The Meme King. Memes Covering AI, Cybersecurity, Acquisitions & everything in between w/ Pramod Gosavi

Pramod, it is so nice to have you on nice on Noise to Signal. The only way to the only way to describe you is as the meme king. And I say that because you've been an investor, you've been a venture capitalist, you have been in boardrooms, corporate strategy, and so on. But one of the things that I really like about you is that you're able to distill complex thoughts in like silly memes that hit a nerve, right? And

I can't vouch for your investments, but your memes are top-notch. There is not there is nothing like it. I sometimes fall off the chair when I see some of these memes because these memes really hit you hard. So typically, I you know, typically the way I do these episodes, I talk, I talk up, I talk to the person, I you know get a sense of what they why they got into this space and so on and so forth. But I want to do something different in this episode. What I really want to do is.

I want to go through your top 10. I want to go through your top 10 memes of all time. And as we go through the memes, narrate that in a storytelling way so that we understand the context around why you did the meme that you did, right? And then we go from there. And the the f so the first one that comes to mind, the first one that comes to mind is the one with Nikesh Aurora, where fake Nikesh Arora is dancing to the tune of of a of a of a Bollywood song.

And it went like crazy viral. When I saw it, I fell off. I for for the first five seconds I thought it was Nikesh Arora. I was dancing and I looked closely. It really looked like Nikesh Aurora. And for the audience who are not familiar with the song, I mean most people who are who are born and raised in India, they'll get this song immediately. But for the Western audience, the scene I'm going to describe is there is a song, there is a song where this is this is a song from movie Kalia.

Amita Bachin is one of the superstars from Bollywood. He's dancing to a song and he's essentially taunting a rival thief. Both of them are trying to steal a diamond necklace from a pretty g f from a pretty lady. And Amita Bachin, who's the star of the movie, is essentially telling the rival thief, Hey, this is not your cup of tea. This is not your cup of tea. And I'll be taking away that I I will be taking away the diamonds. And the context for that, in the case of

Nikesh is Nikesh is essentially telling all these founders in AI that you know AI is not your cup of tea. This is you know this is within this is something I do really well, and you should just give it to me. So that's the context for it. So I'm gonna play I'm gonna play the the meme and then let's get into Pramod for his take on why he did the meme when he did the meme.

Yeah, sure. So

so so so Pramod, why don't you give us why don't you give us your backstory? What was happening? Why were you saying that? Were you going through going were you seeing a lot of acquisitions that were happening at that time through Palo Alto?

Yeah, so first of all I think it's a pleasure to be on your podcast. always been wanting to speak with you and and this is like a really fun top topic here. so yeah, so I have I it's not on my LinkedIn, but I worked for Palauto Networks for about six months during COVID, when I left VMware and I had a pleasure of working with Nikesh on on on on a and A strategy deck. And one other thing that

One of the reasons I joined Palo Alto Networks was basically I was following this company in cybersecurity for like two to three years. And you know, I would choose like my best company, my like a wish list among every category, like in the C SPM, like CWPP, cloud cloud identity. And Nikesh would just go buy that company. And and and literally that's what happened. It's like, you this this guy is just buying every single company that I want to buy.

you know, something is is going on there. And that's how this conversation started. And then I ended up at at Paulo Network. So so this was like last year, beginning of the year, I came across this guy. He's an amazing dancer. He's a very humble guy also, this fake Nikesharona. His name is Marunarona actually. So it's the same last name also. And

So hold on, hold on. You man did you manufacture this meme? Like did you have somebody dab

No, no, no. This guy there no that there's a guy actually I actually gave him credit for that, saying that hey this this became very viral thanks to you. So this is a guy called Varu Varunarura in India. He he he does his dances, he puts a lot of effort into it. So the funny thing is I was just like one day just scrolling through his dances and I must have scrolled like thirty, forty, fifty of them.

I see.

Yeah, see.

But this one really caught my eye, just the way the guy was like dancing and and and lyrics of the song. So the background of this meme is you know, n the way Nikesh operates is like a lot of founders don't understand it. Like a lot of founders think that you they are ahead of the curve, they're talking to customers and they they are building it. You know, all other publicly traded companies, you know, these guys, the CEOs, they wake up like a year after.

Yeah.

Yeah.

something becomes obvious. But Nikesh is not like that. Like he is shoulder to shoulder, neck to neck, looking at these spaces, you know, are the customers buying it. So that's the whole background of this song is like, hey, you know, wherever you're watching, I am also watching it. Right. And he keeps saying that, hey, look, you know, you better sell your company to me because, you know, if you don't sell the company to me, I'm going to have it one way or the other because I am I'm

right there and and I'm best positioned to actually win that market. So the funny thing is like once I post like I I actually had a lot of doubt. Should I post this, you know, this is a CEO of a hundred billion dollar company. Yeah.

It was a very gutsy move. It's a very it's a very gutsy move on your behalf because I didn't you knew Snykesh. It looks like you didn't. You just thought about you know, you saw the song, you posted it out there and it went out. Like if you had a

Yeah, yeah. So it's very gutsy. I tagged him and I thought, you know, I'm going to get cancelled and like whacked and everything. But surprisingly, you know, he reposted it and a lot of founders actually who sold the company, they they came and validated it saying, Yeah, this is exactly how Nikesh operates. He'll come to this acquisition meeting. He's like, I know exactly what you're doing. I know why you are selling. Either you sell it to me right now or I'm going to build it or I'm going to buy other company. And and and and and one of the

W one other thing that like a side effect of this was, you know, like Nikesh the way he operates is like he is like if you talk to his people and and in the company, he's very he's very tough, like he's a very tough leader. He expects a lot from you and that's his aura that's his aura that you're always on like he'll always like challenge you and everything, you know. But but with this song what happened is a lot of people first of all thought it was actually him. So so

Yeah.

So so there's a bunch of people actually thought he was actually him and it actually brought up some vulnerability from about him, like you know, like like I feel like it helped him a lot because he himself couldn't post something like this because his RI is so much. When I posted when I posted it, he reposted it to show his his other side. Other side. And I think there is this other side of him actually that people don't see.

