TNO075: What Does Network Operations Look Like in 2030? (Sponsored)

0:00Welcome to Total Network Operations, the podcast for all packet delivery personnel. Our mission here is simple to bring out great ideas in modern net ops. I'm your friendly neighborhood podcast host Scott Robon. In today's sponsored episode, we're going to talk about the shift that net ops is encountering and what net ops could look like in 2030. It sounds like it's far off, but it's not. This is the tail end of 2026. 2030 is only three years away. And so much is going to happen in those next three years. AI is moving beyond co-pilot and recommendations towards systems that can help us diagnose issues, plan responses, execute approved actions, and even support rollback. Now, one of the things that I got very jazzed about and preparing for this episode is this tension and the difference between what we're hearing about, well, you need to be doing this

1:04agentically today. And technology providers coming to us and saying, this is all right now. And knowing that there are plenty of people who sit in the operations chair who don't see that yet and want to build trust and want to see, you know, how are these agent existence going to behave and what does it take to get to the point where I'm comfortable putting them in production? And to help me through this, we're joined again by Backbox CEO, Rekas Shanoi, and Field CTO, or Fon Kimji. We're going to look beyond the current wave and see what's coming, but talk about how to get from here to there. So I'm very excited to follow up on them with this conversation. Thanks for being with us again, Rekas and Fon. Let's jump right in. So what do you think respectively about that setup, you know, the tension between state of the art today and where agentic AI is going and where it might be in three years from now? Yeah, I think three years is a very short time, but you know, for once in our, you know,

2:08for all of the changes we've seen over the last 30 years, I think the next three years are going to be dramatic in terms of the way the world changes. But I agree with you that our customers say the same thing, which is it's pretty crazy what people think they can do. And most importantly, the first thing I hear from customers is we're in the business of keeping businesses up and running. Sure. We're not in the business of AI. And so we get these crazy things like, hey, how can we use more AI and cut heads? But that's really not the point of this exercise. They're not in the business of just fun technology for the sake of technology. Where's the accountability? Where's, you know, when does AI become accountable to the business and how do we throw it in in a responsible way? I guess that's the most advanced way of looking at it. A lot of our customers go like no AI, not now, not ever. In fact, don't use the word renaissance. It's got AI in it. Like it's like that kind of fear of AI. And so we think the world moves very, you know, at a pace that I think people are still

3:12going to be surprised how fast we move. But we think even Gartner says the biggest net op shift happens now, you know, compared to everything that we've said before for a different reason. And that is if we can build accountability into the system, we just don't have enough people to do what we need to do today around net ops. Right. We'll get pragmatic around it. Let's kind of start walking in that direction. But their fears are not unreal. It's not a simplistic idea of new technology is scary at all. It is this idea of accountability that is scary around AI. Yeah. I mean, Irfan, you have a very customer facing role here. What can you add to what Rick had just said on this front? What are you hearing from the folks that you're talking to? Yeah, it's been great. You know, first of all, you scared me when he said 2030. I was like, oh my god, it's 10 years away. No, it's three years away. That's right.

4:12I hope that's me up on us. And yeah, a lot of the things I'm hearing, it's, oh, great AI. Oh, great AI. You know, it kind of reminds me about years just over a year ago when we started talking about, you know, how are we going to use AI at backbox? I was like, man, another chatbot, not another, you know, quirky image creator thing. And that's the thing with all new technologies, right? There's a pathway to, hey, what's the, what can I do with this? What was it created for? What can I do with it? You know, how can I break it? And then it gets to a point where it's okay, well, how is this usable? And I think we're at a point now. And that's why I think it's, it's a great time to start thinking about our roadmap for two to three years from now is how are we going to implement these technologies at the end of the day? Every technology, every even person at the company, is it tool in your toolbox? And how are you going to use those tools to be successful? And so if we think of AI, you know, it's not just another quirky thing.

