Pushing AI to the Edge: Industrial Interoperability

Hey everyone, we're back here at SUSECON Tuesday-- Wednesday, excuse me, day two. Tuesday was day one, but Monday was day zero, so it might say it's day three, but whatever. We're here.

We're talking to folks. There's a lot of buzz going on this year at SUSECON. And if you haven't been following our videos, I encourage you to check out a lot of our articles as well.

com. Let me introduce you to our next two guests here. To my far right, all right, we have Basil, but it's actually Keith Basil, but- ...

Basil. Hey, man. Yes.

Welcome, Basil. Happy to be here. Thank you.

Tell people a little bit about yourself, what you do, how you got here. So I joined SUSE after the Rancher acquisition. Okay.

I transitioned into the general manager role for SUSE's edge business. Very cool. So you were part of Rancher?

I was. My latest role at Rancher was VP of product for Rancher. So, Liang, I'm drawing a blank.

No worries. Who's the CEO founder of Rancher? Oh, several founders.

Sheng- No, but the CEO, Sheng. Sheng. Sheng, yes.

So I know Sheng a very, very long time. Yes. Yes.

Like three companies before Rancher. Yes. com, and then Rancher.

Oh. I've known Sheng and Shannon- Yeah ... since roughly about 2009, 2010.

Love them. Yes. Great guys.

What a smart guy, too. Yeah, very smart guys. And his new thing, I've talked to him.

But anyway, hey, that's a great start. Yeah. So talk to me.

We're going to get into the SUSE edge business- Sure. Sure ... here in a second.

But first, I want to introduce our next guest, and it's Bart, and if I mispronounce your last name, correct me, but it's Bart Newberg? Newberg is kind of fair. Newberg?

Yeah. Okay. Definitely.

Bart, you heard what Basil said. What's your story? Yeah.

My story. Well, currently the chair of the Margo projects. I am a Rocco employee, a few decades going through different layers of serving the industrial automation space.

Whether it is in open architecture side or not. I'm based in France, here on the continent in Europe, yeah. Beautiful.

So let's dive in here. First of all, I'm not familiar with the Margo project, so let's start maybe there. What is it?

What is it? Well, it's an open source community that is trying to remove hurdles for innovation. And it can be a big word, but the industrial operations or industrial users are currently going through a transition on a journey to autonomy.

More software is going to have to be deployed at their facilities, we call it at the edge, where the digital and the physical world kind of meets. It's going to give them a new challenge, because once the software is deployed, you're going to have to maintain it there. You're going to have to keep it operational.

That's what we call that kind of fleet management task is going to be added to their already existing tasks. The open source community is proposing a way to do that in an interoperable way, as they are recognizing that in industrial automation, there is actually a large ecosystem of multiple suppliers, meaning it can't be solved by one supplier that is trying to answer all the use cases of a given industrial user. So that's a bit what Margo is about, that open source community that is pulling those solutions together, yeah.

Love it. Love it. We're going to come back to this in a little bit, but Basil, I want to run back to you, man.

Okay. There was a time where if someone spoke to me about the SUSE AI-- uh, excuse me, the SUSE Edge- Yes ... " Correct.

Correct. But it's more than K3 now. Yes.

Agreed. So you spoke about Rancher, so Rancher created K3s. Yes.

And one of my early remits as a Rancher employee was to strategize on how K3s or K3s could be productized for the edge. And so that was really the seed of all of these things that we're working on in terms of edge. So fast-forward to SUSE, we've built an internal strategy and taxonomy to define the edge so that we could align our products and offer solutions for those different segments in the edge.

So, the first was the telco space, so we have Telco Cloud that solves for that. It's very similar to the model we have with Margo. With Margo being an upstream community, there's an equivalent in the telco space called Silva.

So we work upstream with Silva, and we pull downstream, and we have a Telco Cloud that fits that industry. And then we have SUSE Edge for the far edge, which is basically Kubernetes, mainly K3s, back to bring it back, as the runtime and infrastructure that we push to the edge so customers can run applications on top of that. And then lastly, the third wave is what everybody's talking about here today, is the SUSE Industrial Edge, which we announced today.

Well, you can't stop there, man. Oh, okay. Sorry, sorry, sorry.

" No, I wanted to give you a physical activity here. Oh, come on. Brilliant.

So let's take it further. So we built this taxonomy, near edge, telco, far edge, runtime. The tiny edge was all about resolving and providing solutions for the industrial edge.

We didn't have a solution. So what we did is we assessed what would win in that space. So things like device connectivity, multiple protocol support, supporting the concept of interoperability, a platform to collect that data, to have historical baselines for the data coming off of the devices.