Yeah, yeah, yeah, yeah. We get it. Yeah.

I guess you humanize him or even

I get it. Yeah, you kind of humanize him. You kind of humanize him in the sense he's a big CEO, he's a big time CEO, but then there is a funny side to him too, and he'll take a good joke if it's in, you know, if it's with the right intent. I don't think you mu you intended any maligness, you just thought it was funny. And I'll admit, when I the f like I said, first five seconds, I thought it was Nikesh because it looks so like Nikesh, right? Like, you know, so if Nikesh would have done the song, he would have easily pulled it off. So it was pretty amazing.

Because as a late as a late Yeah.

Cool. so the next one, the next one I have, I think which is also one of your viral which is also one of your viral memes. And this was the time around when I think there there were these hundred million dollar AI talent wars that were that were going on. You know, Facebook was going in and it was acquiring, it was trying to get the people from I think what is it, the the Think Labs or Think Labs.

Thinking machines and he was trying to get them for a a hundred million or billion dollars and whatnot. And then you made this funny tweet that hey you were in a meeting with somebody and you mentioned AI and then you have Mark Zuckerberg watching in on you. So curious to get your thoughts on that too.

Yeah, so yeah, so the the original like the background of that meme was really, you know, when Facebook was getting a slam for like privacy and there's this photo of him like kind of peeking through the window and everybody's like, you know, like like like Mark is always peeking and and and and like violating for the po privacy, right? So so when this thing was happening where he were like he he was actually trying to acquire those companies and those companies were not getting sold. So instead he said, Okay, instead of

If I cannot acquire the company, I'll go hire the people. And then he was giving out this like really outrageous offer. Like he supposedly hired somebody with a like one point five billion dollars over six months pay package. It was just ridiculous. So then I I posted this funny thing that he you know, he's just like so desperate getting people. It's like he's like literally peeking into people's windows and any time he hears L L N. So it was kind of

Okay. Yeah.

combining like what he's doing on Facebook versus what he's doing doing with in the AI side. So it was like a very good bridge saying that he's like on he's like actively trying to like hire people.

Curious what your thoughts are are, you know, there was a lot, a lot of debate around so this was this is unrelated to this topic, but I'm I'm curious what you think about this. There was a lot of debate around H1B and you know the pricing for H1B and the fees for H1B. And my my take on that was the if the you know given the given the advantage of having top-tier talent on your team, you would be willing to make these kinds of

Yeah, or get people on your team for any kind of money that is out there. And my take on that was this is essentially going to benefit the n AI native companies or the US-based companies at the expense of the world, because all the top-tier talent would be willing to come and work for these frontier models and companies. and the the companies would benefit from it, even though H1B was looking maligned in that context. Do you have any any any c thoughts on that or

Is that irrelevant for this topic?

It's it's relevant actually. Like it actually we see that in like that's one of the thing you know, I I've been working with founders is like in the age of AI, I think, you know, having that that that tende that 10x talent actually makes a lot of difference. So so somebody said this really well during RSA and I think it was this Kevin Mandia. I he was saying that

Top tier talent.

In the a age of AI, you know, you you bring in a tech ten X AI person, it it amplifies everything. You bring a mediocre person, it amplifies that mediocrity also, right? So you bring in a really smart product person, engineer person who who who just like uses a lot of AI and everything finally will be fine. You bring in somebody who is not AI forward and that's a kind of culture and that's a kind of efficiency you're going to get.

And you know, the thing is if you're if you're genuinely in the past in the pre AI world, if you were a 10X engineer, the AI tools could make you actually a hundred X engineer because your productivity with the agents and all the tooling that is available can actually can actually be really you know a winning strategy for you. So that that that memes that meme makes a lot of sense. The third one I want to talk about is this AI, the valuation of an AI startup versus a non

AI startup, like you know, if you're if you're an if you have AI in your name or we even your domain name ends with AI, you're well immediately valued at 2030 X ARR. But if you if you're not an AI AI company for like let's say an agriculture company or some other random company, the valuation is like, you know, in the rocks. Like and the the meme that you have is where you have one with the AI founder with a diamond and the the the non-AI founder is with like aluminum sheet. Looks that looks like a diamond.

So I'd love to get your thoughts on that too.

Yeah yeah. So this is again like a bit of a pop culture. So you know, I I I I I follow a bit of soccer and, you know Ronaldo, this is like Ronaldo's fiance. She she was showing on on on on on social media like this this like a big like a diamond ring that she got for a engagement, right? So and then bunch of other women was showing that hey, I got this thing which looks like an aluminium thing, but it's it's a pretty brick rock, but it's aluminium.

thing, right? So it is kind of funny that he the people are saying he also got it. So I I wanted to show like the discrepancy between AI and non-AI startup. And it's really b big actually. I mean you mentioned like agriculture and all of that. It's not even agriculture as like even like something that like a AI adjacent like SASI. And so we have a we have a bunch of them in our portfolio and a a lot of them are like actually well positioned for AI, but but they're still not getting the same like

I see.

Yeah, love and it's mainly I think what's happening is like the the pure AI, which you know, a lot of them right now just reselling tokens, right? And they're growing very fast, like three X, four X, ten X, twenty X, right? Like just reselling tokens for a loss and everybody is is i thinks that that should be the growth rate, right? And

Yeah, dive dive deeper into that. It means reselling tokens at a loss. What do you mean by that?

