# Talking AgenticOps and the evolution of artificial intelligence, with Akshay Bhargav

## Метаданные

- **Канал:** Cisco
- **YouTube:** https://www.youtube.com/watch?v=3cUBEG4IBOk
- **Дата:** 06.05.2026
- **Длительность:** 30:03
- **Просмотры:** 189
- **Источник:** https://ekstraktznaniy.ru/video/50038

## Описание

In this episode, host AB sits down with Akshay Bhargava, Cisco’s VP of Product Management for AI Software and Platform, to explore the rapidly evolving world of Artificial Intelligence.

As businesses race to integrate AI into their operations, the shift toward "Agentic Ops"—where AI systems take on more autonomous roles—is becoming a strategic imperative. In this episode, we discuss how Cisco is building essential guardrails to ensure AI remains safe, reliable, and scalable. We also dive into the critical risks modern organizations face if they fail to adapt to the pace of AI innovation.

Learn more about Cisco’s commitment to innovation:
Cisco Official Site → https://cs.co/6053CDESl
Cisco Products & Services → https://cs.co/6054CDESm
Cisco Solutions → https://cs.co/6055CDESW
Contact Us → https://cs.co/6056CDESo

Follow Us on Social Media:
Cisco LinkedIn → https://cs.co/6057CDESU
Cisco X → https://cs.co/6058CDESq
Cisco Instagram → https://cs.co/6059CDESS
Cisco Facebook → https://cs.co

## Транскрипт

### Segment 1 (00:00 - 05:00) []

Welcome to Cisco Techbe, the podcast that explores the people and stories behind what inspires the newest innovation. I'm your host AB and I'm very thrilled to welcome Akshai Bargava, VP of product management for AI and software platform. A welcome to Techbe. — Hey, thank you. It's great to be here. — Absolutely. And we got to see each other the other day for another shoot. So, this is kind of like our second round. — Yeah, it's like round two. Yeah. Awesome. — Let's do it. Well, for this one, we're going to jump into Agentic Ops. Um, which I know is sort of like the big thing that's that Cisco is talking about these days. So, kind of want to pick your brain a little bit. Uh, specifically how it's transforming the way we do work. What are your thoughts on Aentic Ops and why is Cisco sort of really deep in this? — Yeah, I mean, first of all, I got to say Aentic Ops is a really, really big deal and it's all about transforming how AI is transforming into this next era. You know earlier AI was all about you know you might go into chat GPT you ask a question you get an answer that's more like chatbot oriented right agentic is more about letting AI act instead of just respond now it's acting on your behalf it's doing operations it's researching stuff and it's executing things — right — and you know there's a bunch of tools now that have become really popular um and people who are using them are seeing exponential productivity you know whether it's writing code like if you use cloud code or codeex or also like new tools like maltbot cloudbot even um anthropic has like co-work so these tools are really powerful now it brings me to what is agentic ops so that's a gentic AI a good analogy is like you know like what devops was for cloud devops was all about like helping you make sure your code is production ready for the cloud think of agentic ops as very similar for AI it's all about like helping AI be safe be secure be trustworthy so that people can deploy it with confidence. — I love that. That's such a great way to bring analogy in because a lot of people understand the other side of things, but this is evolving in real time. So, — yeah, it's changing fast. And I mean, there's a lot of enthusiasm about Aentic AI and more specifically Agentic Ops as an enabler for our customers to start using Agentic AI because they need that trust. I mean, customers are saying, "Hey, don't just give me I can't just take any AI. I need AI that I can trust that my customers can trust. And a Gent Ops is like that operating system and capabilities that lets our customers trust AI. — That's right. — So, it's really important. — Yeah, absolutely. And I like the trust uh aspect to it because that's something that's really important to Cisco overall in terms of all of our technologies. How do we, you know, we talk about ethical AI and you know, there are a lot of considerations there. How are we making sure that the AI solutions and technology that we give to our customers is trustworthy? — Yeah. Is such an important thing and I think it's one that frankly we're very differentiated in Cisco in terms of how we're approaching this and how our customers are able to adopt it. So a key part of it is just recently, you know, the other day I was in um executive symposium and I was chatting with the chief security officer CVS Health. Now CVS Health is one of North America's largest healthcare. — Absolutely. — And they had they see an opportunity um and so they've built AI for their end customers and the value prop of that AI is so compelling because it reduces the cost of care. It provides a better experience for their users. So very powerful AI, but they needed a partner — who could help them make sure that AI is going to be trustworthy and safe and secure and give them the right visibility, observability into the that AI, right? — Um, and so they partnered with Cisco and they talked about how they're using Cisco's AI defense product to help them understand how their AI is performing, right? — And understand if it's safe. And there I mean without getting too technical there's a whole bunch of things that Cisco AI defense does for them. Everything from the supply chain which means understanding where that AI is sourced from running something called validations to make sure like hey what are the vulnerabilities or potential risks that this AI has and Cisco has developed some amazing frameworks um that are industryleading on AI safety and security. So we have this massive framework that says here's all the attacker intents right here's the 150 different kind of techniques or subtechniques and how are your AI in terms of your application your models your agents how is it performing against that right — that's a validation and then in real time we have guardrails to protect it so you put this all together and now you could see why customers like CVS are like I need AI and with Cisco AI defense I have trustworthy AI — I love that and I'll tell you what as I am a CVS to let you all know. So, I mean, I I feel pretty confident now knowing that the company I work for is actually also helping keep CBS safe. So, you know, it's an industry where you really can't afford to have um whether

