Cisco President on Rogue AI Agents, AWS Internal AI Push After Staff Cuts, AI Coding Agent Personas

17:36Stephanie Palazzolo:It's not just me, by the way. The people that I was texting with as well, they also told me to ask you, when is he going to be CEO? So I think people are expecting him -

17:45Jeetu Patel:We'll be able to do time soon because we need Chuck for a little longer, man. He's doing a great job.

17:50Stephanie Palazzolo:Right. Well, Jitu, thank you so much for coming on the show. We really appreciate it. That is Jithu Patel, President and Chief Product Officer at Cisco here on TITV. AWS is not just selling AI agent tools. It is also using them internally. And specifically, it's been using them to help with sales and business development functions as it recently made headcount reductions in those departments. That is according to a new exclusive report from my colleague, Kevin McLaughlin. I want to bring on Kevin to share more about what he has learned. Kevin, welcome back to the show. It's great to have you here.

18:23Thanks for having me.

18:25Stephanie Palazzolo:How is AWS using AI agents in their sales functions? So AWS, like basically every other major software vendor, works with third parties. They're called channel partners, and they range from software providers to consultants to systems integrators. And the role of these partners is to kind of serve as extensions of the sales force of AWS and the other companies. And so they sort of, they get on board with the latest product strategies. They tell customers about this new thing that's coming that you're going to want to buy. And so AWS has a whole corresponding army of employees who work with these channel partners.

19:05And it's in that function that AWS has started to build agents that automate some of the work of these staff.

19:13Stephanie Palazzolo:Channel partners. So these are people that help to sell, help to distribute, help to sort of customize the software so that it works with the end customer? Is that the idea? All of the above, really. They help customers kind of move along the path from beginning to use cloud services and AI to using more sophisticated forms of those products. But really, the main role that these partners play in a lot of cases is just sales. and AWS has a whole, like I said, team of salespeople that work with the partners. Now, one thing that's important to note is sort of confusing, but in a lot of cases, AWS partners are customers.

19:54They are just large customers that have more than just a transactional relationship. They work with AWS sometimes jointly sharing technology and developing things together. So I just wanted to clarify that.

20:08Stephanie Palazzolo:And maybe another point of clarification, I mean, these are like the consulting firms that we think of that work with the big tech companies? Sure, yeah. Some of the most recognizable channel partners are Accenture, Deloitte, Capgemini, companies like that. Okay. So what exactly are the agents doing then with respect to their work with these? I mean, you know, what are the roles or duties, I guess, that they're being tasked with? So we've reported the one piece of unreported information we had in the story was that AWS is working on an agent. And the agent will allow AWS salespeople in cases where customers have deep technical questions to get answers quickly to those questions using the agent.

20:53that you traditionally has been handled that that deep technical research role has been handled by what's known as AWS specialists who are just really they're called domain experts and so they're they go really deep in like one area or a handful of areas like cyber security or networking or things like that and so the upshot of this agent is that AWS says that its specialists will be able to focus on sort of higher level work and strategic work with customers as opposed to having to go and perform the same research to get the same answer like over and over again.

21:30Stephanie Palazzolo:Right. And so, I mean, a bit higher level here. I mean, is it saving them costs? You talked a little bit about layoffs that the company has done or headcount reductions. Remind us again, which exact headcount reductions you're referring to there because it's kind of hard to keep track of them and then you know talk a little bit about how much it's actually saving them in terms of cost if we know anything well the people we spoke to for the story which were current and former employees their impression was that a lot of the automation that's happening uh with aws and ai uh the the in the homegrown stuff that it's using internally um is designed to to take on some of the functions of the roles that were laid off.

22:11AWS was very strongly speaking in the story saying, we're not using AI to replace employees. And certainly, they're not the only companies out there trying to make that clear. But the story was all about sort of, you can automate parts of roles using AI. And the argument that vendors always make is, well, there's always higher level work and they call it higher in the stack. And that's certainly true. I guess one question is, at some point, how much more stack is there to move up to? So that's not something we tackled in the story, but it's definitely a topic that comes up a lot in my conversations in the industry.