5:13It's a tool in your toolbox. Well, is it a screwdriver? Is it a hammer? Is it a laser? Should it burn us through everything? You know, figuring out how to use it. And I think people are starting to ask that question. It's no longer just a management, hey, here's a mandate. So you have an AI budget, you no longer have an infrastructure budget. Right. We'll have a business to run. And so, well, okay, well, now that I'm being sort of forced to do this, the technology is at a point where it's catching up to be able to handle the tasks that help me in my everyday life. We're at a great inflection point now to be able to take that to the next level. Very, very true. If we could step back just a little, I do want to pull on the Gartner comment, you know, the biggest shift in 20 years and accountability. I think the Gartner comment, I think we all resonate with all of us, right? I can't remember a time in my career where we've seen things change this quickly, which I think contributes to some of the comments or fondness made, right?

6:15You know, look, some of the eye rolls, oh, AI again, AI again, AI again. And you know, there's a lot coming at us quickly. It's not all one thing, you know, even, you know, using one shot LLM interaction for generating a script, right? It's not the same as letting an agent loose with guardrails, you know, in a production environment. And we got to keep that in mind, you know, AI is not monolithic. On the accountability point, I can't remember the quote and I apologize to whoever originated it from mangling it, but you can't hold a machine accountable. You really have to hold people accountable. And I wonder, Raka, how you connect those dots in implementing that mandate for accountability. I think it's super important. What do you think? Yeah, I think the scariest thing you can say to a customer is self-healing AI. We had a customer that very, very large service provider that put AI basically clawed on top

7:21of back box just because, hey, I can do that. Let's plug it in, see what it's able to do. And for a system that had all of these checks and balances about audit requirements and how you access systems and what change gets approved and all of that, we had suddenly given AI the ability to go make changes simply by connecting it to a tool. Okay. And all of a sudden, it was often actually looking at things it had no business looking at customer data and so on and so forth. So the thing about this fear is not on bounded, it's actually happening. It was like, there you go. It can be that quick. So the question is, what are those things that our fund was talking about? They can be responsible, they can be managed and so on. The other side of it is that point I was making about workload. And so that's where the more pragmatic approach comes in. It's like, here's all these hundreds of hours of work we're doing. And we find a smart intern that we can hold accountable with an AI.

8:22So don't do stuff. Solve my workload problem in a more responsible way to go solve it, right? So to Gartner's point around the shift in net ops, it's driven by a couple of things. As AI is, you know, also in the hands of bad actors. So the pace at which risk mitigation needs to happen just can't happen with humans in the loop. So what are those things that we can do smartly? And then the second thing is AI itself is requiring a huge infrastructure expansion. So scale, you know, so Irfan had talked to a customer the other day and he told me this, it's just stuck in my brain. They looked at what it would take to handle patching for the systems that they had, which, okay, let's admit it. A lot of people do it once a year, maybe not, right? It's that rad today. So let's go do it the right way. Well, 23 people, I don't think any company no matter how large it is is going to go higher 23 people to do network patching in today's day.

9:25So it gives you a sense of the scale and the problem and how big it's become. And so how do we get pragmatic around that? Solve that problem in a smart way without giving the keys to the kingdom to AI. I, on that point in particular, I think that's a very important distinction for network infrastructure versus compute infrastructure and applications and so forth. You know, that whole, you know, I can, I can, I can spin up VMs or containers on another, on another server. Well, I patched this one and swap and manage things that way. And I don't want to make this sound easier trivial. I know that's complex too. But in network infrastructure, you have a level of fragility and dependence on the pipes working that drives the, I'm going to upgrade, you know, my switch OS and my, or patch things on my switches or routers once a year, maybe twice a year at best, right? So your point is well taken.