Yeah, so if you look at like some of the fastest growing company like Cursor, right? Like it's it's like the the gross gross margin is like minus twenty three percent. And you know, like if you look at every like most of the high growers right now are really reselling tokens like like even like if you look at Cloud Code, that also is like selling token at a loss because they if you take a two hundred dollars per seat, they they give you up to two thousand dollars of of credits for that, right? So they themselves are selling it for the loss, right? So so

So what's happening right now is like all of these LLM labs, I mean they they can do it because for them they have to really amortize this so they can really sell it at a at a at a cost, right? But but these other independent companies, you know, they're just wrapping it around. They have this very lightweight harness and and and some of them are like going with like forward deployed and everything. so I mean, I think I have a different thesis about it. I think, you know, if

If you're doing that and if you go if you get to a scale like a cursor, I think you can see like say it you're building a lot of b value and brand. But if you if you can't get to that scale, I think it's it's going to be very difficult to sell. I think for what I'm what I'm saying is like there's going to be only two or three foundation lamps and there's only two or three slots for them to buy, and and a lot of them will not buy, they will want to build it themselves, then where does this exit, right? So you look at cursor, right? Like

OpenAI doesn't want to buy it, you know, Anthropy doesn't want to buy it. So now they have partnered with XAI, which is like a third, third tier firm, right? But still, since they have that much traction, you know, there there's still some value out there in terms of the customer base and the brand.

And then what do you think happens to the non AI startups, like you know, the sassys? D I mean what's their path to what's their path to, you know, getting the valuation or do you think there's no path for them?

I think they do have to figure out a path to AI. If they can't figure out a path to AI thing, I I I think it's it's not going to be interesting at all. Because I think you know, right now they're just like a like a wave right now when everybody just wants like pure AI assets, like but even if I f look fast forward like two to three years from now, you people will want some kind of an AI in that. So having just like a like a like a SaaS thing, like a very deterministic logic is not going to

Interesting. And then when you say P pure AI, you only count the frontier models as pure AI. You don't count the mo you don't count the the companies that are using these frontier models to build the product as pre pure AI from your point of view?

No, no, that also no, that is also pure AI, but pure AI is like you know, the company is like I would say like Sierra, right? So Sierra is a customer success company, it's it's a pure AI. Glean on the other hand is also not a pure AI. It's it's basically it's more of a search company and they use AI. I mean like they they existed even before LLN. So there's like so think of company that does a

Mm.

cannot exist without LLMs. Like cursor will not exist without LLMs, CR would not exist. Harvey couldn't exist without LLMs, right? So so to me the pure AI companies is like you you just cannot exist without LLMs.

And you do you think like these companies then the eventual exit is to sell to these from foundation labs, the top three foundation labs and they don't have a path to basically become an independent or stay an independent company for a long period of time? If given the given the advances given the advances in some of the like, you know, we've seen this discussion around cybersecurity and AI models. Like, you know, all cybersecurity is dead because these guys are going to come in and take over all the cybersecurity products and so on, even CrowdStrike is dead and whatnot.

so do you think that that's where it is going, where that's where it's headed, where these frontier models get so good at such an exponential rate, these companies if they don't sell into these companies, they're essentially toast.

No, they can build it. I think the like you know, cursor, they w there was a talk about cursor building their own model, right? So the the the the I I think the the thing that so even cursor, I I I I saw a few blogs and even it was one of the A sixteen Z blog, right? So they themselves say that the the the wrapper that they have is actually very thin. So they are just doing a bit of

Mm-hmm.

a bit of like harness and doing it. So in that case, they they don't have much advantage. They're also not taking or they're not they're not like retraining or anything like that. They're just they were just like for the longest time they were just they were just deploying the model. They they were doing some last mile optimization with this harness thing, right? So so what that means is basically there's like no IP in in in their solution. The IP really sits with the model. So so for them to

For them to build a model, they have to basically go and redo what Anthropic has done. I think the challenge with them is like I think they they they made some mistake actually, like when they were going up in order to grow very fast, they were feeding the best model of Cloud to the customer. So the customers have gotten used to that premium quality now. So now if you go and s and and and feed them like a mediocre model, and also it depends on like the culture of the company, right? The culture of the company is very much

is you know very young guys in their twenties just hacking and they were doing something else and they they they landed on this. These are not researchers, right? So I mean it's fine they can obviously go hire a bunch of people to do research, but I think for them to to have a model that is as good as Claude it will take a year. So now they have to figure out like what do I do for a year? Do I keep continuing selling Claude and do do I bring that model in later on? so that's the the challenge that that that they have.

You know, that reminds me of a funny comment from Jensen Huang who said we you know there is a possibility that most enterprise companies become resellers of Claude tokens. And I f I thought it was very funny because you know it is a real it is a real possibility where all these SaaS companies, enterprise companies essentially become res resellers for clot.

Yeah, reseller and also I think like more servicey, like like if you look at like you're in the vulnerability management space, but like in a pen testing space, like with Mythos and all coming in, I feel like there's no like a software pen testing play at all. Like you have to pretty much serve it as a service.

Yeah.

Yeah. So one thing that comes through in your memes is you watch a lot of Hindi movies, a lot of Bollywood movies, and you know that comes through your memes. the one that comes to mind is the Cyber Arc meme, where there is there is a merchant who comes to Amitabh Bachan and he says you should you sold it for two less. And Amitabh says to the something to the effect, if even if it would have doubled the price, I would have still bought it because my mom worked at that place.

Okay. And I think you you quoted this, you you said this in the context of the cyber arc acquisition from Palo Alto. so would love to get your thoughts on that too.

Yeah, yeah. Yeah, yeah. Just on the movie side, you know, l lot of times on a on a Sunday evening, night, you know, you just want to kind of relax and I like I watch a lot of this like old movies that, you know, you know, it's like a no brainer. You just like, you know, s watch them and it's good entertainment, right? So one day I was watching this movie, Threshul, right? So this thing came about and you know, the the the I mean that's a like an amazing scene and a amazing dialogue the way it is done, right? It's like one of the

One of the good iconic iconic ones at all time, right? And then I was actually also I've been working on this on this book on the side actually. Like so it's it's about like how to build platforms and everything. And then as I was like I've been thinking about like why would Paula go and buy CyberArc, right? So so one of one of the reasons is like, you know, they want to get into that identity s identity vertical.

Iconic. It's one of the iconic ones, yeah.

