What if AI could not only generate, but act and interoperate across your digital stack?
At Cisco Live 2025, hosts Daniel Newman and Patrick Moorhead are joined by Cisco‘s DJ Sampath, SVP, AI Software and Platform Group for a conversation on Cisco’s latest AI announcements and its vision for the future of artificial intelligence, focusing on the transition from generative to agentic AI and the concept of an “Internet of Agents.”
Key takeaways include:
🔹Leading Cisco’s AI Evolution: DJ Sampath illuminates his critical role in steering Cisco’s AI Software and Platform Group, providing a foundational understanding of the company’s extensive AI initiatives.
🔹Pioneering AI Innovations: Discover the most compelling highlights from Cisco’s recent AI announcements, as Sampath shares what personally excites him about the company’s trajectory and forthcoming projects.
🔹The Agentic AI Paradigm Shift: Explore the transformative leap from generative to agentic AI, uncovering its potential to automate entire workflows and reshape organizational dynamics. Sampath provides essential guidance for leaders preparing for these profound changes.
🔹Forging the “Internet of Agents”: Unpack the critical challenges and Cisco’s innovative strategies for achieving seamless interoperability in the AI domain. This includes a deep dive into the “Internet of Agents,” the imperative of silicon diversity, and the strategic importance of open architectures in fueling a new infrastructure arms race.
Learn more at Cisco.com.
Watch the full video at Six Five Media, and be sure to subscribe to our YouTube channel, so you never miss an episode.
Or listen to the audio here:
Disclaimer: Six Five On The Road is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.
Transcript:
Patrick Moorhead: The Six Five is On The Road here at Cisco Live 2025. We are in San Diego near the beach. Dan.
Daniel Newman: Yes, we are.
Patrick Moorhead: But we’re not just doing the beach or sailing. We’re all talking about AI networks. But you know, Daniel, the one thing we really haven’t dived into yet are agents.
Daniel Newman: Well, I mean, we’re sort of seeing these layers of abstractions come and then go. And we have entered this era where every company is sort of building the rails for the future. And here at Cisco Live, it’s been a big focus. I mean, we’ve had this kind of ongoing conversation across the technology space about kind of who are the companies that came out fast, who are the companies that came out later, who are the companies? So Cisco is one of those companies that came out very methodically, took its time, found its place. And I think this year at Cisco Live, with all the announcements it’s made and it’s had some big moments up to Cisco Live, it’s really defining itself, in my opinion. There’s quite a renaissance from Cisco right now and a big opportunity for the company.
Patrick Moorhead: Yeah. One of the most interesting statements I heard was the Internet of agents. And Cisco’s really good at creating these. The Internet of IoT. I mean, a lot of these different statements, but I can’t imagine a better guy to talk about this than the person who actually said it. DJ, welcome to the Six Five.
DJ Sampath: Hey, thank you so much for having me, guys. Super excited to be here.
Daniel Newman: Yeah, you heard me say, oh, just because when we started saying the Internet of anything, we don’t know what comes next. But actually, I think AI’s DJ really enabled that because we had that whole period of time, probably in the 2015-20, where IoT was everything that’s right. And I think what it was missing was the intelligence layer.
DJ Sampath: That’s exactly right.
Daniel Newman: And I think it’s coming back.
DJ Sampath: You are spot on. The interesting thing about agents, by the way, is agents was a bad word in security, you all know, and anybody talked about it now it’s a good word with AI. And so it’s really interesting how that kind of changes now with the Internet of Agents. I think really what we’re talking about over here is we sort of anticipate what’s right around the corner. If you think about what happened with the ChatGPT moment, a lot of people recognize that AI is here. AI is now you can start using it to be able to do really interesting tasks. The most interesting unlock for agents happened when they turned on Deep Research instead of ChatGPT.
Patrick Moorhead: Sure.