10:26There that needs to be approached super carefully. And you know, it doesn't, you know, that's like one of the differences between DevOps and net ops, right? I'm a, I'm a big fan of taking DevOps principles and applying the net ops where they make sense, but also recognizing the differences for the network infrastructure. So you've, you've talked about creating efficiencies and you're kind of, you know, what, 23 people that actually just manage, you know, patches and upgrades. That's, that's crazy. That's not going to happen, right? So talk about maybe, maybe unpack that scenario a little more where, you know, how can agents applied appropriately with the right guardrails help us in these scenarios? And maybe, maybe it's not just efficiencies. Maybe it's also, I haven't been able to fill those job wrecks for nine months. Like I need, I need more staff anyway. This can help me with existing workload. I think you pulled on a good thread, Scott, around DevOps and net ops.

11:28So I think one more thing changed in this world about, you know, why the 23 heads are, whatever, right? Right. I've never used to do it before. Why do I have to do it now? Well, I think the fear of AI enabled hackers is real, right? Mythos is not a theoretical idea. Project Glass Wing is not a theoretical idea. What's happened is if we talk to any customer, they will tell you that the piece at which they're being attacked has gone up. Yeah. It's a very real thing. But also vendors, you know, infrastructure vendors are coming in strong and fast saying thou shalt upgrade ASAP. Don't ask me why. Just do it, right? Because of Project Glass Wing. So it's a good thing, but the amount of work that it puts on nest prices that are already over taste is not unreasonable. It's gone a little bit. Hey, why, so that's where we think about, you know, the sort of the, let's get pragmatic here. So if you're going to have a program around keeping your infrastructure up and running and

12:33in keeping it secure, keeping it resilient and all of that, and it's not just your handful of, you know, edge firewalls, it really is all of your infrastructure, then there's a meaty problem that we can solve with AI. And we think about it as, what if you had at your fingertips a smart intern reading the thousands and thousands of pages of vendor recommendations and figuring out pure, very accurately, you know, what actually applies to you, what's actually a problem. What if it said, hey, these workarounds apply to you, you've already applied them, you're in good shape. Now here's the five you need to. That's actually hundreds of hours of savings without going off and doing something in your environment, right? That's an example of like massive improvements that is pragmatic. And when we talk to customers about that, they go like, when can I have that, right? And it's still AI enabled under the covers, but it's not the crazy AI going off and having

13:34access, admin access to all of your infrastructure. And then when you go beyond that, you go, okay, so then if I have all of these, how do I patch these systems? What if AI could create an automation very quickly? And say, here's an automation, doesn't run it, shows it to you in a visual way that you can understand it. And now you've kept control, you're still accountable and you choose what you want to run with, but you've saved hundreds of hours. We think those are the kinds of pragmatic ideas that AI can actually offer you. I think the visual display of information and being able to show people in a concise, graphically impactful way is super helpful, right? Because we're drowning in data today. And I say data intentionally, not all data is information. So that's a good practical add on there. Do you have you thought through or are you thinking through how, all right, like take

14:39an example of, I have a set of CVEs that have just been announced and they affect multiple infrastructure vendors. And I'm an operator, right? And I have brand A and brand B in my network. Tell me how back box can help me navigate, looking at the underlying infrastructure for both of those vendors for the same CVE. What does that look like with your agentic approach? Yeah, it's super interesting because what we did was we mimicked what our smartest network engineers do every day. First of all, to your point, the biggest problem is I'm a Cisco expert. I have some meristas. I have some forties. I don't know what I'm supposed to do. So what we found was typically when they're looking at something, let's say, like it's a give you a simple example, well, why is HTTP enabled on this box, right? Like they're doing that search. Well, the first thing that I do is where else is it, you know, where else is it enabled?