### Segment 2 (05:00 - 10:00) [5:00]

it's a hack or something else, supply chain, you know, people need their medications, they need that healthcare. So, — exactly. And of course, CVS Health is just one example. We have many customers like that. But I think it's a really good one because it's a you know, a regulated industry, just a you know, it's a Fortune50 company. It's a really marquee name. Yeah. And I think it's one that is leaning in heavy into AI because they feel both the need from a competitive standpoint and they also see the opportunity to help their customers. — I love that. I think that one of the challenges with AI is that you have to balance u innovation with risk management. Yeah. — So, how do we do that? How do we maintain that balance while still kind of taking advantage of the incredible technology that is AI? Yeah, it's such a pertinent question especially right now because you know if you ask me this question maybe 10 years ago there's kind of like this um balance bar between like security on one side and speed of execution on the other and you say hey which one do you want and now the reality is the world has changed drastically and waiting for AI to be safe and secure is too risky if you don't lean into AI so the biggest risk that a company might be taking is not getting into AI fast enough. Got it. — And so that is how the complete dynamics of the industry are changing. And why you might say, okay, why is that? Well, it's because AI is moving so fast. It's such a transformative technology that if you're not using it and your competition is you are falling behind right now. Yeah, — the time is now. And the reason is it's not trying to be alarmist. It's just being realistic. Like if you're not moving now with AI, you're falling behind. and uh and you're falling behind at an exponential clip, — right? — So the reality is today when you think about balancing innovation and risk with AI, it's almost like you have to lean into AI and what you need to do is lean into AI in a way that's also safe, right? — And so this is where having the right kind of partner um technology partner, services partners are really important. And one of the things we're doing is we're helping our customers not only lean into AI and accelerate their innovation and their speed of uh innovation, but also do it in a way that's safe and fast. And that's where AI defense comes into play. I gave you that example earlier with CVS. And you know what's happening is right now I see three types of customers coming to you know to us to talk about AI defense. The first is customers who say I'm a business unit in a big company. I can't wait because if I wait I'm falling behind and I might lose in the market and then they're coming to us and say hey by the way we already released this AI can you help make it safe and secure and uh can we buy your software like today right — so that's one camp it's real and it's happening — the other is there's customers who are you know saying hey we want to make sure that all of our applications I'm a platform team I'm a security team this is kind of like the CVS case I want to make sure all of my apps are safe and secure and I want to make this a best practice and incorporate AI defense into that. So that's another example. And the and then the third is you know customers and service providers are saying we want to make sure that we help our end customers be safe and secure build you know the software with best practices and we want to incorporate AI defense into their best practices. So, we're seeing all three motions happening and happening fast and and we're definitely leaning in to help our customers and nothing gives us more satisfaction than seeing our customers have, you know, safe and trustworthy AI out there. — Absolutely. And, you know, I want to just dial in a little bit more on that uh the fast aspect. I of course your team is really in at the heart of all these innovations with within AI, but is it still surprising to you on a daily basis just how quickly AI is evolving and how quickly you have to kind of keep up with it? — It is and I'll tell you just both on a personal level as well as professional like I love to just write code. But I'm a, you know, I'm a software engineer by background, computer scientist back in the day. And just seeing how these tools are changing, evolving is, it's amazing. Like just before the holidays, I would use some of these vibe coding platforms. And I would run into a few issues. It was still magical in many ways, but I was still running into issues. And then just uh two weeks ago, I built an app end to end and um I literally had zero issues. The software was able to build it end to end. I wrote the requirements clearly and there were zero issues in the software development. I mean this is mind-blowing like how good it is. Um and so that's kind of on the personal side just kind of how I'm experimenting playing with the technology. The other thing is on the professional side we're drastically changing how we build software oursel at Cisco. I mean Cisco we are leaning in big in terms of um AI development both in terms of capabilities for our customer but also how we build software. Yeah. you know, AI defense and this was shared

