AI Meets Networking with Thomas Scheibe of Aviz Networks at Cisco Live US 2025 – Tech Talks

Welcome to Tech Field Day. Yeah. Hi everyone.

I'm Thomas Shira with Networks. I'm the Chief Product officer for the company. And I'm Tom Hollingsworth, event lead for networking here at Tech Field Day.

Thomas, it's great to see you again. Same here, Tom. Um, it's been a while since we've heard from of these networks, but you just happen to be here at Cisco Live and we had an opportunity to kind of opportunity touch base and see some of the cool stuff that you're working on.

What is it about the rise of AI and networking that's making it so important for, uh, companies like Avis to be able to kind of make that more simple to use? Yeah, Good question. Good question.

And quite frankly, it was, was an amazing show. Um, AI and networking. There's really two aspects, right?

And we have it in our tech line. It's networking for AI and AI for networking. What really is happening, a lot of those things that companies want to do is all talk about digital transformation is happening.

And really AI is accelerating, is on two different planes. That's why our tech line one is networks need to get rebuild upgraded for AI workloads. We know this.

It's happening, it's happening in cloud, it's happening in the enterprise. We have the perfect tool for this because these networks really, a lot of them will be built around either Sonic OS or around ulus from Nvidia. And we have a beautiful orchestration tool this to build, monitor and deploy these networks.

So we showed this a lot, a lot of positive feedback. We actually have customers having this deployed today. Uh, that's one piece.

The other piece, which is the, I think the office for a lot of people looking at AI as a productivity tool, right? Which is really around, I have a network, it could be a current network, but I want to use AI capabilities to accelerate. How do I manage, operate and troubleshoot?

And that's really around the power of AI to get to data that you have today, your existing data, how you look at your network, how to monitor it, how to get to this data in a better way and use it in a better way. I think it's very important that you called out something, and I think a lot of people aren't really understanding in this industry, is that there is accelerating people using AI for other things. Yeah.

And then adopting AI to be used with what you are trying to accomplish. And well, I mean that's the part of the reason why we here at Tech Field, they have two separate AI focused events because there are companies like these that are building technology to accelerate AI based infrastructure. But then there's the aspect of, now I can take what AI is building and use it to simplify operations and automation and things like that.

Which do you see is going to be more important to your customers in the long run? I I think it's both. It's, it's realistically it's both, right?

Because the way I think about, and that's why we're starting off, was we're really a networking software stack. What we see happening, right? This transformation is along these two lines.

Companies that run networks will rebuild their networks every so often. And AI is the driver that sets a lot of the architecture constraints going forward. How do you build these networks, right?

And so one of these is clearly, hey, we wanna do this like the cloud provider do, right? We wanna run an open source network operating system, uh, called Sonic. Microsoft does this, meta does it in their own version, but it's, it's very clearly done.

So people want it because say, Hey, if these cloud companies build these large AI cluster and operate them, we want to have the same efficiencies. So that's one angle. And to your point, once you do this, you can use this now, right?

And we're talking about LLM and training clusters here, right? You now can use this to apply this. A lot of different verticals, industries, one of these use cases is to apply it actually to network operations itself, right?

Because just because you build a network, somebody has to operate it, right? Today, you basically have, and I love a good friend of mine who just reminded me of the term click ops. You click on dashboards and new eyes and you're trying to figure out how to get fast, where you need to be.

Why not just use a language interface and ask the question you wanna ask and let the LLM, the AI piece take care of it to get you to the data you need to get to. And so that's the second piece. And that's why I think both of these will happen.

People will keep building these infrastructures to deploy more and more AI workloads and then they will use these AI workloads to actually get stuff done, right? One of these pieces is get them operating infrastructure much better. And I think that is something that can't be understated enough, is that so many years we've spent as networking engineers and architects and operations personnel learning the terminology, learning the interfaces that we need to use in order to make the thing do what we wanted to do.

And then I go and I turn on the TV and I watch Star Trek and Star Trek, people just ask the computer to do a thing like a, a person would. And the computer knows what I need to do to make that happen. Like when you tell the computer to raise the shields, it doesn't need to know the address of the socket to bring up the SHIELD dashboard.