22:49Stephanie Palazzolo:And when you went to Amazon and, you know, suggested your, told them about your reporting, what did they say to you about this link between, oh, headcount reduction and agents? I mean, what was their response? Yeah, they said very clearly in the story that we're not using AI to replace jobs. That was really their stance. And I think, like I said, we're not disputing that in any way. But certainly, I think at some level, when AWS can use AI to automate parts of roles, it also showcases AWS's own ability to – because a lot of the stuff that it's developing, it's using its own AI products for. And so I think at some point, this can be a big applied AI story that they can tell, showing investors that it's not just about selling AI to customers.

23:36It's also how you use it as a large company to create more efficiencies. And no one's better at finding efficiencies than Amazon.

23:44Stephanie Palazzolo:Let me ask you one question before you go here. You talk to all these sources in and around the Amazon ecosystem about these agents and about these changes to headcount. On one hand, people at Amazon are designing the agents. They're building the agents, right? This is the company that is getting agents out into the world. And then we have this sort of other parallel reality where it's like, yeah, maybe agents could do my job. I don't know. And so I guess the question I have for you, Kevin, is there fear within these companies, the companies that are literally making the agents, are the employees themselves kind of scared about what it is they're building in a way?

24:28Well, I definitely spoke to some who were, in fact, scared this long term. What are the implications? But I don't think they're alone. I think in any sort of knowledge work, I mean, as we've written about extensively and things like AI coding, I mean, I don't know that anyone can really look at what's happening with AI who has any kind of role in a company and say that their job is completely safe. Now, that said, the hire in the stack argument, I think it does hold water. There's definitely a lot of real busy work, you know, responding to sales leads, looking into sales leads, like another agent that AWS launched last week.

25:10This basically helps its salespeople look at the sales leads that partners get, which in many cases don't have a ton of data associated with them. They're sort of incomplete. And so the AWS salesperson could look at that, spend an hour looking and find nothing that's actually worth chasing. So instead of wasting that hour, the AI can sift through that and tell you, hey, here are the ones we need to prioritize.

25:35Stephanie Palazzolo:Great. Well, Kevin, I want to thank you for coming on. That is Kevin McLaughlin, our enterprise software reporter, here at The Information. Thanks. We are just three months into 2026, and already Tech M &A is not living up to its expectations. The Informations Deal reporter, Valida Powell, went to New Orleans last week to attend an important conference for M &A insiders. She joins me now to talk about what she learned. Valida, welcome to the show. It's great to have you here. Hello. So you went to the Tulane Corporate Law Institute Conference. That's like what? The biggest conference in M &A land?

26:09Stephanie Palazzolo:Is that the idea? It's the annual gathering for dealmakers to talk shop. Okay. And so you're at this gathering. Everyone's talking about M &A. And what's the consensus? Is people are happy or not? So like bankers and lawyers are never not happy about the state of M &A, but then also like it's kind of the attendees are more focusing on like all industries. But then people are saying that M &A is not living up to their expectations like this year. But then there's like other bright spots like other than tech acquisitions. So in our stories like on Sunday that we wrote about of how the M &A volume has gone up like quite significantly in the first quarter, but it's mostly driven by like SpaceX and the acquisition of XAI.

26:54And for other than that, like the number of deals has fallen because like people have these fears around like, you know, how AI is racking their valuations, you know, what should they do to stay ahead of the competition? So there's lots of uncertainties around the world these days, which is not great for deal makings, according to bankers and lawyers.