### Segment 3 (10:00 - 15:00) [10:00]

at the AI summit like you know we had all these amazing a builders of AI there like Sam Alman from Open AI and Jensen Wong and you know Matt Garmin from the CEO of AWS all these great builders were there and one of the things that G2 shared on stage was that AI defense is one of the first products that is written 100% that's going to be 100% written with codeex which is open AI's um you know codew writing tool Right? — Which is pretty amazing because you think about AI defense. It's an enterprisegrade product. It is a cyber security product and it's written 100% by AI. I mean like if you told someone that 5 years ago, they tell you that's crazy. That's crazy talk. And now it's real. Wow. That's amazing. And of course that also means though that even though it's able to do that, we still have guardrails in the terms of like having people be able to monitor that code making because I think a lot of people think, oh well, if AI is building it, that's it. Aren't you worried about that? But we there's a human element that I think people miss with regard to the AI conversation. Wouldn't you say? — That's right. Okay. That's right. And it's I'm so glad you brought this up because, you know, a lot of times people will just hear me say that and say, "Oh, they just put a prompt in and they made the whole product. " No, that's not the reality. The reality is there was a lot of thought and discipline on how you can make get AI to write such a complex product that has a really complex system architecture and do so in a way that AI could write that code. So we've invested a lot and I mean frankly I think that's tremendous IP that Cisco has in terms of how we do spec driven development, how we pressure test those specs and how we get AI to test itself to deploy itself. All of that is really IP in how you give the input to AI so that AI can be successful, right? Because if you don't have all that hard work upfront and all of that rigor, then AI will not be able to build you a, you know, secure, reliable, performant kind of product. Absolutely. — So that's really the IP upfront is in that process. But the beautiful thing is once you do that, you can literally say, "Okay, AI, here's a spec driven development. Here's all the details. Go. " Yeah, love that. I love that. You know, it just it's core to how we at Cisco overall think about AI, which is really, you know, it is it's a tool that is used in conjunction with human capability and it's really there to help you do your work better as opposed to just doing it by itself. I want to go back to the AI summit, uh, which was greatly successful for us and for our partners and our competitors. And this is where I'm kind of going. It's really important I think that we do work with customers, partners and other companies that will be considered competitors to sort of advance our technologies. What are your thoughts on that in terms of the importance of working with even competitors to really achieve the same common goal? — Yeah, I think like look the ecosystem matters a lot and being a leader in AI having the right mindset and openness and platform orientation is critical. Look, no one's going to solve the AI. There's no one company who's going to solve all of AI all by themselves and especially not when you think about it with the end goal of helping every enterprise, every organization achieve optimal AI, right? So I think it's really important to think about the ecosystem and at Cisco we're doing things that are you know really forward thinking and leaning into the ecosystem and partners and even the even competitors and you know as part of the ecosystem right — so a few examples I'll give you one is of course you mentioned AI summit is bringing together builders the greatest builders of AI and you know that was a phenomenal show the other thing that we're doing is we're leaning in to open standards. And so one of the things we're doing is I talked about it earlier, but you know, we have the AI safety and security framework. — We've built that at Cisco. We have our AI research team that invested heavily in researching and thinking about the holistic set and building the most comprehensive AI safety and security definition. And now we've given it to the, you know, the public. We've made it open source. It's out there. we've let um you know testing organizations like Nest um OASP and others leverage it and we've also built it into our own products but more importantly we've opened it up so that's an example of open standards helping the whole community because we want everyone to have you know safe and secure um AI software and the best way to do that is to have a common nomclature have a deep understanding of the different types of threats whether it's you know multi-turn or you know or single turn type of attacks or prompt injections. Understand that holistically. The third example that I'll give you is that as we're building a lot of our products, including products like AI canvas, AI canvas is really is a magical product that brings together the central