15:39So now we have AI sitting there watching our network admins do those things and find those things on these other devices and comes back and says, Hey, if you're worried about that, here's five other devices. Can I create an automation to fix that for you? And so it's sort of that mimicking that human behavior and giving you smart recommendations. And what we found our customer saying is somebody was reading my mind and actually doing some of that triage work for us. So that's an example. And that seems simplistic. But when you think about the thousands of different versions and devices and all of that, even from a single vendor and then you multiply that by the multi vendor, the complexity of that problem is hit that point where it's not human manageable. And you just keep hoping you're a really smart network operations engineer doesn't get hit by a bus. Right. Well, there's things as well. Like, you know, if you think about, you know, there's various, I think we're talking a little bit earlier in the chat about this various levels of maturity of what customers are willing

16:42to accept for AI and their insurance, right? So even if you take it a step back, one of the things we implemented was AI on the back end. So as part of this is, you don't necessarily have to even run it in your environment. Just this triage portion, right? If again, to that to one earlier, if I've got some Cisco, some 40s, some Ristas, whatever, if you go to all their different websites to figure out, what do I need to do? How do I need to do it? They're all different formats, different colors, different, you know, some dark mode, some build, somewhere over here, some over there, some are written differently. You got to parse through all that and figure out, okay, well, what do I need to do? What do I actually need to do? And does this affect me? And so what we've done on the back end, even before it has the customer environment, is use agents to just go and query all that information. Now, you could go today and build that yourself if you're comfortable with building cloud agents, go and ping all the websites, it's just what we do. Sure. Yeah. Pull all that data, normalize it, and just give you a view. Here's what the vulnerability is.

17:43Here's how it affects me. Here's what I need to do about it. Here's what the vendor says I need to upgrade to this version or this patch level, or here's a configuration item to go in and make the changes. And then where we took it one step forward is say, here's all your affected devices because we know what the configurations of those devices are. And to go and fix that, just click this button to create a remediation task. And that's where you have the AI model go and create that remediation. It's a here, I've now created this remediation task. You're the expert go vet it, make sure it's correct, all the steps are correct, and then go run it. That what would take days, weeks, you can now do in minutes. Yeah. I think that's a perfect example of how to build trust in agent systems. Right? You've got to focus on offline, non traffic, impacting capabilities that solve a real problem. Right? And over time, somebody does this for a year, 18 months, and they find like, hey, me

18:46clicking OK to continue on that remediation task is almost a rubber stamp. If I've done that for 12 months or 18 months, I might be ready to say, OK, I've got 99.5% confidence that the things you recommend really need to be done. You can go ahead now and start automatically creating those remediation tasks. Does that fit your vision or workflow that you see coming or correct me if you have a different way of looking at that? Yeah, I think that's the Nirvana where you're able to trust it on its own and do it automatically. Again, to your point about accountability earlier, the person who's accepting that trust is the one accountable, not the machine that's doing it. Right. So you've got to make sure that the source data is correct. Right? And that's one of the biggest things when you're building LLMs, and that's something that individuals can't necessarily do because you think getting the large library of data where if you're trusting a public LLM, that's pulling data from all over the place.

19:47You don't necessarily know where that's coming from. You've got to be able to trust the source. And so one thing I'd encourage listeners to do as you're talking to your vendors and I've worked with vendors for over 20 years on both sides of the coin. Sure. Ask them that question. How are you building that model? What data are you putting into that? And so that was one of the key principles for us as we were building this out is the data that the agents are going to be building stuff off of has to be accurate, has to be clean. Otherwise you have garbage in, garbage out and good luck. So everything built, if you're building a network automation for Cisco's, you've got to have clean Cisco automation, that the thing's learning from. Similarly for other platforms and make sure it's able to identify across the board which ones which and give you the correct output. But that's the super important for trusting accountability. But I think it happens. Let's call it that. So cognitive decline, right?

20:50So I've been hitting that button over and over and all of a sudden something went wrong. Then ROWA, what do I do? Right? To our bounce point of like, hey, AI is going to agree with you. Oh yeah, that happened. Let me just go another way, right? Is this going to happily, very politely go do that? And then what? So to us, the provenance of what that, where did that change come from? Why was it important and all of that? That's when you start looking at all of that. More importantly, if you've got more and more experts just sort of clicking on it and not really knowing and so on, how do you give them at their fingertips when stuff happens, their ability to take back control all over again? And that's where everything you build should leave you with that sort of visibility both visually and all of that without having to become coding experts all over again. They've stopped doing that. They've been clicking the button and all of that. Now stuff happened. How do we give them control one more time? So those are the problems we tend to think about.