And I think that that's important for the next generation of people when we get to them, just like with modern PCs or Macintosh computers, I don't need to teach people how to, uh, you know, code applications to work on that infrastructure. It just kind of happens. It's, it's somewhere in the middle.

But I like your analogy a lot and I, I, 'cause I wanna also demystify a little people look at it saying, oh yeah, it's just magically dust stuff. What you just said. What really happens under the hood, and I just picked this example, raise the shield, right?

What actually happens under the hood, the the AI language model takes is raise the shield and translates it down. Shield means on the backend, this, this, and this. Mm-hmm.

Now, I know to raise, I need to take an action. I need to get certain data from where I need to understand what is the status of this shield. Maybe it's already up, right?

And I don't raise it anymore. It's not, and how much, maybe the question comes away how much you wanna raise, but that piece can be built into, right? So under the hood, a lot of the stuff, what you do today to operate a network, the data is the same.

You're gonna use the same data. It's just how do you interact with this data is significantly different, I think, going forward, right? Because the time where you have to learn who you are, you have to learn the clicks, you have to learn the syntax of every vendor tool that will go away.

It's literally raise the shoot or raise the bar. We have half jokingly a race AI took my, uh, job to the next level. And so it's really raising this, how you interact with your infrastructure is different.

And so yes, you don't need to all the know, all the little, uh, caveats, all the little pieces, they should be taken care of on the backend, but you need to be still very precise what you want, what action you want, and then a translation happens for you. Yeah. Yeah.

I think that that cannot be understated enough is that the job itself isn't gonna go away. Yeah. It's the way that we interact with the job, right?

That makes it a lot easier to do. And I think that that's something that a lot of people have hesitation about in the industry is they're worried that AI is gonna take that job away. Um, I've edited enough AI written material over the years.

I don't necessarily know that that part of the job's gonna go away. But how do you see AI kind of accelerating the job of your average operations person? Is it, is it just gonna be their, how I have a lot more time kind of sitting around on their hands or you think that they're gonna be tackling other problems?

We all think we will have time to sit our hands. It's never that, right? I'm coming outta networking.

It was like bent was if you have a big number, it's enough and it's never enough. And I think it's the same here. You will always be busy, right?

Because there's always something else you either want to integrate, whether you're trying to figure out how to make a network work for certain use cases, application users, there's so many different things. You just don't get to this, right? Because you today are buried by a lot of learners reporting, auto reports, all of the stuff you can make much, much smoother.

Quite frankly, you're buried a lot. Whereas like trying to figure out when there's a problem, what is really going on? Give me a baseline.

A lot of this AI can take care of. It can be your helper, your co-pilot, whatever you wanna call it that pulls the data together that is relevant for your situation. So you can actually make a decision faster.

Not the co-pilot. You're gonna be still the pilot, but you will get the data faster. And so that will free you up to get other things done.

I, I have never had anybody that that's worried about having too much time. They're more worried about they can't actually work on the stuff they wanna work on because they're too busy with mundane tasks. Sounds to me like the real value of of doing this is that AI is going to give me context around things that should be done and help me understand why they need to be done.

And I don't have to spend all of my time doing a whole lot of unnecessary research. That's Certainly one big chunk way I will have. Well, it sounds like you guys have a lot of really cool stuff going on.

Um, where can people go to learn more about it? Come to our webpage, uh, go to perplexity, go to an open AI tool, ask how to do this. You probably will find us, come to Avis.

Uh, we have demos. You can run a copilot, uh, on our webpage. You can try it out.

You can get the software. Again, we are a software company. We don't need you to buy a lot of hardware.

We are deploying a software. But yeah, please contact us, uh, across the portfolio. Absolutely.

com/tech field day. com for more details on that. Thomas, thank you very much for taking some time to come and talk to us about all the cool stuff that you've got going on and make sure that you stay tuned for more great things coming out of Cisco Live and Tech Field Day.

We'll keep you updated. Thanks Tom.