edge provides the level of redundancy, of failover. You lose connectivity, some system goes down, you’ve got another one which isn’t – they’re even physically disconnected from each other. We’re also working now with, I’ll say, one of the largest French railway systems, and they’ve spent a lot of money and time over the last decade putting IoT sensors on the tracks, on the engines, on the turn signals, on the turnstiles, in the stations themselves, all the environmental monitoring and the buildings, you name it. They’ve got sensors on everything. They’ve spent a lot of money and a lot of time. And these are systems, again, which you have to be inherently secure in and of themselves, but now the problem or the challenge that you’ve got is: they’ve got two dozen siloed systems, one is monitoring track state and performance, one that’s monitoring how many people are going into the train stations, one that’s monitoring the temperature of the engines on the locomotives. And they say, “Now, in a secure way, we have to orchestrate all of this together” with all the people.
The people that are repairing these machines and manning the counter at the front desk. These things need to be able to talk to each other and communicate in an intelligent way, so I know that if a track is about to be unusable because it’s starting to vibrate. The train that’s on that track needs to be moved off. I could put another train in place, except the nearest train is just in a cool down mode right now. So, I need to take the second to nearest train, bring that, and then meanwhile I need to know how to tell the attendant in station X that this train is going to be delayed so don’t rush to gate 5. That’s using IoT sensors in a true sense and orchestrating a very complex set of events, event interactions. That is very interesting I think and also needs to be very secure.
Adam: And also, the interface to that example, the orchestration. People always talk about user experience, and user experience on an application like that or an application you’re rolling out is awful, then people are going to say it is broken when it is not broken at all, but the user experience sucks. So, what do you do about making sure that you’ve got a good user experience? What’s the feedback with that: a company like yours uses with its customers?
Blaine: Well, now we’ve circled back to human-machine collaboration!
Adam: I couldn’t fool ya!
Blaine: You know, again, I would say security is one thing that we talked about that often gets lost in the prototyping phase. Human-machine collaboration, or the user experience to use less jargoning terms maybe, is the other thing that gets lost. Because you have to push something together really quick, so you think about how the application is going to flow and what its going to do and maybe you don’t think as much as you should have about who is going to use it, how are they going to use it, how are they going to interact with the system, what’s the UX, right? It doesn’t matter how you build your user experiences; whether you use tools that are built into the platform, or maybe you’ve got your favorite mobile application tool kit that you’ve been using for a long time. But I would say to your point: bring those in the design phases as early as possible, hopefully even in the POC phase so you can get real people trying out real applications. Again, the folks that sat in the demo rooms today, that wasn’t a demo application. That was a real, live, and running application, and that’s what you need to have running in POCs so that you can have users literally try these things out and say, “No that will never work because of x and y.”
Adam: I think also, one piece of advice. We got a couple minutes left. I think Blaine and I would like to share, and especially Blaine, is – start small. Don’t worry about a huge enterprise-wide application with huge ROI and huge TCO. I mean, I think you [motions to Blaine] would definitely embrace that.
Blaine: I would. It’s interesting that the phase “digital transformation”, I haven’t heard it used once in this room or in any of these sessions. I know we are demonstrating at another nGage session on digital transformation which we’re at. So, I love the concept. I think it’s absolutely valid. Most companies here, virtually all, do need to fundamentally transform to avoid being disrupted. All those things we know about. At the same time, the concept is scary and it’s too big. It’s too much. None of us, or I don’t think anybody in the room is Chief Digital Transformation Officer of their company.
Adam: Not today
Blaine: Not today, and those people do exist. You know, at larger companies. I’ve talked to some of the people in companies that do have one of those people. But, even if you do, you have to start somewhere. And so, start small, figure out something that you can actually do, but don’t make it a meaningless POC or science experiment. Do something that will add value right away.
Adam: A problem – solve a problem.
Blaine: Yeah. Solve a real problem. Right.
Adam: So, we do have a minute left. Any questions that we can answer for anybody? Spencer’s up and mobile with the microphone. Wow, we’ve stunned them. I think it’s after lunch, what do you think? [laughter] It might be after lunch. Oh, wait! From the head table.
Questioner: I am also interested in companies like yours who are very customer facing. I’m sure you’ve seen some big surprises and disappointments as you’ve worked with organizations for a whole lot of reasons.
Questioner: But I guess I am always interested in: where’s the momentum, right? When you think about healthcare, when you think about transportation, retail, where are you starting to see the momentum happening across the various customer systems that you take a look at and why do you think that is?
Blaine: I see two sort of fairly common use cases. One, whether it’s even a healthcare thing or the technical communications infrastructure example, they’re both sort of “field service” related use cases. There’re obviously scenarios where you have people and machines that need to work together to fix something or prevent something from going wrong. So that’s a very common use case in the IoT world. A second one that we are seeing commonly is called an “enterprise nervous system” use case. It’s more like the railway thing, which is IoT, but it’s more about connecting systems of systems and making all these individual, siloed applications that have been built over years and decades finally work together. So those are the two common sets of use cases I would say you see pretty commonly.
Adam: Awesome. Well I know we are out of time. Thank you, Blaine, very much. Thank you very much.