21:50And when you build it the right way and you build the same accountability, you also build very cool integrations into other systems that are also AI inagentic friendly. So where is that ticket that said this is the problem and this is the solution and all of that? And what is it being fed by? And we think that the quality of the data you start with and your provenance all the way down and to be able to show this was the change and this was who approved it and this is how they did it and this is why that's the way it is. Suddenly becomes even more valuable, not less. It's not like you're abdicating all of that data to AI. You're actually building it in a way that gives a human visibility. Sure. I do think there's even a path with in this scenario where I am ready to accept automatic remediation after a certain period of runtime that I can still keep flesh and blood humans accountable and responsible. You might have somebody in the net ops org, could be a front line person, could be manager

22:55director that says I authorize this system to now go in this automatic mode and still hold the person accountable. These are things we got to think out from a workflow and accountability perspective, but that doesn't seem unreasonable to me. Once somebody's got a level of confidence that these agents are doing what I need them to do and I'm now ready. Assuming I have the right type of error handling, 99.5% of the time they're great, but that 0.5%. How do I handle those new scenarios and pop it out to a human when there's ambiguity or the agent or system doesn't know how to handle it? That seems like a reasonable framework to me. What do you think? It's huge. Also a path to cover your end. Having the first foundation to anything we do is back up the data. Make sure that it works. Make sure you can roll back to it.

23:57I got to roll back. I got to give me the previous state, the last known working state. So even any automation we build, even pre-AI, we've started the backup. So we've got backup happening daily, whatever that is. But before you run this automation, back it up. Test it. Then run your automation that was either human created or AI created. Then test it again, back it up again, then you're done. So you have the full workflow. So you're not just saying, oh, I trusted Go. No, I trusted it. But I have a backup plan just in case. And I have off ramps in case something goes wrong. I also think that the human will spend more time on the business side on the strategy side, which is what's also often falling down. So one of my customers that I met the other day said, they're very large healthcare business. So for them, that network isn't just a network, isn't just a switch. And so on, it's actually in the surgery room.

24:58And that thing better not go down when surgery is happening. Humans die. They think of accountability and business in that way. So when they're in that situation, what's actually acceptable? What's this thing that is this patch that needs to be applied? And so on has a whole different level of meaning. Doesn't mean that they don't do it. It's just that they've got a very smart network operations engineer who's spending their time understanding the business context and applying the business context. Versus, you know, when it's finally down to, yes, it's been tested. It's been backed up and now we're ready to run it. That's the stuff that they want to get out of that business of doing. Let's not do the mundane. But the business context and how do I infuse that into AI that I think is going to get really, really interesting over time? 100% agree with you. And like if we, if we generalize this into, okay, how does this impact what's happening to people in net options and net ops teams?

26:00I see this as an opportunity to elevate. And you've called out increased focus on connections to the business. I think that's super important. That's something I avoided like the plague for the first couple decades of my career. I just want to play with the tech. I don't want to care about what the account and say, guess what? You need to care about the business and what the account and say, whether you realize it or not. So tip of the hat to you, Rhica, for calling that out. And I'd even bring it to, you know, you get to think more like an architect or a, you know, a resource manager or an agent boss, right? And I'm not equating all three of those things. But you get to rise above and say, I need to care more about what the end to end behavior looks like. What happens in this particular workflow? And I'm managing different tools to make sure the workflow, you know, whether it's patching or something else that I'm using agents to use. I talk about this all the time, you know, here's an opportunity to think like an architect and to have like more, more of a bigger picture here.