You know, they have network endpoint, they have SOC and then identity something that is missing. And then the whole thing was like, Hey, should they buy like a pure play player, like a smaller player? They should they buy a governance player, like a sale point, should they buy like a authentication, like a like a ping identity, or should they buy s something else? But then or like a PAM player, also like a smaller p PAM player, like a like a strong DM and and the more I thought about it is like, look, you know the

It's already consolidating, like this this thing is becoming a platform. So for them to to basically to become like a big player in identity, they need the they need the whole platform. They need like all of the legs because if they like Okta itself is not able to compete with Microsoft on that authentication, right? But CyberArc is able to compete on a PAM side, right? So they needed like a strong platform play. And if you look at it like that was the only asset that was available. So that's why

That's why I posted that meme is like I think CyberArc actually really sold it very cheaply because it's also coming from like like you know, like the communication wants to dominate the world, like so the so he wants to go and say like, Hey, I'm a one stop shop for all of your needs. and for the longest time they didn't have identity, so they couldn't sell identity and with with with agents and everything, identity is becoming important. So so it's like a very near and dear to him. Like so for him, you know, his aim is to

So he I felt like you know he he would be ready to pay like any amount of premium. And and that's why I posted that and and that also like I think people yeah.

went viral. And I'm curious what happens to the identity space now that CyberArc is part of Para Alto. In my sense it severely disadvantages the other players in the space. What do you think is the what is what do you think is the outcome of that deal with CyberArc?

Yeah, I think on the at least on the on the on the Pam side, it really disadvantaged like smaller players, like you saw strong DM selling itself, right? So so Pam used to be like a best of breed play where people would, you know, they that's like security thing and people would go buy a good asset, right? So and it was also very fragmented space. But now with Apollo coming in, I mean th like

Anybody who needs a needs a PAM is going to sell them a PAM, right? So for smaller players it's it's going to be very, very difficult in that. it's T B D on the governance side and on the authentication side. So on the governance side, you know, they CyberArk bought a smaller player like a few years ago. They themselves were building a platform, so they've also bought like a smaller authentication player. It's still T V D to me.

if they can actually win against Okta and Microsoft in the authentication game, can they win against Sailpoint and on the on the governance side? I I I have a feeling that, you know, I think they they'll be happy with what they get on the PAM side. They will be happy to maybe they'll acquire a company in governance and they might be happy to just coexist with Okta and Microsoft on the authentication side. I don't think they will want to they will want to

I mean they just cannot actually with that asset they just cannot win that market.

Yeah, one of the th one of the geniuses of of Palo Alto or Nikesh is Nikesh has been saying this story even when he started at Palo Alto that he saw the entire cyber security space as a fragmented set of tools and he just went on on a buying binge, especially even in the cloud era. He just went and bought a bunch of cloud tools and made it the Prisma cloud. And he's just it's a rinse and repeat strategy from his point of view, at least the way I see it.

comes in, he looks at the industry, it's completely fragmented, buys the top layer, goes to the next industry, and he keeps repeating that same strategy over and over again. And it's really it's it is very hard to integrate all these tools, but he is Palo Alto has been able to do that work really well in terms of buying companies, integrating that into their platform and then selling it and become that becoming that one stop shop. So it's pretty amazing that I I I think I agree with your comment that you know he would have paid any price to buy CyberH. It's just that no one was no one valued CyberH enough.

when they were trying to sell. No one saw the big picture, even though it was like right in front of you.

Yeah.

Yeah, yeah. And I think, you know, like Okta should have bought them long time ago and it ended up becoming a problem for them later on because they they were they started competing with them. Like so I think when they were like really, really small, it was not a very sexy asset, not a very like forward looking acet. And Okta thinks, you they can build this internally themselves, but I think they should have st like just suck it up and

Yeah, I agree.

Because now I think, you know, they were they were creating problems with them like even standalone and now with with Palotto it's just going to become a bigger thumb.

I get it. The the other the next one I want to talk about is this concept called forward deployed. I think it's like the greatest title hack of the year. You add forward deployed to your title and immediately you become like this AI engineer, AI god who can transform things immediately. And one of your what one of your memes was, you know, you have all these founders with all these tools to to do something better and then you have the guy who won the gold medal for shooting.

He just has like he just looks at a bare eye and he just goes and shoots and and shoots. So and you said it is it is based on real life events. Yeah

Yeah, yeah. Yeah, yeah. So, you know, I mean I post a lot about security on on on LinkedIn, but I I I spend a lot of my time on the AI ops. Last two years I've been looking at the intersection of AI and traditional industries. Like one of the investments I made made like in fintech, other one was legal tech, and last last yeah a year or so we've been looking at like logistics, supply chain, procurement, manufacturing, software development and everything. so

This kind of came out of again like a real experience. So we were looking at this company like in in fintech or logistics and you know, a lot of this investors are like clamoring for like the best thing that they want is like, hey, somebody from Facebook, Google, come in and say, Yeah, I want to disrupt this whole thing and I you know, these guys are in the stone age, I want to bring them to like the you know, Facebook age and they they come in and they say, I like they they build pretty good software.

they might hire somebody or they might have an advisor like like the FinTech was a good example. They they might have somebody from audit and they're like, I'm talking to him. Same thing on healthcare, I'm talking to him, but it's still like not the same thing. Like these guys are just asking them what to build, but both in terms of the product and go to market, like they they just like no synergy with with with with with the with the thing, right? And and then they're also they're trying to like use pure software. They're like, look, I'm going just like

like create this product that just works like charm out of the box and you don't have to care about it. But with the problem is like a lot of these sectors, they they have so much complication. They have like so many corner cases. And that's where I put this meme is like you know you can have this forward deployed person, like you you created a software which is like 80% there and you you you say, okay, take this software, you know, you you can optimize the last mile, you can run through some of the loops, you can like drag it multiple levels and just get it work, right? And and and sometimes what happens is there's this one use case

you just optimize it, you know, sometimes just particularly for that case, like you know, you were seeing like some healthcare guys who were like hard coding like some of this stuff, like hey, like you know, this thing comes out again and again. I just don't want to spend tokens on it. So so there's some deterministic logic also there. And then you know, it gets done. So so that that's the whole motivation of the meme is like basically like I I truly feel like, you know, AI is like most a l a lot of the benefit is going to be the last smile optimization. And

And for that, you know, having this forward deployed, like customization is the is right way to do it.

