NATAN: Erich, how's it going?


ERICH: I'm doing well. How are you, man?


NATAN: I'm good. It's great to have you on the show.


ERICH: I'm very happy to be here.


NATAN: Awesome. You and I go back quite a few years as we tried to figure out this crazy decade. I don't know if it's, like, closing the decade of industry 4.0 or whatever it is, but spend a lot of time talking about how standards kind of play into that. So we're going to talk a little bit about that. 


ERICH: Of course. 


NATAN: But, you know, last time I saw you was in Hannover.


ERICH: Correct. 


NATAN: How about Hannover this year? What did you think?


ERICH: Yeah, I think it's slowly getting back to its former glory, just from a number of conversations and people I met. I would say it's about 80% where it was before the pandemic; you know, Microsoft had a big booth there. We got a ton of good leads, so that was very positive. The conversations that I had were all set up beforehand. So I was kind of booked solid for the entire four days. 


The fifth day was kind of slow because there was a train strike, so most people left in a hurry on Thursday night, you know, but Monday to Thursday was packed. And, you know, I've had a few panel discussions. They were also quite interesting, all around standards, all around where we are as an industry. 


So I think we're really beginning to see a lot of adoption for these new standards that were built over the last several years, you know, the days where, you know, you could just do POCs and then leave it they're over. People really want to use this stuff in production now. And, I mean, obviously, OPC UA, you know, it's been around for a long time. So it goes from strength to strength. What's new on the OPC UA on the end of the market is really all the cloud initiatives. I mean, I kind of started those a few years ago, but now it's like, you know, AWS is there; Google is there. 


NATAN: Right.


ERICH: So it's really much broader than it was before. And we have a lot of the automation companies on board as well, actually leveraging the cloud via OPC UA. So I think that's also something that we've seen a lot of this year at HMI that was new.


NATAN: Yeah, you know, it's so interesting and such a good lead into our discussion on standards and the like. I was in Hannover. First of all, it was really nice to see that they now bundled more or less this infrastructure and software side like in, you know, hall 15, 16, 17. Because it used to be, like, you know, you walk to this company for that industry. There's, like, a software company focus there. But the reality is, I think, it's a reflection of what's happening in the market that everything is converging, you know, things like, hey, Snowflake was at Hannover for the first time. How about that?


ERICH: Yeah, and there's a lot more. All the software companies have now discovered Hannover, right? It's not just the automation companies or machine builders anymore. It's really, you know like you said, the halls 15, 16, 17 was software companies only. And, you know, I think that's a big change as well. And it's a good change, digital transformation, right? As for digitally transforming these factories and these plants, you know, it's the software that's the new ingredient, right? 


NATAN: Totally.


ERICH: And similar for, you know, I'm also doing a lot of work with the Metaverse. There, of course, you're not just adding the IT but also some of the game technologies into, you know, traditional manufacturing, which is, again, it's a mind shift. It's a convergence of technology. And I think we're going to see a lot more in that kind of direction. And [inaudible 03:44] start talking about, you know, generative AI and all the benefits that it can bring to the table so that all these new technologies are converging into what traditionally was just an OT play. And now we have an OT/IT game engine AI play. And these technologies are clashing and creating this new reality where, you know, you can build things in a completely different way, right? 


NATAN: Yeah. And, you know, I want to kind of build on that thought. And I do want to jump into standards, which, you know, I think we got to explain to the listeners what you do. 


ERICH: Okay.


NATAN: But let's wrap the Hannover thing for a second. It's like, I was walking there, and I'm like, oh my God, we have arrived. I think this year I feel comfortable the first time saying it that the message of open architectures that have blueprints...and you've seen it at the Microsoft booth, obviously, like you said, AWS, Google, Cognite, all these, like, companies. And they say, like, you know, you should control your architecture, and you should do best of breed, and you should bring the ecosystem into it, and so on. So that was, like, one thing. 