### Segment 4 (15:00 - 20:00) [15:00]

management for all of Agentic Ops in a single place. And it's built with AI. It's built to be multiplayer and it has generative UI in it. So, and you could so you could literally say, "Hey, tell me what's wrong with my network. " and it draws you a network diagram and pinpoints where the problem is and recommends a fix and say do you want me to fix it? That is a product that is so powerful when you bring together not only the Cisco portfolio products but also bringing in the ecosystem of products that customers have in their environment — right — and the third piece is don't forget customers we have some of the smartest customers and best customer in the world they have IP too they have remarkable IP and so also being able to let our customers not only bring in the tools they use but also bring in the knowledge and skills that they have and bring it all into this one beautiful canvas where they can solve the problems that matter to them or ask AI any problem that matters to them about infrastructure data center. So yes, we're leaning hard into the ecosystem and hopefully I gave you three different examples of how we're doing that. — Oh, absolutely. And I think it also just underscores the importance of making sure that the customer is happy because ultimately we want the customer to have not only just great experiences but simplified user experiences. — Yes, exactly. We want them to have simplified experiences. be able to harness the power of AI and do so quickly. And most importantly, we want them to be able to do that in a safe and secure and trustworthy way where they feel comfortable with it and they can make their and customers comfortable with AI. — Absolutely. While we're on the customer topic, what would you say to customers? What advice would you give to those who are kind of in the beginning of their AI journey and are looking to adopt more AI within their infrastructure or just need AI solutions for their out you know business outcomes? — Yes. The first thing I would say is congratulations because that's exactly the most important question you probably need to be asking your company today to have success over the next 5 years, 10 years and beyond. So congrats. You need to be on this journey. Number two, yes, you need to lean in, but let's do it thoughtfully and let's put together help you put together a plan to do it thoughtfully. Um, there's kind of two dimensions to this that I would encourage customers to think about. The first is just understanding that AI is also a journey. AI is iterative. AI is changing and also your understanding of how you adopt and use AI and how you build your own skills is going to grow and increase over time. So, the most important thing is to get started. It's kind of like, you know, you go to gym and uh, you know, people make lots of New Year's resolutions. Hey, I'm going to go to gym. I'm going to, you know, I'm going to be so strong. healthy. But the most important thing is actually first to start. — Just go there. — Just go to the gym. Like drive there, go in the door and do something, right? — And then the first day, first week, you do something and then it starts to become natural. a habit. You start to learn like, hey, I got I can work out every day. I it's part of my habit. And then slowly you start to build that skill and now you say okay going to gym is after week one or two hey I can do more than not just go to gym I can start to get better and more proficient at certain aspects right and now I can do you know I can lift more I can run longer or whatever your criteria is AI is the same way it's like a muscle in the organization you got to first start leaning in and then you got to start understanding what are all of the parameters and there's new parameters coming in new tools new capabil abilities, new skills and you got to learn how to work with those. So that's one dimension is you start and you start iterating and have a growth mindset with it. The other aspect is you understand that there's certain capabilities of AI whether it's you know prompt response or agentic AI or moving towards human in the loop and then eventually moving to you know guided autonomy and full autonomy. So there's kind of a spectrum of even the AI, you know, how much control you give to AI. As you get more comfort, you can move down this continuum. — That's right. — The third thing that I would recommend customers is make sure that you have the right guard rails and validations as you go through this pro process. So in case you make mistakes with AI, they're not detrimental. You have the guardrails to protect you, use a product like AI defense. It will help you. Um and then the last thing that I would say is that understand like what are the customer problems, business problems that you are trying to solve and make sure that whatever you do with your AI experimentation, you lean in deep into solving a product or a problem meaningfully. Yeah. Right. Like some people will say, "Oh, let me automate one step of this process. " Well, automating one step of a process is not that valuable if the next step needs a human to review it. Now, you just create a bigger backlog for a human to review. — So, think of a endto-end process that you could fully automate so that you can start to understand and have a visceral feel for the power of AI when you let it automate something holistically. So