And I I agree with your point where you know if you see the same pattern over and again over and again, you don't have to spend a lot of tokens. You can get to a deterministic model because it's the same thing happening over and over over again. And for the corner cages, you can always switch to an LLM and then go from there. So pretty pretty on point. The other thing that you rile against is inflated LinkedIn profiles, you know. So that's something.

that r seems to rile you up really bad. What is going on with the LinkedIn profile memes? And that you know it goes back to the same, you know, like picking up like I one of the iconic characters from the Bollywood, Gabber Singh and you know so and then riling from there. So love to get your thoughts on that too.

Yeah. So yeah, so so I

Yeah, yeah, so that's like a very iconic like I mean I didn't create this that me but I got it from some place where I thought it was relevant because you know, you see these people on l on LinkedIn, they are just basically, you know, like the titles like, you know, like ten ten ten ten sentences long and they like, I can do MA, I'm a change catalyst, I'm this, I'm an advisor, I'm I'm like a global leader and so many different things and like you know, literally you have to go through the

Global leader, global leader.

like profile and see what they're doing. And to me I think it's just like a really they're just doing like a bad service for them. So like right now, like you know, like I've hit like thirty thousand like followers, like connections every time I have to like figure out like should I connect this person or not. And and to me right now, like the the biggest red flag is when they have too much stuff going on. Like like to me I think you just pick what you are working on right now. Like yes, you might have done ten different things in the past, like you are at SAP

specialist like ten years ago, you you work in you know product like twenty years ago, but you're hardly coding anything right now. So so to me I think it's like like I think that that that Gubber thing thing is like really appropriate. Like hey, what what am I a decoy, right? I I kill people. That's it. You know there's like nothing nothing like as clear and concise as that. So literally I think a lot of times I

meet these people and I literally ask them like, okay, what is it that you do on a day to day basis? Like like to forget these titles, forget this, like what is it you do on a day to day basis? And I I feel like people should just like put it out there like what they're doing.

Okay. And the next one is very near and dear to my heart. this is the this was the this was the mean, and I think this is something that you created. I don't think this existed before. This is genuinely something that you made out of thin air. There was a bridge that was built in Mumbai somewhere, and it essentially had a single lane. There are two lanes merging into a single lane, and it was the built- the bridge was built by obviously not engineers. It it is clear that the bridge was not built by engineers.

And you came out, you put a meme out there, and the the meme was essentially, you know, the AI companies and the legacy companies fighting for the same budget. And it is so appropriate and so apt because regardless of what you build in AI, the budgets still are the same. I haven't seen the budgets increasing exponentially to I mean, for some companies they have increased exponentially, but not for every organization. So every even if you're AI native startup versus a legacy tool, you're still fighting for the same IT budget, security budget.

And the meme was like a real classic. So, what was going on when you read when you did that meme?

Yeah, so s so internally, you know, we this this whole SAS calipse, right? Like people are thinking like the SaaS companies are going to die. And and I saw this really interesting presentation and it kind of like resonated with me quite well because I think if you look at like what happened in the in the cloud era, like I I used to work at ca at VMware, so so the on prem thing did not die at all. Like, you know, it's still like a cache machine and and and Rotcom bought it. It's not growing, it's it's kind of

Declining a bit, but the the decline is not more than like a few percent, five percent every year. It's still it's still kind of like it's going to go very, very slow down, right? And then most of the growth what happened on the on the workload side, everybody went to the cloud, right? So all of the spending went to the public cloud, like Amazon Azure. And and this slide was kind of said the same thing is like the SaaS thing is going to plateau out and a lot of the spending is going to be on the AI side. So now if you look at the AI budget, right? So a lot of this AI startups are saying that hey, everybody's going to spend with me.

And a lot of these SaaS companies are also going and saying that, hey, I'm I'm coming up with an AI co-pilot tool, they're going to spend on me. And it's the same budget, right? And that's why that that bridge says is like, hey, like both these parties are building AI apps, and both of them are going to go after the same budget, and you're going to see this traffic jam where they are competing with each other and there'll be some consolidation, some some companies going to die, right? and that's the reality right now, is like

And that's one other thing I think the one of the nuance there was really, you know, you need to be like in a really s strong competitive situation to get your budget. Like if you're doing something like a copilot on top of a SaaS or something, you're not going to survive.

W do you you know, especially for cybersecurity companies, you know, th the ones that are in the SaaS SaaS side of things, do you think a new generation of software come through that is AI native? Or do you think the existing legacy companies will figure out a way to make to survive into the next wave?

some of them will will do. I I think you you have to look at from like a unit of work, right? Like hey, I'm doing some some work, right? So if I'm doing some work primarily on top of a a SaaS, right? Like like a like a sales force, like sales is a good example. Like if I'm just like doing sales operations or revenue operations, all I'm doing is I'm just working on a on Salesforce, right? So so that is something they can just replace it by bringing in AI and

someone else cannot win it, right? So anytime I think it it it is there's some workflow knowledge that that comes up on the other side. and I I can give and then also when it traverses other other systems like like like even like system of engagement right now like like emails and like Slack and other things like anytime you're you're doing some work including other other things.

like a real workflow. That's where I don't think they'll do a better job. And they'll still they'll still go and say, hey, I can do everything like you know, like service now has the studio where they're like, okay, look, bring in anything that you want, email Slack, everything, and and you can create the same thing with Salesforce also they they they they they do that. And I think to some degree I think they will be successful in in saying that hey, you know, if you if you're just trying to automate between your email and Salesforce, okay we'll we'll do it for you. Right. So so I think

I think anything that is like shallow and one or two degrees, I think they'll be able to do it. I think where startups can add a value is when there is a lot of a lot of complicated things done at at a higher level. Like like for example, I was looking at this company that does like it's just a very simple thing. They just create a code for a industrial device, right? And the customer comes and says, Hey, I want this circuit breaker, give me a

It's just very difficult to just go and just do like a query on a ERP. Like the you know, you have to look at you know, what the customer has bought before, like what how much how much does he spend, like how what is his history and even the the the request is like very s s very simple, like you have to translate that into a real like a product spec and then you have to eventually come give like a proper code, like like a document, right?