The other thing I've seen...and this is, like, you know, you talked about the refreshing change in Hannover, and this is a very sort of a, I'd say, German-centric quite traditional venue that has been evolving. And then you see the signs that the Hannover Messe folks put on. They had, like, signs, like, with a, you know, a picture of a kitten saying, "Today what the fuck is tomorrow's wow." And I was like, oh my God, they're finally getting it. And then they're saying, like, "Have a beer with an engineer." 


So the comment you made that they're finally, like, ready not to talk about all these big words like digital transformation and whatnot. They're like, hey, let's go build because it's here. And the other thing is, like, no one is asking about what do we want to think about cloud? Everybody is like, that's obvious. 


The last thing is, like, walk the halls and, like, the frequency of IoT per capita [laughs] has dropped dramatically. No one is saying IoT anymore. And maybe that's the segue into our standard discussion. And so maybe first of all, you are Chief Architect, Standards, Consortia & Industrial IoT, Azure Edge & Platform at Microsoft. 


ERICH: [laughs]


NATAN: I think that's one of the longest titles.


ERICH: Yeah. I always joke I need to write my title on the back of the business card as well because there's not enough space in the front, you know.


NATAN: Please explain, in great detail, every section of your title before we begin.


ERICH: So, first of all, I started off I built the Windows Industrial IoT team from scratch. A few years later, then I built the Azure Industrial IoT team from scratch. And I've been running both teams for a long time. But my hobby was always standards work; consortia work, you know, bringing standardized interface, standardized data models, all these things into Microsoft products, you know, and differentiating through open source. 


So a standards organization and I represent Microsoft, and several of them, they get their kind of market adoption through reference implementations that are open source where people cannot just read the spec, but they can actually try it out. They can try it out. They go to GitHub. They download, you know, the reference implementation. They start playing with it. They integrate it into [inaudible 07:02]. So that's kind of how it works out there. It's not just about the spec. It's also about the open-source reference implementation. 


So, you know, I started doing more and more of that. And I enjoyed it so much that when my CDP at the time asked me if I want to do that full-time, I jumped on it. I basically made my hobby my job. And I've been doing that now for three years, never looked back. I think it's great. So the engineering team is run by somebody else. And I can focus 100% of my time on basically advising the teams within Microsoft on how to build standard-compliant software. 


And, obviously, I continue to build those standards in those consortia. I mean, they want our help. They're more than happy that we're contributing so much open source. You know, we're the number one contributor of open source to the OPC Foundation. We're the number one contributor of open source to the Digital Twin Consortium. And the list goes on. So there's all this activity that we're driving towards standardization, both inside and outside of Microsoft, that I think is critical to move industry for all from the, you know, workshop POC area to production.


NATAN: Yeah. So, before we go even deeper, I want to kind of talk about the elephant in the room. Let me explain what elephant it is. So, back in the day, I was part of standard consortia myself, you know, the Java Community Process, for example, The Cronos Group, you know, that built all the standards for the mobile industry, things like OpenGL ES. And I was representing Samsung, so kind of had a similar job to you, different industry, I guess. 


And, you know, we all know, like, people get pretty cynical and sarcastic about standards pretty quickly. And because they're, like, supposed to be these places where we collaborate and define the industry, and all that kind of stuff and, like, you know, draws experts but also draws the politicians, like, just to be honest about it. And then it's, like, becomes the word soup salad, whatever combo, and lots of meetings, and lots of fights. So, what are the big things that people...You know, before we dive into the actual interesting standard work, what are you seeing? Like, you're there.


ERICH: Yeah, I think there's generally two types of companies participating. There's what I call the lurkers. So they just show up, and they don't say a word. And they just basically, you know, want to know what's going on. But they're more concerned that they may miss out on something, right? And then there's the companies that are actually actively contributing. You have much more highly ranked people from those organizations actually involved. And you can see immediately how serious a company really is about a particular standard by the type of people they send to their working groups.