DJ Sampath: You could now go out and say, hey, go do this thing for me, figure things out. I got a PhD about 15 years ago and I forget what I did when I got my PhD, but if I had to go do literature survey, it took weeks, months for me to be able to go pull all of the documents related to that, sit down, read it, summarize it, analyze it, put a story together, and then that point of view might not even be used. I might just choose to say, you know what, I’m just dropping this, I’m going to move on to something else. Now. What used to take months can be done in a matter of minutes. I literally say the same thing on ChatGPT with the deep research thing. It goes out, it does a whole bunch of things and comes back. Agents can operate autonomously. They’re powered by an LLM or an SLM. They’ve got the ability to use tools. Like they could go browse and do a bunch of different things. And they have memory. Now, that’s a really important combination. And so as you start to think about agents.
Daniel Newman: It’s interesting. David Solomon from, I think he’s CEO of Goldman Sachs, came out with something that said 95% of the S1 can be done.
DJ Sampath: That’s right.
Daniel Newman: And it was along the lines of it used to take a team of six, something like three or four weeks to do the work, and now 95% is done and the only thing that matters, the entire team, is that last 5%. But let’s take a step back. Let’s go back really quickly. You’ve got this big role, this charter. Now you’ve got AI software platforms, and you’re at a company that’s kind of in a, I would say in a significant sort of redefining of itself. It’s pivoted to ARR, it’s pivoted to software. And so maybe at one time it was a big iron, but you are the center. So talk a little bit about the charter, the role. I know he’s humble, shaking his head, but he knows it’s true. Talk a little bit about the role of the charter and kind of the way you’re thinking about developing this part of the business at Cisco.
DJ Sampath: Right. Just a quick brief history over here. I got into Cisco about two years ago with an acquisition right now as CEO, founder of a company called Armor Blocks. Back in 2017, we started working on trying to take these natural language algorithms and to be able to see what you can do with that in cybersecurity we were partnering with this company, a few folks that had just come out of Google and they were talking about AGI fast forward. That company ended up getting named as OpenAI, as we all know. So we were working with them very early on. And so fast forward to when Cisco came along and said, wait a minute, you folks actually understand how to apply AI. You’ve built a solution, you built a product. It made a ton of sense for us to move into that security business group and sort of build AI technologies for that particular business. And so about four months ago, what happened was there was a realization that, listen, we’re going to start doing a lot more things with AI software and we have to bring the platform together in a meaningful fashion. And as Jeetu took on the role of the Chief Product Officer and he was putting some of these pieces together, he felt like this is something that he really wanted to start thinking about as a top level initiative across the entirety of Cisco.
Now, when you think about AI software, there’s a ton of work that we do on the AI infrastructure side of the house. We’re just talking about a few minutes ago, but how we’re doing, thinking through the networking, the compute, the silicon. But there’s this huge stack that is really, really important from a software perspective to be able to go out and build. I’ll give you a quick example. Securing the use of AI is super important. We launched a product called AI Defense. And since then we’ve been talking to hundreds of customers at this point in time that say, hey, listen, one of the reasons why we’ve not gone further in adoption of AI is because we need to make sure that we’re securing the use of AI better. We have the right kind of controls. And so we called that part of it right because we knew that we had permission to play there. But when you combine them and stack that with the AI assistance that we built as part of this organization, we quickly recognized that you have to start connecting these products together and you cannot build a platform that is not AI native in this day and age. So we set out to build a platform that was AI first, that had agentic frameworks baked into it so that you can go out and leverage these agents to be able to go out and do a whole bunch of different things for you and then wrap that all together in a product that we talked about and launched today, which was the AI Canvas. So the canvas, sort of, if you think about it as bringing all of these pieces together, gives you a place for humans and Agents to collaborate and all together. And that’s something that is tremendously exciting. So this group, the charter right now is to be able to innovate and build durable AI products and build it in such a way that Cisco comes out as AI forward, AI native and fundamentally alters the value equation for our customers.
Patrick Moorhead: Yeah, I didn’t get Canvas until I actually saw the demo and when I saw it in action, I’m a visual learner and I just got it. And I think that’s one of your biggest announcements here. What are some other announcements that you made here? Maybe prioritize them in terms of enterprise value?