Yes, the ERP can do it, but it's it's just too much for them to go and like like create this flow for each and every customer, right? So so I feel like in that case like ERP could stay like a dumb system of record and and and you could have a a tool like a best upgrade tool that operates on top of it.

No, my my question was more towards like cybersecurity solutions, the defensive cybersecurity solutions. You have all these different silos, you have AVM players, you have pen testing tools, you have you know Unified View, in you know, the CTEMP tools. But these are all most of these tools are like all pre-AI. Right. So my but the question I have for you is do do you think a new wave of tools get developed that completely reimagine how defensive cybersecurity solutions are built, or you think the existing

Players will figure out a way to transition into the new world. Do you think because if you if if you go back, if you look, everything is very siloed from in terms of like how the assessments are done, how the risk assessments are done, how compliance is all of these they have their own silos in their fief dumps. And but they were built in a pre-AI world. Do you think there is an there is a way for like you know, completely reimagining how because from an attacker's point of view, things are the cost and scale of

exploit development is essentially going to zero. So the attackers can just go in, prompt your way into building an exploit, compromise systems and so on and so forth. So but the defenders are now sitting there with their old tools in terms of you know running a scan, finding, prioritizing and doing all these things, creating tickets, which could be outdated in the age of in the age of AI. I'm curious what do think happens in the age of AI for defenders who are trying to defend against these attacks. So do new solutions come up or the existing solutions will find a bridge to

Get to that point.

I I think some of them are just like not like on the on the vulnerability management side, right? If you look at the vulnerability scanners, they were just like scanning and just like correlating like an asset to a a C V, right? I think those things are just not going to cut it at all. I do feel like you need

Even A scanners could be built with AI now. So even the AI there are the AI native scanners could be built much more cheaply, are much more effective in finding the gaps other than the older tools.

Yeah. I I think I think where the existing companies can do very well is like when they have a lot of data, right? And that's where I think some of this platform companies like Mal Auto and CrowdStrike, the more data that you have actually you're able to correlate things faster. Because what's going to happen is like, you know, the the the attackers are going to they're going to basically you just keep on attacking, attacking, attacking and

They are going to try to just weave through the products, right? So the more you have like siloed product, I feel like that's actually a liability. So, like for example, if you're doing authorization in one platform and PAM in other platform, and those platforms don't talk to each other, you have a hole there, right? So so that's where I I think that and also you need to think like think like the there are two two

The two two levels of platform. Like one platform is like you what Palo Alto has is like, okay, I have network, I have identity, I have everything. Like that's a higher level platform. It's it's is both from a selling perspective and and on the data side also, which they have not integrated that much right now. And then there's also like a like a platform like like a like a attack vector platform, which is like an identity, right? So if if you have like everything on the identity, like you know, I authorization.

IT DR, PAM together and you're able to have like a 360 view of everything, every employee, every activity, then that strength strength strengthens a lot more than having like you cannot just do like author get an authorization from a from a player like a strong DM and then you're using Okta for IM and then you're using Sale Point for governance.

I get it. the next one is also very near and dear to my heart. this is the one for and again goes back to your memes on Bollywood movies. this is about the this is about the Durandhar movie where Ran Beersing, you have one with AI for security versus security for security for AI. And the folks in AI for security, they are well trimmed, they're wearing a nice

They're wearing a nice shirt, nice tie, nice p nice pant, and then the the the one the founders in security for AI, they're all ragged, they're all you know, they're in a war, in a battle with blood on their face and so on. I'm curious why you felt that why you felt the need to write to to do that meme.

Yeah, so yeah, so in in security, you know, I I've been they're spending time on those two two areas, like AI for security and security for AI. And the more I talk to like the AI for security, so if you look at a lot of the AI for security, you know, they are actually selling AI. Like a lot of this like AI sock companies, they they they're basically bringing LLNs in. And they and they come in as like, hey, we are an AI expert, right? And that's why, you know, you see this guy with the with the tie and everything. So this is like

Ranwir seeing when he was a nice boy, right? So he was this this very very charming and and and and not like a bad guy, right? So he was he like he wanted to go to military and serve and everything, right? So that's that that's the personification of AI for security, like you have to go in as a kind of an academic L expert and I'm going to bring AI to it, right? But on the other hand, like on the on the security for AI, I think you

Charming, very charming and young and

interpreted little bit different and I think both the impr interpretations are correct. It's like that thing is like a like it's very crowded. Like every single day, like twenty twenty five companies, like every company, like every other day I see a pitch saying I'm doing AI governance, AI governance, AI governance. First of all, I think there's a gang war going on i within AI security itself. Like people are just fighting with each other. And second of all I think, you know, in order to s like there's two gang wars going on. Like one is like you know with

the with the hackers, with the employees who are using it and misusing it and then there's a gang work going on within themselves also. Right. So that's why I kind of put that thing like saying that, hey look, you cannot like this very be very softy, like a like an AI forward company to survive in in in in AI security.

The way I perceived the meme, and correct me if I'm wrong, the way I perceived the meme is AI for security, the path is much more clear in terms of the use cases, in terms of the problems we can solve. and that's why, you know, you have the clean shirt for AI for security. But the path for security of AI is uncharted. You don't know what is gonna work, you know, is it prompt you know, today it is prompt injection, tomorrow it is something else, and it's never clear you know

other day it could be AI governance. It's never clear what is of value to the end user in terms of security of AI. That's how I perceived it. But it looks like that's not what you're going after or or was it? Maybe it was like a there was like some second meaning to it.

Yeah, I mean y y you could interpret either ways, yeah.