27:13Stephanie Palazzolo:Right. And I mean, one of the things you mentioned in the story is when you talk about AI, you You talk about the state of the world. You talk about the volatility. These are all fears that are weighing on people. The other thing you mentioned is activist investors was a topic at the conference. What did people have to say about the threat of activist investors getting involved in companies? Before going into activism, there's another area that's stalling dealmaking. On the other hand, for example, Glassstone Place Partner, which is a PR firm, have a survey before the conference that talks about like what deal makers expect to drive deal making in 2026 and one of the areas like private equity but then sponsors like these private equity sponsors have to borrow monies to finance their big acquisitions you know they don't have lots of cash on their balance sheets and then the recent kind of crisis or like kind of right down in the private credit markets you know like like names like blew out like they have been kind of like retail investors have been like redeeming funds and like the loans costs for the software companies have gone up because they don't like investor are not sure about the long-term values of these like software companies like on the ai competitions and this is like an area of like challenges for um deal making too on especially private equity side that's been very active usually in the software world so so this i mean this is people literally saying this is the credit markets are a huge challenge for us right now to get any of the deals done that we want yeah it's like the borrowing cost would be much higher which is not ideal for people right okay uh what about the activist investors this this was one of the most exciting parts of the column i was i was excited to talk to you about it so um lots of like software companies or stock prices drop like significantly for example like in hubspot they have fallen at 50 50 like the last years and activists are like you know um more incentivized to look at all these companies which have like you know stable growth and they have like you know free cash flows to see if they can make a play to turn a company around um so they like bankers and lawyers who advise on like defense saying companies again activists will are saying that there's more companies that activists will look to take states as this company either they want to push for changes or they want to push for a sale of those companies right so so people people are expecting more of these activist campaigns to to start in the coming months is that the idea there's already like a few that's kind of played out like um last earlier yesterday um elliot investment which is the big activist over tech state like the chip design software maker um some nazis and they think they're trailing behind, they should make more money on this AI booms.

30:05So I thought this is an idea that activists will look to, to see like, we know who's trailing behind and what they could, what public campaigns can launch to push for those changes, either so, you know, cost structures like, you know, expanding their sales or in some ways, you know, they could also push those companies for like a sale to, you know, private equity sponsors. For example, like, you know, Starboard took a state in. Clearwater Analytics in last December, and then they pushed the company to sell to private equity sponsors later in the month. So I felt like, you know, this will be a trend that's relevant in the tech space.

30:43Stephanie Palazzolo:Now, last question for you, Valida. I mean, you talked about the data here with M &A, and I encourage people to read the story because the data really does show the extent to which M &A has not been happening this year compared to other years. But are there any bright spots here? Are Are there deals getting done in any pocket of the tech sector right now? So the data center space or the infrastructure space has been really heated. Like we reported last week, a few weeks ago, about N-Scale buying this company called AIP. And then there's another private equity sponsor taking an energy provider, a big deal there.

31:25And I felt like this is an area that people have been expecting that like adjacent to the tech space and how like, you know, they're supporting those AI build outs. And that's a bright spot for deal making. Great.

31:37Stephanie Palazzolo:Well, Valida, I want to thank you for coming on. That is Valida Pau, our deals reporter here at The Information. There is a curious trend taking hold in the world of AI coding agents. developers are starting to instruct coding agents to take on different personas acting like more of a team and less like an individual programmer our ai reporter stephanie palazzolo wrote about that for our ai agenda newsletter i want to bring her on to talk all about it hi stephanie how's it going hey it's going good okay let's talk about coding agents so you you wrote about this idea that ai agents should take on multiple different personas, personalities?

32:18Stephanie Palazzolo:What's going on here? Yeah. So essentially this kind of trend that we wrote about in this morning's newsletter is a new way that developers are using coding agents to get better performance out of them. And so I think maybe this is best explained with an example. And so, you know, in the past, somebody might have given a coding agent a pretty high level task, like, you know, make a new login window for my app or something along those lines. Now, instead of doing that, people are basically spinning up multiple agents to take on different kind of personas that might pop up in every stage of this process.

32:54So first, they might ask one agent to act like a product manager and create a doc that kind of outlines what they want this new feature to do. Then maybe they'll tell another agent to act as a spec agent. So basically turn that product doc into more specific instructions that really lays out kind of what technical work needs to be done. Then you might ask another agent to be like a tasks agent, which turns that into specific coding tasks. Then finally, a coding agent to do those tasks. And then lastly, a review agent to review kind of all of those different coding tasks that the previous agent just did.

