Exploring the Power and Potential of Enhanced AIOps with Nokia
Clayton Wagar, leading the AI practice for Nokia’s IP division, framed his presentation as creating a “through line” connecting the history of network operations to its AI-driven future. Recalling his own start managing a 911 data center with manual, CLI-driven processes, he traced the evolution of the industry as operators moved from mastering workflows in their heads to writing them down and eventually using automation. Wagar emphasized that as AI is introduced, it’s crucial to be prescriptive about its use and to understand its two main facets: the “plumbing” aspect of building massive networks to support AI, and the application of AI tools to network operations itself.
To illustrate the challenge of applying AI to mission-critical systems, Wagar told a story about a 1943 discussion at Bell Labs between Claude Shannon and Alan Turing. They faced a choice: build a computer like a human brain (a neural network) or like an adding machine (a deterministic system). They chose the adding machine, not only because the technology for neural networks didn’t exist, but critically because telcos and governments required predictable, deterministic outputs, not a system that might hallucinate. This historical context highlights the primary challenge Nokia addresses today: reducing AI hallucinations to make the technology safe for essential, real-world networks where reliability is paramount.
Wagar then connected this to the modern concept of autonomous networks, such as the levels defined by TM Forum. He proposed that these frameworks were largely developed before modern AI and assumed a deterministic path, whereas AI introduces a new, separate plane of capability. He pointed to Google’s public journey toward autonomous networking, which leverages custom-built AI agents to move beyond simple event-driven workflows to a truly autonomous SRE model. Wagar concluded by positioning Nokia’s strategy as learning from these leaders, blending AI and traditional automation to inform their product development and establish new best practices for the industry.
Presented by Clayton Wagar, Principal Consulting Engineer. Recorded live at Networking Field Day 39 in Silicon Valley on November 5, 2025. Watch the entire presentation at https://techfieldday.com/appearance/nokia-presents-at-networking-field-day-39/ or visit https://techfieldday.com/event/nfd39/ or https://Nokia.com for more information.
Transcript
I'm Clayton Weger. I lead our AI practice in our, uh, IP division in, uh, what's called Cloud and Enterprise Strategy. So I know, um, originally you were expecting Mike Bouchon the celebrity.
I'm Mike today, uh, I work for Mike and do a lot of the work around ai. Um, as we were preparing the material here for today, we had some exciting things to show you that comes in the form of demos, and I want to kind of create a little bit of a through line to that, you know, and my, my kids are in drama, and so we talk about through lines and creating the connections between things. So, you know, a AIOps, um, the, the term itself, uh, I had been around like many of you for a long period of time, 32 years.
I started my, uh, my journey in being on the ops side of things, managing a 9 1 1 data center in Atlanta, Georgia, when I was 22 years old. All layer two for that. So that was quite a splash of cold water.
I constantly think back to that initial time when it was largely the information I could master the CLII can master my brain, the workflows, I can master my brain. And then we started writing those things down. And of course, we've brought to bear over the last many decades, tool sets to help automate our job.
And certainly when you look at ai, there's two facets to it for me. And you're gonna see both of those facets today. The first is, how do we apply networking technology to grow AI capacity?
That's pretty much a plumbing, uh, you know, uh, motion. We go and we build very, very large networks that are, uh, very large on scale and very complex to some terms. You're gonna see a, a, a bit a bit about that.
Also, we have this idea of throwing the word, the, the, the letters AI in front of everything like AIOps and thinking that it's something magical when we need to be really, uh, prescriptive about how we use AI tools. So, um, I was recalling, it was recently reminded of a, um, a story, 1943 and Bell Labs, Claude Shannon, father of information theory, modern information theory, Alan Turing, um, British scientists who obviously this is a, against the backdrop of World War ii, they were having a discussion at Bell Labs on how do we build computers, right? Computers didn't exist.
You had some logic circuits and things like this. How do we build them? And there's really two choices.
The first of which was to build it like the human brain, what we now call a neural network, and understand all of the, the, the, the, the particulars that go into that. The other ones to build like an adding machine, right? So it's perfectly predictable and gave us the same output every time.
And, um, really two force functions there that caused us to go the way of the adding machine. The first of all, the technology didn't exist. We now know it would take 80 years before we would get the technologies which allow us to simulate with some regularity, the human thought process.
Also, quite frankly, the practical requirements, right? You were building for telcos and for governments who wanted a deterministic output every single time, right? When you're breaking codes, you don't want to have someone hallucinating and then saying, oh, you're absolutely right, and giving you the second answer.
Um, and so I, I tell that story because it's important to understand that, uh, you know, Andy mentioned this idea of autonomous networks being defined. Here's a famous one from TM Forum. There's others that are out there which are somewhat similar.
You can think of it as levels of a self-driving car. The same idea. There's less and less human intervention or interaction as you move your way up the stack.
But I would propose to you that largely this was developed in an era before modern ai, right? This probably assumed tools and systems that would be brought to bear that would allow us to continually automate away functions that we knew existed. Things we could write down, workflows that we could take from our brain and put even all those corner cases into some sort of system.
But now AI's changed that. And one of the, one of the things we do when we talk about the AI enablement of our products that you'll see today is reducing those hallucinations. Because obviously we're dealing, uh, with mission critical networks, we're dealing with even, even you could consider, most networks in the world, um, are instrumental to operate on our day-to-day lives.
Certainly, we all know Maslow's hierarchy of needs. Some put wifi even below shelter, right? Or, or maybe battery is even below that for some of the, the folks like my kids who don't charge the phone as much as they should.
But the idea that we have this concept of autonomous network lives separately in many respects to AI's application against that. If we simplify that and we go from, again, something very manual, we can be using AI on the left hand side of this, we're gonna have completely manual operations, but perhaps use AI to create our playbooks or to create the, the type of, um, you know, the canon of information we need to manage our networks. And then as we step up, we can bring AI to bear as a tool across any of these different levels.
So there, they're two related, but different planes wanted to set that stage today. Before we get into, um, some of the demonstrations, uh, one of the things that I take a look at here is learning from leaders, right? So, um, we are very fortunate to have Google in the world for a number of different reasons, but Foley, who runs network over there, uh, he's given public presentations.
He also was kind enough to do it at a recent, uh, Nokia meeting, talking about their journey in autonomous networking. It's a little different than the TM forum framework that I showed you, but across time, they showed us some human driven workflow, event driven, which is largely where a lot of the industry is today. The, the vendor community is largely event driven now because of a couple key things.
Number one, they have a massive software development capability, of course, but also they have some of the, the world's leading edge ai, uh, researchers and, um, you know, you know, applied product. They're able to bring that Gemini expertise, that DeepMind expertise and run a truly autonomous network. And so this is, uh, sort of a public video that they did that talks about how they use ai.
You can see the concept of AI agents here, right? We'll talk about that today. Here's a playbook that's been automatically generated because of the impairment that's in the network.
When we talk about using ai, this is really a vision that, that I personally relate to, right? The idea that they can bring to bear this, this capability of AI build custom AI systems, right? These are not off the shelf Gemini.
These are, they've been clear that they're custom systems that they built for themselves and they're able to bring all this together and now have a very different SRE Type of motion than they would have had even two or three years ago. So you're gonna see some bits and pieces today, some demos for us. Um, some of it's ai, some of it is not.
You're gonna see us using those two pieces together to inform how we see, um, our product development, uh, motion, as well as what we think are best practices in the industry. And we hope.