NATAN: But, Erich, tell me, what are people fighting about now? That's what we want to know. We know the standards are great and doing all these things. But where's the blood? 


ERICH: So the fights are usually around the details, right? I mean, especially if they've already started implementing a standard and you're still developing it, obviously, that can cause churn. Quite frankly, when you are an early adopter, and you want to, obviously, take advantage of that because you're first to market with a product that is standards-compliant. You must also take into account that things may change, and you're one person around the table with one voice and one vote, right?


NATAN: Right.


ERICH: So you need to be 100% agreeable to the fact that things may change. And you have to go back to your engineering team and say, "Actually, we need to change it again to keep it standard-compliant," and that's a challenge. And, I mean, usually, the fights are around the details. But, quite frankly, most of the time, we all want the same thing—interoperability. Because interoperability gives you scale, and scale leads to revenue, right? 


NATAN: Right. 


ERICH: Everybody can build a Snowflake. Building a Snowflake is easy, right? 


NATAN: Mm-hmm.


ERICH: But it's not going to work with anything else. And especially in the manufacturing industry, where you have a hugely fragmented market, interoperability is king. So you must be interoperable with your competitors, with your partners, and with your customer systems that already exist. And if you can't get that done, you're not going to be successful, right?


NATAN: Yeah, no. I don't know if you would agree to that. But let me try something characterize. So I think we're past the first stage where the key infrastructure cloud players are entering the stage. It's awesome. Where were you guys? Okay, we were waiting. 


ERICH: [laughs] Yeah.


NATAN: That's great. But, you know, we have the traditional sort of automation incumbents like the Siemens and Rockwells of the world, and those guys, you know, great companies building awesome products to automate everything. But each one of them own, like, a fifth dom [SP] of legacy protocols. And that has been sort of the way of the world. And then they have, like, the galaxy of Siemens versus the galaxy of Schneider, and so on, right? 


ERICH: Yeah. 


NATAN: And the reality that it changes the supply chain and the distribution methodology, and the integration folks, and all that kind of stuff. And, of course, people span and do all these kinds of stuff. But we kind of lack true openness, like, in the full sense of the word. And I guess what I'm trying to ask you is, like, do you believe, like, from where you're sitting that the real shift to open protocols...like, let me explain, you know if we did not have TCP/IP, it would have been, and ethernet, it would have been pretty difficult to imagine HTTP as we know it and all its evolution, right?


ERICH: Yeah.


NATAN: Are we at that moment, like, from standard and their revolution, like, from your perspective? Or is it still, like, too early?


ERICH: It depends. I mean, certainly, I think the era of the walled garden ecosystems is over.


NATAN: And it's just like, ba-bam. It's over. 


ERICH: It's over, yeah. I mean, you know, look at what happened at the field level, right? I mean, you got to differentiate what level, you know, in the industrial automation [inaudible 13:23] we're talking about. So I would really think that at the field level with OPC UA, you know, field level effects, you know, field level communication, the fieldbus war was ended, right?


NATAN: Yeah. 


ERICH: And I think, similarly, on the OT network, OPC UA became the standard even earlier. I mean, nobody questions that anymore. And even all the automation companies have understood that, and not because they woke up one morning and said, "Oh, let's change strategy." No, it was because the end users, the manufacturers, demanded it. 


I know so many large manufacturers who have OPC UA in their requirements specification. They say, "I won't buy your software unless you give me a standardized interface based on OPC UA." And they even go further than that and say, "Well, that's the interface." But I'm also only interested in standardized data models because, quite frankly, you know, people ran around in the '90s and said, "Oh, data is the new oil," right? They were wrong. It's not the data that's the new oil. It's the data model.


NATAN: But can I provocate you maybe, like, I know you know the answer and, like, you're going to destroy me, but I'll try anyway. Why do we actually need OPC UA if we can, like, just have MQTT native and hardware accelerated? 


ERICH: That's a classic question. 