33:29So it's kind of all about like breaking up this, you know, larger kind of higher level task into smaller ones, and then assigning different agents each part of that task and telling them, oh, pretend to act like this type of role or person that would be doing this sort of task.

33:47Stephanie Palazzolo:Now, this is kind of interesting because I know that as it relates to chatbots, I mean, we've done a little bit of report. Well, you've done, you did all the reporting. I just talked about it here. But, you know, I recall you having written about chatbots taking different, and this is not different personas, is this is a different personality. So, you know, maybe you have an answer that's a little snarkier, an answer that's a little happier, maybe one that's a little more introspective. I mean, is the idea here not so much that, that it's more like segmented by roles or is it that the coding agents might actually have different ways of giving you the product or maybe it's, I don't know, snarkier or something like that?

34:31Yeah, so I think there's like kind of two different like related concepts here. So I think first there's this idea of AI chatbots or just, you know, AI tools that people talk with that have different personalities. And so I think part of that is kind of, you know, for consumers, for instance, who might have a certain preference for a type of personality they want their AI chatbot to have. So maybe I want, you know, ChatGPT to be more concise or more funny or more kind of straightforward with me. And then that's a little bit different from, you know, what we're talking about here with coding agents, where it's less like personality and it's more like pretend to be a product manager, pretend to be a, you know, senior engineer.

35:17There is like some overlap, though. So we saw, for instance, XAI, whenever they first came out with their Grok 4.2 model, a lot of developers pointed out that whenever you looked behind the scenes, there were kind of these four different agents that were running all at the same time. And so, you know, one of them would kind of be more of like this contrarian agent, which some might say is kind of a personality trait. But basically, its main point was to kind of push back on anything that the other agents said. So, you know, question things that the other agents wanted to do or bring up maybe alternative paths that the other agents had a thought of.

35:53And that kind of, you know, made the end product better. And so I think all of this is like very related and, yeah, definitely very interconnected here.

36:01Stephanie Palazzolo:So what do you think this development and this trend says about the usability of AI overall? Because on one hand, I see the argument that, hey, I need to structure out my thinking. I would need a team with five different people. It's easier for me to think about what each different agent would do. On the other hand, you could make the argument that it's taking more instruction. I'm having to do more of the thinking. I thought the agent was supposed to do all this for me. Yeah, I mean, it's a really tough question. I think there's like two camps, which you kind of laid out there. I do think that, you know, on one hand, it feels a little bit silly to, you know, be in a world where people were talking about super intelligence and AI that can help us colonize Mars and, you know, educate every child on Earth.

36:46And then meanwhile, you're having to tell like an agent like, oh, pretend to be like a paranoid, you know, engineering manager that questions everything that it's being told or something like that just feels a little silly. I think at the same time, though, some people would kind of point out that, you know, if you were working with humans, you'd have to do the same thing, right? Like, my, you know, my editor can't send me one sentence over Slack, like, do this story. Like, I think it, you know, there's a lot of back and forth that happens where I'm talking through, like, a story idea with my editor and breaking things down.

37:19So I think that process is something that you even have to do with humans. but I think a lot of researchers are saying that this is kind of a you know band-aid solution for now but in the future they expect the models to be able to do this sort of thing by itself so instead of the human having to be like okay here's this you know there's this broader task like let me break it down for the model into steps one two three the model itself will be able to do that kind of division of tasks and then assign each of those tasks to the agent without the human having to interfere too much. So researchers are saying that that's in the future, but this is kind of a band-aid to deal with where the tech is right now.

37:58Great.

37:58Stephanie Palazzolo:Well, Stephanie, I want to thank you for coming on. That is Stephanie Palazzolo, our AI reporter, here at The Information. That does it for today's show. A reminder, we are on this stream Monday through Friday at 10 a.m. Pacific, 1 p.m. Eastern. I want to thank you all for tuning in. We really do appreciate your viewership. Make sure to subscribe to the information on YouTube and follow us on X, Instagram, TikTok, and check us out wherever you get your podcasts. I am already excited for our next show tomorrow. Have a great rest of your Tuesday. Bye-bye for now.