27:03What other generalized things like if we go to 2030 again, remember, if on it's only three years away, what is a net ops engineer look like now with these new tools and managing agents? What do you, what do you think? What are your customers speculating on characterise that for me? Yeah, I like to think of things, I look to look back at history and see how things made us evolve and how we can drop parallels. So if you go back, you know, what does it feel like a hundred years ago, at least, whether there's things called typewriters that turn into computers, that turn into laptops, remember those things. Yep. And what it did was it made things faster. So if I was trying to write a letter, communicate across the board, you know, a hundred years ago, I had, I still play with Penn, it's more of a fidget toy now than an actual tool. But, you know, yeah, I had to learn how to articulate that and that would take time if I made

28:04a mistake. I had to go back and fix it and maybe start again, right? It was incredible. Then came these typewriter things. I don't know if you saw in line, I don't know if it's true, it was not or not, but the old Steve Jobs letter that said, we're now using this thing called a computer. Stop using a reason we're processing. We're not using this old typewriter things anymore. Right. Because it made it easier to move faster, to fix mistakes and be clear and more optimized. Sure. And so now if you think about that and try to drop parallel with what's, what is a network engineer architect going to be doing? Well, one, it's, do I need to sit there and, you know, on a Friday night with, you know, your drink of choice and some pizza and the upgrading devices, you know, a couple every every Friday, Friday and Saturday night. Or can I automate that? And now I take it a step further and say, hey, wait a second, I don't need to necessarily be up all night. I can schedule it, you know, set it in, forget it.

29:05Remember the old infomercials. Right. And this was showtime rotisserie barbecue, I think it was cold. Anyways, I digress. You're making me hungry now. Sure. Lunch is a long time off. Yeah. So you take that now then level it up further to your point about understanding the business aspect of it. Now, as an individual engineer or architect, I spent all my time doing this one task. Oh, wait, they still have my rest of my day job, these other 10 tasks. But if I can streamline that and become a manager of these tools doing this effort for me, I am now monitoring it, speaking it, making sure it's running correctly. Now I still have to be the expert in that because I need to know if this one is doing it correctly or not. I need to know if this ran correctly, I need to be able to verify. And you now become a manager of agents or processes to streamline your job and make it faster and smoother.

30:06Now, we've got all sorts of different technologies that we can draw examples from that made our lives quicker and easier. But I like to use the driving example, right? Before it was, you needed to be so focused, you had to press the clutch, get the accelerator down, shift correctly, right? Then it became automatic transmissions to where you can barely even find manual cars anymore. Now, it's automated driving cars. The focus on driving isn't necessarily because you need to be engaged in it as much, but the focus is there because you need to make sure it's working otherwise the lives are at stake and be if you're not, you're probably going to get a ticket and maybe a license suspension. So there's other reasons to be focused on it, but you still have to maintain the knowledge of how to do those tasks. And so while things aren't getting more automated and more going, you still got to know what's happening. I will say, the whole driving and evolution of driving and analogy has hit me very, very

31:08hard and practically the last few months. I have an older vehicle that is great. I'm a Toyota guy, 200,000 plus miles, it's doing fine. But for longer trips, I'm starting to rent cars that have lane assist, that have cruise control with radar. And I really have found it does take toil away from the automobile operator job. And I am not as tired and I am, I'm fresher for those meetings I'm trying to get to in DC. And when I come home and don't, you know, instead of giving my wife the worst of Scott at the end of the day, I'm a, I can engage in a little more conversation because I'm not, I'm just not as exhausted. And I feel like that's a real important concept to grasp when we talk about how things are impacting and going to continue to impact what we see in network operations. I, I, I, thank you for, and for, for inserting that. That's, I've been wanting to say that out loud for a while.

32:09You just gave me the opportunity to. So, Welcome. And even if you think of like like network operators are kind of in the, in the knocks side a little bit where they're using like observability tools to see, he is my data flowing from here to there correctly. Yes. Okay. That's not what happened. Something changed. And so you need to understand what changed to change. Well, where is that data? Well, I've got a bunch of backup data I can see from this timestamp to that timestamp, which was in between where this flow stopped working. What changed? And you can then pull that data in. So you can then integrate tools together, you know, what your observability tool is, what your resilience backup tool is. And now you have AI agents that connect those together and pull that data saying, hey, wait a second, you know, at 2 a.m. this stopped working. Well, query the tool that knows that and pull that data and just stitch it all together. And you're able to do that type of analysis much, much faster than you were previously. Before we leave this topic, I do want to get to, you know, what you're building toward.