DJ Sampath: Yeah, I think the biggest one, if you ask me, is about the devices that we announced for the campus and branch. This is one of the largest refreshes that we’ve put in motion. These are devices that are ready to be able to support the future of workplaces. As we think about it, when you start to think about this movement from chatbots to agentic applications, these agents are going to constantly communicate back and forth to be able to do a lot of different things. If you have WiFI at home, you remember the day, back in the day when you say, hey, listen, I’m going to start with, I just have this single laptop with this computer that I need WI FI for. But if you go in today and look at the number of devices that connect to the WI fi at home, you’re practically like a small business. You’re going to have hundreds of devices connecting to it. If you now extrapolate that into the era of agents inside of the enterprises, you’re going to have these things that are sitting inside of devices, your robots, they’re going to be this huge inflection point of your iPad’s going to have about 30 agents that are going to constantly want to check and update and do a bunch of different things. So we’re entering into that transformation and all of those devices have to be reimagined in terms of what they can do and they need to have safety and security baked into it. You need DPUs inside of these switches, these smart switches, to be able to execute on that vision of bringing security, infusing that into the network. You’re going to need the next level of WiFi capabilities. You’re going to systematically, as you’re thinking through this whole platform, it needs to come together in a single dashboard. That’s how we announced the combination of Meraki and Catalyst starting to come together. And that’s a huge win for a lot of the enterprises. So the themes are the campus and branch is a big set of devices that we’re announcing. The next big one is unification. And if you heard the one more thing that Jeetu announced at the keynote, it was all about something called Cisco Cloud Control, a control plane, a management plane that brings all of Cisco’s assets together. It’s a huge win. And so we’re starting to make progress on that front. And of course the third big one in my mind is a canvas we’re going to in service of that agentic era and bringing the platform together. This is going to be a really big thing. I could keep going on, I could repeat the keynote, but I don’t think I’m going to. I got more for you.
Daniel Newman: I got more for you. We have a big topic here. We’re going to talk about this pivot that’s going on. So heading into the year we surveyed, we talked to a large number of CEOs, billion dollar companies, and above all of which were overwhelmingly planning their year around going all in on AI board level, meaning this was not an IT initiative, this is a board level initiative, kind of like digital transformation was supposed to be, but didn’t quite get there. Of course we had this big liberation day and all of a sudden every company’s new strategy became how to figure out their supply chains and tariffs.
DJ Sampath: That’s right.
Daniel Newman: Although it seems like, you know, and again could be different by the time you’re watching this, but it seems like there’s starting to be a bit of a calming effect in terms of people figuring out that trade deals are getting done. So we’re back to AI again. Good news. Yeah, Generative, you know, but in the last like, let’s say you said the ChatGPT moments in the last two and a half years, three years now almost, we went from kind of like, oh my gosh, we have an LLM that can, you know, basically next era of search to like, we can completely revolutionize your Enterprise, give you 24 by seven autonomous agents that can talk across all of our systems. You know, there’s impacts to work, there’s impacts to workflows, there’s impact to the software industrial complex, there’s impacts to the infrastructure stack. It’s completely new and now looks completely differently. So we’ve gone, let’s just say from generative AI to agentic AI and the automation is going on. I’m really curious, what do you see of all this that’s most transformative? And as Cisco’s out talking to all these Enterprises, the same CEOs, we probably talked to how Are you advising them to get ready for this? Is it augment? Is it replaced? Is it displaced? Does it expand? Does it do more? What is it?
DJ Sampath: I think one of the biggest things is, you know, a lot of folks are waiting for AI to sort of settle down. And I think that is.
Daniel Newman: Is it ever, ever? Never.