NATAN: You know, our listeners don't see you, but Erich is now smiling widely.


ERICH: Smiling, yes.


NATAN: Natan is like brace, brace, brace.


ERICH: The answer is yes, we need both. So people always think that you know, and there's a few companies, especially in the U.S., that want to only talk about OPC UA in the OT network OPC UA client server. But for the last five years, OPC UA has been extended to also work with a publish-subscribe, you know, communication pattern, you know, that we call OPC UA pub/sub for short, and it leverages MQTT. It's based on MQTT since OPC UA is generally transport protocol independent. You can use different transport protocols, but, like, the factory standard is MQTT. 


So, from that perspective, of course, you can use OPC UA with MQTT. And that's, in fact, what all the major automation companies now use. And Amazon and Microsoft, so the two big cloud companies, also standardize on OPC UA with MQTT.


NATAN: I see. And how does that read on things, like, say, MTConnect? And it's a more general question. It's like, is OPC UA and MTConnect together, and separately it's pretty much our core protocols for the next couple of decades at least?


ERICH: Yeah, and there's a...you probably know this, right? But maybe the listeners don't. I mean, OPC UA keeps getting extended with so-called companion specifications, companions. So, within the companion specs, you can do two things; you can either define a new data model, for example, for a particular machine type, which then gets standardized through the OPC Foundation. Or the other big use of companion specifications is to actually create a mapping from a different standard to OPC UA, and that's what was done with MTConnect. I think they're up to version...I forget what version they're at now. In other words, it was just a revision recently for MTConnect to OPC UA mapping. 


So, if you have a machine that is MTConnect-compatible, you can buy software off the shelf that will convert the MTConnect data model into the OPC UA data model. And then, of course, you can use OPC UA pub/sub with MQTT to effectively send MTConnect data to your analytic software for analysis. So it's compatible.


NATAN: Or to your no-code/low-code for operation software, say.


ERICH: Right, exactly. Exactly. 


NATAN: You know.


ERICH: And, I mean, the MTConnect Institute, the OPC Foundation has a very good relationship. We're working together. And I think, really, there is no conflict here at all. In fact, they're collaborating, and, you know, again, it's all around interoperability. And I have several customers actually that use exactly that combination of MTConnect and OPC UA and OPC UA pub/sub to the cloud, right? 


NATAN: Yep. 


ERICH: So, no problem there.


NATAN: Where do you see this going? Before we play our favorite, like, end-of-show game, which is the buzzword bingo.


ERICH: [laughs]


NATAN: Let's start with, like, the hottest buzzword right now that, you know, burning through every Google search every day, which is OpenAI and ChatGPT. Are we ready for that?


ERICH: I think we are. And it really depends what you use it for. So we're actually working with Siemens on using OpenAI to automatically generate the data model for fixed function industrial assets that probably don't have OPC UA built in because they're so small, think of energy meters, pumps, valves, sensors, that kind of stuff. And wouldn't it be great if we could automatically onboard those kinds of devices and map them to OPC UA automatically? 


NATAN: Yeah.


ERICH: And right now, what people need to do is they need to find the spec. They need to read the spec. And then, they need to manually map. I don't know; those devices usually have, like, a handful of tags. Let's say they have 50 tags, right? 


NATAN: Yeah.


ERICH: You need to manually map these 50 tags, get their name right, get their address right, get their length right, get their unit right. [crosstalk 19:04] And manually type that into your favorite industrial connectivity software. Wouldn't it be great if that whole process could be automated? And that's exactly what we've done together with Siemens using ChatGPT version 4 and our Azure OpenAI service, and we just demoed it at HMI. We got great feedback from the audience. We got lots of follow-up, you know, people blogged about it. 


So I think we're onto a winner here. And I think it's a real advantage, even if the model isn't perfect, even if it's 90% perfect or correct, let's say. You just generate that into your editor into your, you know, connectivity software. You eyeball it, you fix it, and then you onboard to device.