Awesome. and the the the ninth one I had is this you did something on SVB, where all this money is getting exchanged between the money is getting exchanged between the frontier models and the the the banks and NVIDIA, all of these are essentially circling the same same money to get their job done.

it it was done in the context of SVB, but I think it is still relevant in the context of what we see today with what is happening with OpenAI, what is happening with you know Nvidia investing in OpenAI, and then Enthropic investing in Nvidia, and then Grok, Grok giving the money, I don't know, the giving the compute to Anthropic. All of it's the same money that is just rotating in that you know, Mach 7, Mach 8 ecosystem. And

I'm curious what do you what do you think of that?

Yeah, yeah. So I think the original meme was actually like that's like the my most like liked meme is during when S V B collapsed, right? So so if you look at what S V B was doing was you know, they were they were basically banking their startups and and that's like a that's like a one way thing. Then they are also giving loans to the employees and the company based on the equity. Right. Founders

And the founders too. And the founders were getting cheap loans cheap loans too.

found us two cheap loans too. So it's kind of this circular thing where, you know, I I got this loan. It was also even like not just like loans are like different level loans, like all the way from like so you can just go and say I have so much equity in this company and, you know, I want to buy like a two million dollar house in Polo and they'll give you that thing.

And they you they would even give like personal loans based on you know your equity or something like that, because they always thought like, hey, this guy's net worth is so much, like but everything was like a paper money, right? And what ended up happening is they they they took all this money and they they they invested in like what they call as like a high duration loans, right? So so and when when the economy becomes soft and and and some of this people started asking the money back and and liquidating.

just couldn't take the money out and that's why it it kind of crashed because they actually had a liquidity issue that you know it was kinda it was like what they call as a time mismatching like you know this these guys needed money on a short term basis and they had invested in long term loans right you see something similar happening and and we you see that in some of the companies like where there's really conflict of interest where like Nvidia wants like like they they actually want a fragmentation of models right so having

two companies like OpenAI and and Anthropic is not good because they eventually can go build their own hardware and then they get displaced, right? So so they NVIDIA keeps on investing in this like RL and like open source LLM models. You can see like fundamentally there's like no path to them for success at all on a standalone basis because they don't have a killer app.

And they and they and they do that at the at the at the infrastructure layer as well. They invest in CoreVee and they're essentially trying to make sure that AWS and Microsoft doesn't don't become so big of a customer that they can dictate terms to that. So they go in and they invest in Core Weav and all the smaller infrastructure players they go and support Elon. Hey Elon, we want you to build a data center with NVIDIA. So they're trying to basically make sure that they are not

Okay.

Exactly.

Yeah.

Beholden to some of these companies, and they can still continue to have their 80 mars 80% or 70% margins on these chips.

Yeah. Yeah. So so you see a lot of that, you know, they they're investing a hundred, two hundred million dollars and that money comes back to NVIDIA. So guess what this company they spend on, right? They will basically spend on on GPU. So that money comes back to NVIDIA. And so there's just like a very much of a circular loop going on where NVIDIA sales go up and they're they're generating this cash. This they invest in the startup and then the startup spend money again. And literally some of this is like forced money because

If you if you go and like value this company like saying, Hey, do you actually have a business model? And you you say you don't have a business model, I'm not going to fund you, they're not going to buy NVIDIA chips. But NVIDIA is just throwing all the caution saying that, Hey, look, as long as this guy's like the money comes back to me and my valuation goes up, I'm I'm fine.

I I want to ask this jaded question. Do you think we are in an AI bubble?

it's kind of a very different bubble, I think, the more I think about it. Like, you know, people keep comparing with like the dot com boom, right? So one of the problem with the dot com boom was like it just took time for the companies like like you know, the infrastructure was there, but you know, you know, there was like, you know, this pet shops and everything, but like people are not ready to buy online, the infrastructure was not there, right? So so so that was the problem there.

Yeah.

And now I think I feel like we have a reverse problem, which is like, you know, like you know, if you take an example of Mythos, right? Like a mythos is available now companies have to go clean up all their vulnerabilities. So they're just like going and racing to just clean it up. And after they clean it up, what happens, right? Like the they're not going to spend that much. So so if you just look at like all vulnerabilities get solved in twenty twenty six, twenty twenty seven is going to be very incremental. And it's going to even not be incremental because you're going to use AI to s

to fix them at a start. So so like next year around, like there's no like there's no spending on that, right? I think and you can see that a lot with with everything, right? Like like even like this lawyer thing like Harvey, right? Like, you know, this they're charging a lot of money now, people see the value, but next year around they're going to say that, hey, look, I can use Microsoft, I can use like Claude to do that. So I think the problem

with this cycle is it's like a like a almost like a global maxima like of spending and then everything is showing that it's going to be much smaller in the future. But in the in the dot com it was the other way around. It was like long term it it was better, but in in AI, the long term actually is worse than like now.

Do you think the value then if that is the situation, do you think then the value essentially concentrates into the top ten, top seven model infrastructure companies and because they can serve all these needs and the smaller companies will die on the wine? If that is the if you're saying because all these products and all these models can do it, I can do it in Microsoft, I can do it in Cloud, I don't need Harvey. Do you think that then consolidates into the top top ten, top fifteen?

Yeah, it's a kind of

companies and is that one of the reasons why there is insane amount of funding? You know, a trillion dollar valuation, one point I I just read Anthropic had a one point two trillion dollar valuation today or yesterday. And it went it was six hundred billion dollars last, you know, three, four months ago.

I I think the the the the premiere models will do will will actually do well, right? It's like a modern like a Google kind of a thing. but one of the things that I keep debating is will they be focused mainly on like high end use cases, right? Like you know, coding is like a high end use case. Or can they can they also do like this co pilot app or like a small workflow, right? Like would you want to spend like an expensive token?

for that. Like somebody said this really well is like the token cost is is different for different use cases, right? So it's the same token from Cloud. You can actually like resell that token actually for a profit because the use case use case is higher, but if the use case is not higher, you'll actually sell it at a loss.

Yeah, we do.