And then what's really cool is that we have what's called a workflow engine that lets you drag and drop and build workflows to orchestrate and pull data from those devices, and that feeds the operational awareness dashboard. So you know what's happening on premises in these industrial situations. So it's a phenomenal win.

It's very elegant for us because we've taken that platform and we've added it to our SUSE runtime. So we have the complete stack from Linux all the way up to operational awareness in the industrial space. I love it.

Very coolBart, you heard how he referred to working upstream, downstream. Yeah. Margaux, the other telco.

Silva. Silva. Yeah.

Yes. Talk to me about how you view that relationship from the Margaux point of view. Yeah.

Right? Because what's upstream for him is downstream for you and vice versa- Yes ... a little bit.

That's it. It's a good point, and that's really what Margaux is all about. " You cannot go and say, "Oh, yeah," that kind of challenge, as I mentioned, managing and a heterogeneous fleet.

Try to do that as one, even two, even three vendors there. It just isn't going to work. Well, it can work, but they're just not going to be business reality that can be solving and being financially addressable.

So the only routes to go there is to kind of do it together. And that means, yes, upstream is downstream for the other one. That's the whole point, but that it's really that win-win if you get out of it.

Because in the end, whether you want the users, they want to get that ability to have a single pane of glass, if you want to call it like that. Mm-hmm. Regardless of who are the partners or the suppliers that are providing their pieces to the puzzle to get to that software-defined future.

They don't really care. They say, "It just has to run. It has to run now, it has to run tomorrow.

It has to be updatable," et cetera. And I don't want to have what I would call 10 screwdrivers to get my wardrobe to fit together because the left panel is with that type of screw and the right panel with another type of screw. You just want to have that one screwdriver making.

We've all been there. Still can mean that the left-hand side panel come from different suppliers, doesn't really matter. And for the suppliers themselves, it also doesn't really matter that they might have competitors using similar, I would say, enabling pieces in their technology stack.

That is not going to hinder them from innovating. In fact, contrary, it's going to help them actually to innovate because they can focus on where they differentiate, and they don't have to put the energy in pieces- Mm ... that actually isn't valued by their customers as something that is differentiating, where they're willing to call it pay.

That's a big word, but that's the reality of doing business, where they're ready to pay for. So the things that aren't... I might say, things that are just enabling, let's just do that together.

And that's what you talked about the Project Silva. Yes. We have that kind of connections in the background, like learning from Project Silva as Project Margaux.

Why? Well, they just started that a few years ago. Because of the industry, got the awareness earlier.

We just don't want to reinvent the wheel with Project Margaux. That's an important- One thing I alluded to in the keynote was that the industrial space, if you look at it from a technology perspective, is very siloed. You've got proprietary systems.

A Siemens system may not talk to a Rockwell system, may not talk to a Schneider system. Yeah. So the concept of interoperability is great, and the best way to say this is that nobody can do all of it as a single entity.

And this is why the relationship with Margaux is so critical because it's like nobody's smarter than all of us when it comes to solving the broader interoperability challenges. So we're super excited and big champions of the spirit of what Margaux is trying to achieve. I like that.

Nobody's smarter than all of us. Yes. Hey, a big story around the SUSE Edge story is the recent acquisition- Yes ...

you guys did. If I mispronounce it, it's Losant? Losant.

Losant, excuse me. Talk to us about that one. Well, it goes back to my original statement of the three-wave strategy we had to address all of the edge segments.

The final one was, again, that tiny edge, the industrial space, and all we had at SUSE before the Losant acquisition was we can push Kubernetes out to where it's needed most. But we relied on customers and partners and systems integrators to build the industrial solution on top of the runtime. What we wanted to do was bring the complete stack to the table, and so we evaluated several potential companies, and we saw that the Losant acquisition would accelerate us into the space.

To give you some idea of the filtering we did, we wanted to have a solution that was cloud native friendly or very adjacent or something we can easily run on top of Kubernetes. So we checked that box. We wanted to have full white label branding because SUSE's partner network is very large and very strong, and we wanted to basically have a turnkey solution to just give to our partners and integrators to go and do industrial services and solutions where they have more of the deeper domain expertise, and we just become the folks that facilitate that structure.

And so we checked the box across the board from a business perspective and from a technical perspective, and we transacted the acquisition of Losant. We closed on that, I think February. Mm-hmm.

Late February. And from that time period to now, April 22nd is when we announced the SUSE Industrial Edge product, which is the Losant capabilities on top of the SUSE runtime that we already had in place. Excellent.