NATAN: That's exactly the gist of demos I think you saw as well that we, Tulip, put on with Microsoft where...exactly with this sort of interaction paradigm that is fundamental human in the loop. Because, you know, I think the hype train that is going very fast now it's like, oh, well, ChatGPT will do it. And it's, like, so ridiculous because it's, first of all, you understand what this thing is good for. And, like, your example is so perfect. 


Like, the finite nature of, you know, small amount of tags highly structured for the meters or for, you know, those devices lends itself well to, like, generating something that a human can refine, finalize, and there the game is just okay. And it's just a good author that saves time, but you still need to check it for all the things to make it work. 


We're doing the same. And we put together apps that you can just use natural language. And because it has, like, in 3.5 and 4, like, a large enough human language models, you know, you don't need to define error codes. It will just say, "Oh, this is an end on event," or "This is a machine down," or "This is a timing issue." And it will just start classifying things that your industrial software should classify. So that's, like, a simple example. 


But the stuff, you know, in our low-code side of the house that will come out at some point is things, like, oh, we allow people to build custom widgets in Tulip. So, yeah, it's just, like, we'll have another field that says, "Build me this special gauge. I want it red and green with this kind of parameters." And it will just do its best to give you something that works for your description. 


You might need to iterate and deal with the prompt engineering, as they are starting to call it. But, at the end of the day, you'll get a piece of working JavaScript, and HTML, and CSS, and, like, you'll tweak it. And you hit publish, and you have a new widget in Tulip. And you didn't need to take it to a shop or hire an IT person to do it. And I think that's going to happen faster than folks...


ERICH: Yeah, I think, Natan, it comes down to seeing it for what it really is. It's a tool. It's a tool to make your mundane jobs easier. It's not here to replace you. It's here for you to focus on the things where you can add the most value. And I think that's exactly how it should be used, and then everybody wins.


NATAN: And that's a great note to jump into the last section of our show today, the buzzword bingo. Are you ready?


ERICH: I'm ready.


NATAN: The way this game is played, you don't need to think so much because, you know, it's kind of supposed to be, like, funny and off the cuff but somewhat serious too. I just say a buzzword, and then you give me one or two sentences what that means to you today or maybe what it will mean in the future; you pick. 


ERICH: Okay, cool. 


NATAN: So here we go with the first one: digital transformation.


ERICH: We're on a good path. 


NATAN: All right. Digital twin.


ERICH: A tool to define your asset more accurately. 


NATAN: Metaverse.


ERICH: A great way to invent new use cases that will help people get safer and more secure in their job at the factory or on the plant.


NATAN: Amazing. So I think you win the prize of not being negative about any of your buzzwords.


ERICH: [laughs] Yeah, no, I mean, I got to say, especially Metaverse people say, "Oh yeah, it's just a fancy dashboard." Well, tell that to the expert that needs to do flight training to be shipped out to an oil rig because some machine at the oil rig is no longer working. You know, and if a person on the oil rig instead can wear an augmented, you know, headset like HoloLens and the expert who can fix the machine sees what he sees, you know, on-premises and can talk him through fixing the machine without having to fly out there, I think definitely much better.


Because I don't know if you know, but flight training for going out to an oil rig involves being fully suited up, you know, buckled into your seat, and then submerged in a tank of water where you have to free yourself, and that's [crosstalk 23:58]


NATAN: Yeah, evacuate the helicopter. 


ERICH: Yeah, exactly. And, I mean, that is certainly something that I want [laughs] to avoid, you know, so...


NATAN: You know, every time stories like that, they just ring the, you know, give digital transformation meaning because we think about offices and back end whatever. And the world we live and operate in is so physical. 


Erich, on that note, so good to have you on Augmented.


ERICH: My pleasure.


NATAN: I appreciate the time and the insight. It was a lot of fun. We should do it again sometime. 


ERICH: Absolutely. It was a lot of fun. 


NATAN: And we'll see you soon.