DJ Sampath: Amen. Exactly right. I think. I think we are. I think that’s a mistake. If you are waiting for, like, hey, wait a minute, you know, there are so many things happening. Maybe we just wait to see who’s going to be the emergent winner of agentic frameworks, you know, oh, is it going to be Google or is it going to be here or there? Like, you know, so all of those types of questions, I think, made sense about 10 years ago when the cloud stuff was happening. And I think you could choose to sort of, like, wait it out and things were moving at a pace where things would settle in, at least for a brief period of time, and you could make some of those decisions. I don’t believe that it’s going to be true for AI. It’s moving so fast that you’re going to have to jump in and participate and make some opinionated calls to say, listen, we’re going to pick a set of things to go out and do, and we’re going to accomplish this in a record time because things are moving really, really fast. So I think that is a common thread that we’re starting to see and we’re doing it ourselves. If you think about the number of agents we have today, which is the number of agents we’re going to have by the time this calendar year is done, I think it’s going to be an exponential difference in terms of the number of agents we deploy. And I think when we talk to these enterprises, the other top thing that comes to mind is there’s a lot of hesitation in terms of starting to adopt AI, because everybody’s thinking about privacy, everybody’s thinking about, hey, if I put something into the model, how do I know what went in and how do I take it out? The funny thing about the models is that they’re easy to train and easy to teach, but they’re extremely hard to make them forget something that they’ve learned. Right? So you cannot make a model forget things easily. So that makes it really challenging in terms of, hey, do I never train a model of my data? And then the question then becomes, is the model that I’m running in the cloud really my model, or is it the cloud provider’s model? And that becomes a very existential question, because it is fine to put data inside of the cloud, because you’re storing data, which is commodity in some ways. You’re retrieving your data, but your data is not being manipulated in any different way when it’s in the cloud. But with AI, it’s kind of different. You’re getting tokens of intelligence being emitted out of these models, and it suddenly becomes really important to know who truly owns that model.
So the notion of having an infrastructure where you can control what goes into the model, what goes out of it, becomes really important. And when you combine that with safety and security, which are two very different issues, as you know, safety is all about the model. Making sure that the model has the right type of data is not hallucinating, it’s not. Doesn’t have biases. And security is all about attackers attacking the model to make it do something entirely different. Prompt injection attacks, context window overloads. So we’re taking care of all of those things in combination with AI that you own is top of mind for a lot of these enterprises, especially if you’re talking about large financials, large hedge funds. And hedge funds are building their own foundation models. What you saw with DeepSeek, we’re seeing a lot of those patterns and we’re advising them, saying, listen, let’s help you with safety and security first. Let’s give you the right type of infrastructure so you can build your models and you can use it from an inferencing perspective. And this is where we’re partnering with Nvidia really closely. When you think about what we’ve announced with Nvidia, it’s about a secure AI factory. That is something that is really exciting from an enterprise perspective because they want to set. It’s a cool concept, AI factory, To me, when I first heard it was really interesting, I’m like, the way they thought about it, way Jensen articulated it, he said, you’re going to use this infrastructure to be able to create a lot of AI applications, AI agents, and this is basically a factory. Use this to create all of the AI stuff that you want to do. And then when we saw that, we saw an amazing opportunity. We felt like, hang on, we understand security better than most. And if we can use that opportunity to be able to be the de facto secure AI factory player along with Nvidia, we can truly accelerate AI adoption inside of the enterprises.
Patrick Moorhead: Gosh, I know the industry doesn’t need a new acronym or a new term, but it’s almost like a trust Operating system to get everything working.
Daniel Newman: You have one every day.
Patrick Moorhead: Do I? No. Is that my one? It took me this long.That’s good. Hey, we talked a little bit about the Internet of agents and the vision and I think the company you brought out was very provocative and what looks to me like a very valuable type of vision. But this stuff is hard, right? And you layer on top of that the optionality of silicon. I mean, silicon choice is good, but sometimes it can be hard. And then we’re talking about interoperability with open systems here. It seems to get pretty complicated. How are you making what looks really difficult that I know is difficult. How are you making this the easy button for your customers?
Daniel Newman: He just tells it in an agent’s way.
DJ Sampath: Okay, I’m not even here. This is an avatar.
Patrick Moorhead: We’re both waiting for Dan and Pat bot to replace us. It can’t happen too soon.
DJ Sampath: No, but Pat, you’re bringing up a very, very interesting and important point. The easy button. When you think about some of the best products that have been built by a consumer enterprise, pick a product that you like the most. The ease of use of that product is probably what drove you to like the product a lot. The user delight comes from something where you intuitively understand how that product works and you don’t have to sit down and figure things out. I’ll give you an example. When the iPhone first came out, do you remember that slider that you had to slide to unlock the phone? I didn’t have to teach my 3 year old enough to be able to figure that thing out. It would just glow and he would just go in and slide it and unlock a phone, which is kind of very cool. Easy buttons are meant to be that way. And that’s something that we have to absolutely work towards doing. What you’ve seen today is basically a step in that direction. When we talk about the canvas. The canvas, the whole point of it was to be able to give you an easy button in some ways. Because as humans, we just talk and we communicate using language like we’re doing right now. What if we could take that interface and seamlessly translate that into a generative UI that brings the things that you need together that automagically goes out and has agents. And you are not wrong. Agents go out and once in a while.