### Segment 5 (20:00 - 25:00) [20:00]

pick a customer problem, automate it end to end. — Yeah. And I think that last point, they're all incredibly important. So, thanks for sharing your insights on that. That last point is something I've heard other leaders talk about which is don't just get AI for the sake of AI because it's something to get. You need to really think about you know the why you need AI is so much more important than oh I just want to have it — right. Exactly. And I think like I see sometimes people you know do these mini experiments. Oh AI is not that useful. I say no. AI is incredibly useful, but the problem is you may not have structured your experiment in a way where you're allowing AI to flex its muscles to do all the things that it can. — That's right. — And so that's why picking the right customer problem and use case is also important, especially when you're getting started. It's one of these things when you see the power and magic of it happening, then you want to lean in to use it. You'll be like, "Oh, now I can think of 10 more places I can use it. " — Oh, yeah. Absolutely. And I think you might be alluding to this, but we see this in just our everyday personal use of AI. I mean, you know, it's whatever you put into it, it's going to give you back more and it's going to kind of grow with you. So, it's a great thing. — Exactly. — Love it. So, clearly, Agentic Ops is redefining enterprise automation and I'm sure there are some incredible use cases that you could give us now. I'd actually love to hear your take on some of the best use cases for it now, but then when you think 10 years into the future, um what possibilities do you see when it comes to AI and Agentic Ops and all of that uh technology? — Amazing. You know, I was just listening recently, you know, at Davos. they had a very similar question like when is AGI coming like you know and um and they had the expert Demis who's the you know the head of uh Google deep mind and the founder of anthropic and they're giving their point of view on like when is AGI coming and what's the future is it 5 years 10 years 3 years — and you know it's magical just to think about how the possibility of how AI could you know have not only regular in human intelligence but even a step beyond it right it's pretty amazing to think about that but you know when I bring it back to it yes then some people say 3 years 5 years some people say 10 years — right — the reality is when I bring it back and I talk to customers I say guess what whether AGI happens or not you have the possibility to transform your business right here right now — with the technology that's there so yes we can speculate about the future and by the way the future is only going to get brighter. The, you know, the famous saying is that the AI that you're using today — is the worst AI you're going to ever use in, you know, in the future. — Oh, wow. Yeah. — So, AI is always getting better. So, that's kind of the power of AI. — At the same time, if you want to get business impact, — the AI tools are there for you to exponentially get better at your business right here, right now. — Right. And some of the type of use cases that I'm talking to customers actively about, you know, span a variety of areas. If you know, Agentic Ops can be relevant to you if you're doing security operations, right? Like AI canvas being integrated into all of the Cisco security products and being able to say, hey, you get an alert that, hey, there's some suspicious activity happening. And AI will automatically do all the research for you to figure out is that suspicious activity worth investigating? What's the root cause of it? Oh, it looks like there I've detected this laptop has, you know, um C2 server activity and communication that's happening and then giving you a recommendation that oh, I've checked with Talos and it is legit that is a real, you know, um a real threat and a known threat and then coming back and saying I recommend you contain that endpoint right away, — right? — And then do you want to continue? You say yes, I need to. So yes, and it'll go and contain it right away for you. Then you it'll say by the way do you want to take a look at like how this happened in the first place so we fix it so this issue never happens again you say yes I wouldn't want to make sure it never happens again then it'll say okay here's a recommendation plan I went I looked at all your firewall logs this — some of your settings are opening your security permissions too wide — I recommend let's tailor it just to the people who need it right — should we do that — yes and I say okay I'll do it and they'll say hey congrats we're Well done. We fixed this issue and we'll make sure this issue never happens again. That's an example of security operation. IT operations amazing automation opportunities. Hey customer, file this ticket automatically respond to that issue, file, you know, file it in the system, respond to the customer, fix the issue, close the ticket. Amazing. Um, we already talked about a lot of the phenomenal use cases in software development. Yeah. Right. Not just building software, but how to test software, how to deploy software. So you know I feel like there are so many use cases. I could go on and on but the reality is that if you go and talk