Let me bring something else up. Yeah. " Yes.

"We got to bring AI closer," cheaper, right? Than- Yes ... going back on these frontier models and paying those token costs.

They're killing us. Correct. How does that all play?

We saw the AI factory here, but that's not necessarily an edge product, is it? The output of that, the derivatives of that, could be applied to the edge. So let me give you a high-level view of how I see AI affecting the edge, particularly industrial.

So the platform that we acquired, the SUSE Industrial Edge new offering, it has several internal pillars. There's infrastructureThere's device connectivity, there's the workflow engine, and then there's the dashboarding and kind of observability. We actually can look at each one of those pillars and apply augmented AI to each one of those things to give us an aggregate of something really cool and really innovative going forward.

One of the things that we're talking to with one of the industrial manufacturers is having the data around the device families being fed to build a SLM instead of a LLM. Right, small language model. A small language model that we can have on premises to augment some of the MCP capabilities for all the other pillars.

" And that's the level where we want to get to in the future. But I've always said that we have to have the baseline connectivity, all the fundamental building blocks in place before we can layer AI on top of it. Agreed.

We're getting there, though. We are. Great thoughts.

I see it. We use Perplexity Computer. I use Perplexity Computer.

Yes. Just last week, maybe it was the week before, because traveling screwed me up, but they came out with Perplexity Computer on-prem, where you can run it on a Mac Studio or something. Okay.

I'm sure you'll be able to run it on a SUSE Edge device too. Sure. And that, again, what do we want?

We want it closer so it's faster. We want it closer so maybe we could lock it down, it's more secure. We want it closer, so we don't pay the vig, right?

Yeah, on tokens. Yes. On tokens.

It's getting crazy. How do you guys look at that in Project Margo, though? Well, you bring it up.

If you want to run it closer to the physical world, for industrial users, it's actually suddenly going to become their task to maintain that workload, as we call it. Call it an app. It's kind of abstracted away, whether it is a machine learning or an AI app, doesn't really matter.

It's suddenly going to become their business to maintain it. Yeah. Meaning, go and do it, Mr.

Industrial or Mrs. Industrial user. Are you actually geared to go and do that?

Is that part of your typical mission? No, you want to take advantage of it. " Meaning it has to become way more simpler to have a single pane of glass to be able to do the life cycle management of it.

Otherwise, it's going to have to stay in the clouds, and we're not going to be able to get it to the point where it's more productive for us, maybe more cost effective, et cetera. So that kind of using AI at the edge, it's one of the use cases- Yeah ... that are, call it justifying the engagement that a lot of the players have and a lot of the actors have in the Margo community.

Yeah. Absolutely. Yeah.

Look, I first became involved with Edge four years ago with... Mm-hmm. It was really growing out of Wasm.

Yes. If you're familiar with Wasm and all that. Yeah, sure.

It's one of the technologies we're looking at. Yeah. But this AI on the Edge thing has really...

To me, that's the killer app for the Edge, guys. Yeah. Right?

I think- Yes ... it's driving what Apple's doing. Correct.

It's driving what a lot of us are doing. It's... No, no.

Not that it's not a killer app in everything we're doing. Sure. Here's a statement I used to say, and I say, applications go where they're best treated.

Yeah. And that's many times at the edge. In the AI sense, I've modified that a bit, and there's this concept of called, what I'm using the term, enterprise intent.

So what is the intent of the organization for that location? So I'm going to refactor my statement and say enterprise intent, particularly around industrial- Mm-hmm ... is what we're pushing to the edge.

And AI is going to manage that intent and take agency of implementing the intent for the organization. I love it. Guys, we're running low on time.

Okay. People who want to find out more about Project Margo, where do they go? org.

That's easy. And they- Starting point for what I would call the more HTML lovers, they can get routed to the GitHub. They can just go and experiment with the open source.

That's the place actually where that can be- You know the Git, is it GitHub/Margoproject or? It's a GitHub... It's Margo.

Just a Margo- Margo. Yeah ... very simple Margo.

org. org. Yeah.

Yeah. Dazzle, where can we stay on top of that? com/edge.

Slash the edge. Yeah. Not the edge.

com space edge. com. Yeah.

Easy enough. Yes. Gentlemen, thanks for coming here on Techstrong TV.

It was a pleasure. Absolutely. Happy to do this.

Happy to have you here. Yes. Hey, man, we got a lot more coming at you today.

We got more tomorrow. We had a bunch yesterday. Stay tuned.

You're watching Techstrong TV.