33:13But are there any other significant impacts you see on people in the operator seat? You know, any other impacts to the role in the next three years? I'm not fishing for anything specific. Just like, is there anything else that you're, you see coming up that we haven't called out yet? I think the biggest thing is there's a myth that we're going to have fewer network operators. I think that's probably the biggest thing. I think we'll have network operators handling a much larger remit. And I think we'll have more exciting network operations jobs. You know, there's sort of this general idea that network operators are really always overworked, which hasn't changed in 30 years. And that's true. But I think the job gets more interesting, hopefully less overworked. But I cannot imagine an environment where we are abdicating everything. So I do think the role changes, the remit grows dramatically. People do more. But it's not a case. I don't see a scenario based on, you know, the customers we're talking to, especially

34:15in large enterprises where you have fewer of them. I just, that's the one I would just like bet on that we don't have fewer of them. Yeah. One of the other pieces as well is if you think about to that point of growing teams and activities and stuff is budgets. If you start thinking from a financial perspective, again, leveling up kind of where you're headed in your projects, businesses were really driving funds towards cybersecurity and less towards infrastructure. And infrastructure seems like I need to do these things, but I don't have money because all the business money's being towards other IT areas. And IT at the end of the days are cost center. While we enable the business, we're not making the business money, but we're enabling them to do their thing. And so infrastructure teams not to do stuff, well, hey, all of a sudden there isn't AI budget in pretty much every business. And if you think about it, you don't necessarily have to go buy, blog, or open AI, or something to fill that budget. If you can say, hey, there's legitimate business outcomes I can achieve by a tool that's

35:17using AI, I can leverage some of that AI budget to pay for it and make my life easier. So all of a sudden that budget to automate things that I wanted 23 FTEs for, I don't necessarily need because there are tools that can help me with that, that leverage the budgets within my business, within the sort of constraints of what I'm allowed to do. And just thinking a little bit differently, a little bit outside the box to be able to manage day to day activity. Yeah, you're not the only vendor in technology supplier to raise that just in the last few months. Sometimes it's just easier to procure a product or system that does it using AI and other mechanisms versus trying to fund it via DIY projects. I'm not trying to discourage the builders out there who do want to stitch and build things together themselves, but one size does not fit all, right? And so your point's super well taken there. Yeah, and if I could give an example as well, I was talking to a couple of different financial institutions, both very large, both similar size competitors of each other.

36:24But one was really struggling with automation and trying to figure out a solution, is there tools I can use and they ask, Claude, is there a tool that can help me with this? And that's how we got engaged with them. Whereas the other one, we have been to talking to the meta conference and they're like, actually, we have 20 automation engineers. They're just coding. They help ensure every facet of the business. They help finance. They help here. They help other technology. We said, hey, can you help us build network automations? And so we're, that's the approach we're taking as the business. And that's okay. You make that decision for what makes the most sense for you. And there's different ways to approach it. That's totally cool. For sure. Well, so we've talked a lot about where we think the puck is going, both today, things that we see and our customers have been dealing with. You're obviously driving toward, I think you mentioned road map or fauna, I don't mean to make you do a road map readout. But what should be we watching from back box for what you are trying to drive in the

37:28markets that you serve? Yeah, the biggest thing for us is operational efficiency. We understand that network operators, we're purpose built for network teams. And we want to make your lives easier. And so the biggest thing for us is getting customer feedback. And that's one of the reasons I love my job is, is really trying to understand, what's the challenge? Let's peel back. You know, that's those five wise. You know, what's the challenge? Why is it this? Why is it that? And keep going till we understand, what's the root of the problem? And how can we fix that? And how we make that easier? And so if you look at kind of where we're going with with automations, leveraging AI for making your life easier, doing things that are not, yes, AI is the cool buzzword in this world we're here to talk about. But there's a bunch of stuff we're doing that's not AI to help with that too. Sure. And streamlining and making things more efficient and making it live easier. And that's super, super important for us is really listening and understanding and then and driving those use cases home. So for us, I think just do so how does that play out?