### Segment 6 (25:00 - 30:00) [25:00]

to a customer and this is why the enthusiasm is so much about AI is you could pick any part of the business um you know any business uh process and there's opportunity to make impact. You can pick any customerf facing experience there's opportunity that AI can transform that. So that's how pervasive and how compelling AI is. And I think to the other aspect of like where do you see AI you know down the road I feel like today we talk a lot about AI and we talk about AI in kind of a singular term you know artificial intelligence it's a you know a singular you know singular thing but actually AI is going to become so pervasive it is going to move from not just being in interactive form in agentic and automation it's going to become in physical AI right every robot every machine every satellite you know data centers in space will have AI. There's going to be this place where AI is going to be proliferated. And even AI, when we talk about agentic AI, if you have aentic AI, like I built an app recently, it has 20 agents. Now, am I going to say that it has 20 AIS? Well, actually technically each one of those agents has a purpose. One's an orchestrator, one's a scouting agent, one like looks at blogs, one you know, correlates, you know, interesting tidbits together. So each of them is like their own AI. So AI will proliferate. It will become very widespread. And I think we're going to talk about AI as just a part of everything. It's just a component of it. — Yeah. — And I think that's the future we're very quickly heading towards. And it's so exciting. And at Cisco, we could not be more thrilled to be part of this journey and not only be part of it, but leading this journey and helping our customers lean in and get the maximum impact out of AI and do it in a safe and reliable and secure way. — Oh yeah. And I love that outlook. You explained it so well. It's almost kind of like the internet. It's like when it first came out, worldwide web. Oh, check the worldwide web. Now we just go online. Like now you nobody really references the worldwide web anymore. Um, but no, that's a great way to look at it both in present day um perspectives as well as looking to the future. So, let me let's step away from work really quickly because I want to close out with uh more of a nonwork uh topic when you're not busy and I know you're really busy and you're kind of like just you have time to yourself. You're you're not at work. You want to find your zen moment. You want to just take a load off. What are some of the activities or hobbies that you find joy in participating in? You know, some of the things that I have a lot of I got two boys. We love sports. We do a lot of sports. And you know, I'll tell you one thing that's very interesting how AI has changed some of my habits. I was getting on the plane and before I left the plane, I you know, I had to say, "Okay, AI, while I'm on this plane, can you please build this build these enhancements to my app? " And so that way by the time I get off the plane, I know that AI's done all my work for me, right? or uh or you know AI, hey, prep. I got this customer meeting. Prep for this meeting so that by the time I land, I know I'm ready to go. And it's kind of funny because a lot of times when I go with my kids, we'll go play something. I'll be like, "Hey guys, give me just two minutes. Let me just tell AI a few things to do so that I know AI is working while I'm playing outside. " And I feel like that's part of how I think AI has become intertwined in my life and my hobbies. It's like it's not something, you know, it's like even when I go to do a hobby, it's like before I leave to go my hobby, I'm like, "AI, I want to make sure you're working while I'm having fun. Make sure you're doing my work. " — I mean, that's like the perfect use case for AI because, you know, you're still using your own like, you know, proprietary thoughts, but someone's doing the heavy lifting and you can enjoy, you know, your family time. — Yeah, exactly. But that's the thing I love the most, family time, spending time with boys. We play a lot of basketball. My boys like to swim, they like soccer. Unfortunately, I'm not very good at soccer, so I feel like I'm also learning soccer as an adult now. Um, but yeah, those are some things I love to do. — I love that. Well, that is, I think, a perfect place to bring the conversation to a close. Ash, thank you so much. This has been such a pleasure. I think it's been really great for anybody out there listening who doesn't really know a whole lot about AI or those who do, just to kind of get a perspective on Cisco's approach to AI, why it's such an exciting time to be part of this revolution. And I just want to thank you for your time. I appreciate it. Thank you so much for having me. I think the more that we get the word out about AI, the more people learn. And I think the thing that I would say is lean in. Get started today. If you're not using some of these AI tools, if you're at Cisco, if you're at one of our customers, start today. Start using tools like Cloud Code, Codeex, use them today. And if you're a customer or if you're in a large organization, think about how you can use AI with trust and safety in mind. And Cisco is here as your partner. We can help you. We have tools like AI defense that can help you on this journey. Get jump started and make sure that you get to a successful outcome very quickly. Love it. Thanks so much. Hey, — thank you. All right. —