Yeah, I've heard I mean I've heard rumors around this. I don't know how my how true these are, but you know, for example, the tokens that are related to financial engineering making money to you know for banks trading, those could be valued much higher. And we we may not even be able to get access to these models because the Citadels of the world would buy these models, access to these models for billions of dollars. And because they can make billions of dollars on those, on those models. So it's it's it's you know

know it's a wait and see let's see how this things pans out in the future. The last one I want to cover, and I don't know if you've had like personal experience with this, but I've definitely perceived this. There are there are a lot of good companies that are out there, especially in cybersecurity. There are very a lot of good companies, but they don't get the same love from Wall Street. And then there are some companies that are very good at the marketing side or the analyst relations side or some other angle aspect of it and they get all the love.

They get all the love, all the praise. And then you had this meme that went out where there is this guy walking around shaking hands with a bunch of CEOs of cybersecurity companies. And each one gets a different treatment as they go through as they go through the handshake. I'm curious if you have experienced it personally or have you seen such situations develop in the you know in your career.

Yeah, I think the I think it's becoming more and more important. and I saw that firsthand, like at at VMware. So mm you know the company used to spend like executives is spend like insane amount of time on on on messaging and marketing, like and and and the the execs actually used to own that rather than just giving it to the marketing people. Like like a CO like Ragu will like keep thinking like, Hey, how should I message this to customers, like if I buy something?

Like what does it mean in terms of the platform and all of that, right? And and that really resonates very well, right? And in cybersecurity also you see some of these CEOs, like I I would say like the best CEO is not even like Nickesh, right? Like I think the if you look at the best CEO, like Cloudfair guy. So he has the highest multiple in in security, right? So he goes and, you know, he he messages like, you know, he's a very rare asset and like everything is coming to him, like with the edge and everything.

has all this data and everything. Like so it is becoming very, very important. Like I was just talking to a friend at Okta the other day and it just like, you know, Todd doesn't do a very good job. He's not very like vocal. He doesn't talk about like why investors should invest. Like what is it doing long term? Like this committing and it's becoming very important like both from

customers as well as investors as well as employees because people want to see what's out there. Like what else are you doing? Like so so previously it used to be like hey, you just execute you you you you create results, but it's all like like backward looking, right? Like if you if you want I mean the investors out there literally is saying that hey I want to invest in cybersecurity, what are like three or four stocks should I invest in, right? And for them they don't they don't want to

you know, do like this quarter by quarter. They want to go, you know, develop some confidence in the team. And a lot of times it's like confidence in the team. Like like what Nikkei says, like, okay, what you see right now is not even what it exists long term. Like I'm going to just keep buying and expanding. So believe me, and he's very vocal about it. And one of their employees actually said this really well. It's like he says, you know, like he's just out there talking everywhere.

That is his like way of communicating. So he's like, okay, if he goes talks in the conference, he says something. It's not that he's just c talking for the conference. It's also relevant for you. Like he went and spoke in a conference that's this particular category is going to die. And literally there are people like in the company working on that category and they're like, Holy shit, what is this guy saying? And then he's like, Yeah, he he he's saying what he's saying, and it's true, it's not just like for that conference, right? So yeah, so I think the the the background of the meme is like this is a very famous like

Barack Obama beaten greed, right? So when Obama took office, he went and shook hands and literally this is what he did actually. So it was a it's a very epic meme where he he went and like shook hands and then he hugged like the African American people, like his own people. So like the people that you love and and you like. So so that's what is happening on the investor side is also like people are relating with whom they can relate to, you know. They also like people who spend time with them, like, you know.

know some of these other companies like Fortinet and like Zscaler and LetNet Scope, they're they're their their team doesn't spend that much time with analysts. They don't like humor then they don't share information. They and also it's a it's it's a bit about also sometimes it's all about, you know, saying that I'm going to do it and and doing it. It's creating that trust, right? So so if an analyst say, okay, what are going to do next two, three years? So some of these CEOs are not sure like they don't want to commit, but that's also the problem that if you don't commit

They feel like you cannot deliver.

Yeah. Pramod this was this was a fun interview. ha I have to ask you, what's the next big meme that is dropping from from from home? Like what are your what's the next big meme you're working on?

Yeah.

Yeah, so the the the the tragedy is I think a lot of people don't get it. Like a lot of people think that, you know, I have a team in India or someone like that doing it and and some of this like the meme that becomes viral actually that literally takes me like ten, twenty seconds. And I also don't know like what people are going to relate with or not relate with, right? So there there there's some memes that have spent a lot of time and like, you know, perfecting it, everything I put it out there and it doesn't go anywhere, right? So so right now I think I'm I'm more like a fail class. So like if I get an idea, I post it.

and sees what reaction people have it if it's like a

Have have people have people accused you when do you actually work because you're memeing so much?

Pramod Gosavi (01:00:02.198) Yeah, it was it it was happening like two, three years back because I mean the the thing is like some of this idea just come like free flowing for me. It doesn't take me too much time, but people f think that hey, you know I mean they have some people actually they they spend a lot of time. like l the like you see some people on Twitter and all, like like they are just like on point for everything. So to me I think I do think that they're not spending time because if you're going on every single trend, like every single thing.

Because I myself, like if I want to create something, I know it it takes time, right? So so if you want to like comment on every single thing out there, it is a lot of time. I was not doing that, like it basically like whatever comes to me naturally, like whatever I see, I do it. Like like even if I do it like like initially I wanted to do it like once a week. then i sometimes it becomes like once a day. Sometimes, you know, there's a l lot happening. So I think multiple times over day. But now I think

I don't have any kind of frequency so anytime I see something I just post it.

Mehul Revankar (01:01:02.36) In the the the good thing is you get so much inbound in terms of interactions with the with your with your companies and investors. you know, you have this raw data fire hose coming your way. So it's also source of some of the memes that come out. So it was it was super fun to interview pr interview you pramoth. Keep those keep those memes coming. I'm going to stop the recording now, so hold on. Don't leave.

Pramod Gosavi (01:01:23.512) Yeah, yeah.