38:30There's two areas where we see massive, massive gains on efficiency for customers. So the first one is how do we give you decisions in your hands, right? Make better decisions. So that's an AI enabled feature. But really the fact that AI is there is secondary to the value that the feature provides, which is the hundreds of hours we take from you to make better decisions. And then the second is where we say, well, how do we make your life better? And making life better is around the area of like, can I, you know, for engineers who need to be both subject matter experts on every device and every vendor type and so on. And also be good at coding. We take that away from you. You don't have to be all of that. We give you automations that are very custom tailored to a specific vendor's device behavior and so on and then give you something in your hands. So it's sort of making life better, making you less of needing, needing to be that expert on all things and giving you something that still keeps control in your hands.

39:34The next piece is really about how do we fit into your ecosystem in a way that is complimentary. So when something fails, how do we automatically create a ticket? How do we enrich your get how about we enrich your CMDB? How do we become that source of truth? All those things that, you know, by connecting to other devices and making everything work together better, we save an enormous amount of time and energy that is today's sneaker net and, hey, do you know what happened? Do you know what happened? Kind of things that we solve in meaningful hundreds of hours of manual labor today that's not interesting to anybody today. Yeah, I would layer on top of all of this too. I think to the comment on, you know, elevating and expanding our remit for what we need to operate and manage. I think the amount of network infrastructure that's going to continue to be built out over the next three, five, ten years will be massive.

40:38Much of it driven by AI workload processing, but also just the way that networking is embedded into almost everything we use. So I would encourage our listeners, anybody who's concerned about their job being AI to weigh that is not a future I see. I see it as another set of tools that are going to help you do all the things that we've talked about here and more, things that we haven't even envisioned yet, because it's really hard to see around the corner, right? We can we can easily visualize, well, this is how it's going to impact my job today, but it's harder to imagine what does the next version of my job look like? And that's exciting to me, you know, or funny, you talk about things you love about your job. I love this part about my job to talk about things like this with people just like you who are bringing real solutions to the table. So I'm so appreciate you coming on today. You know, we've talked a lot about what we think the future of NetOps is going to look

41:41like and we haven't said it directly, but you know, it's going to it's not going to be whether or not you use AI, but it's how you use it and use it well, right? I think we've we've touched on that many different ways today. For back box, it seems to me, you know, this means you're building toward a more automated future where AI is actively engaged to help your customers and your network teams make better decisions and move faster. Any other final comments from you before we call it a day? I just tell our network operators like raise the bar, right? Don't accept simplistic clawed as your strategy. I don't think they're thinking that they're thinking clawed as being thrown at them, right? I would just say raise the bar. We built a network operations teams over these years around accountability. Well, raise the bar, require it and and see how vendors like us help you with that. And if we don't, don't don't fall for that AI tool. Yeah, I would say the, you know, the teams that are going to thrive in 2030 and beyond,

42:43it's not going to be the ones that resisted AI because like you're saying, it's another tool, right? They're going to be the ones that figured out how to sting control of it and use it to their advantage. And so that's what we really focused on. Very much agree with you all on that. Ulricha and her fun. Thank you so much for coming on the total network operations today. Really appreciate your time. Thank you for having us. I'm free. Everyone who tuned in to listen and watch. Thank you for being with us for yet another conversation here on Total Network Operations. Please send us your feedback. Hit me via DM and LinkedIn or send us official follow up at packupusures.net slash follow up. Thanks again. I will see you next time on Total Network Operations.