Patrick Moorhead: Once in a while, he’s a Ph.D. and you’re not. And you got it right. How about that?
Daniel Newman: Hey, you know, there’s plenty of jokes about that. You know,
DJ Sampath: We’ll be here all night, especially about the things that I get wrong. And it’s unbelievable the number of things I get wrong. But every once in a while.
Daniel Newman: We’re analysts just to keep you off course. Like, we only remember the things we got right.
Patrick Moorhead: Exactly. We call them victory, we bury the rest, and we make forecasts so far out nobody will remember keeping us on our toes. That’s the game.
Daniel Newman: All right. Hopefully he remembers the question. Back at it.
DJ Sampath: Yeah. So when you think about what we’re doing with that, we’re making sure that you fundamentally are bringing that easy button to the customer. And it’s not just about the software stack, it’s about the whole thing. It’s about thinking through. Silicon diversity is such an important topic because one of the things that you can potentially find yourself in is in a vendor lock situation, because if every single one of the bill of materials that you have depend on the same supply chain, it’s pretty likely that you’re going to find yourself in a very interesting situation when it comes to pricing, when it comes to availability, and there’s going to be a single person that determines exactly what that’s going to be. So Silicon diversity is absolutely paramount. Yes, it is more complex, but it’s our job to reduce that complexity and give you that easy button with things like the canvas to be able to simplify the whole operations of that complicated stack.
Patrick Moorhead: I mean, you’ve done it before. I mean, I think Meraki was an easy button for the Edge. I think UCS and Intersight was an easy button for Compute.
DJ Sampath: Duo was an easy button for MFA. Seamlessly would just pop up.
Patrick Moorhead: Yeah. So that’s good because a lot of companies work very hard and they take pride. I mean, Cisco used to be the company that was very proud of CLI. And you have changed since. I’ve been doing this job for 14 years.
Daniel Newman: Years, Pat, I’ll give you a little compliment here. But the era where being an analyst that actually understands all the layers of the stack has become paramount in the AI era, like there was a time when you knew chips, you knew networks, you knew storage compute, everybody knew one thing. In this era, you’re almost, you know, no offense out there, you know, go home. But you’re almost, almost not useful. I won’t say useless because that hurts feelings, but it’s almost not useful if you can’t, because you’re really. What you’re talking about is the stack. And in this era, that stack is really the ability to build it correctly, flexibly open where it needs to be fast, and of course benefiting from the language that eventually will be self-creating. Right. You know, synthetic and grows. It’s super exciting. DJ we could talk to you all day.
DJ Sampath: Yes, likewise.
Daniel Newman: I think there’s some other things going on here. I hope you’ve had a great Cisco Live. And let’s do this again soon.
DJ Sampath: You bet. Thanks so much for having me. It was an absolute pleasure.
Daniel Newman: Yeah.
DJ Sampath: All right.
Daniel Newman: And thank you, everybody, for joining us here. Check out all of our Cisco Live coverage. The Six Five is On The Road. We are here in San Diego, California for Cisco Live 2025. Lots more coming your way. Stick with us, but we got to take a little break. See you later.
Author Information
Daniel is the CEO of The Futurum Group. Living his life at the intersection of people and technology, Daniel works with the world’s largest technology brands exploring Digital Transformation and how it is influencing the enterprise.
From the leading edge of AI to global technology policy, Daniel makes the connections between business, people and tech that are required for companies to benefit most from their technology investments. Daniel is a top 5 globally ranked industry analyst and his ideas are regularly cited or shared in television appearances by CNBC, Bloomberg, Wall Street Journal and hundreds of other sites around the world.
A 7x Best-Selling Author including his most recent book “Human/Machine.” Daniel is also a Forbes and MarketWatch (Dow Jones) contributor.
An MBA and Former Graduate Adjunct Faculty, Daniel is an Austin Texas transplant after 40 years in Chicago. His speaking takes him around the world each year as he shares his vision of the role technology will play in our future.
