EP: 08 From Data Security to AI Resilience: Vasu Murthy (CPO, Cohesity) | What the futr

Welcome to Episode-8 of What the futr.

This episode of What the futr features Vasu Murthy (Chief Product Officer, Cohesity) in conversation with Sandesh Patel and Chris Brandt.  Watch them explore the rapidly changing relationship between data infrastructure and artificial intelligence.

From the idea of data centers in space to the concept of AI resilience, this conversation dives deep into how enterprises will manage, secure, and scale AI systems in the years ahead.

Stay tuned if you are interested in AI infrastructure, enterprise data strategy, and the future of intelligent systems

Subscribe for regular episodes.

👉 Explore the platform: https://futrconnect.io

👉 Business inquiries: sandesh@futrconnect.io

Transcript

00:00:00:05 - 00:00:23:06
Sandesh
All right, well, today we have a special one especially for me. In my entire career, I have spent at helping customers store data, protect data, and restore data. So this conversation is going to be following for me. Today we have Vasu Murthy, Chief Product Officer from Cohesity. So we're going to talk about two amazing topics when we talk about data and AI.

00:00:23:08 - 00:00:27:21
Sandesh
So get ready. You know.

00:00:27:23 - 00:00:33:23
Sandesh
You got to change the game.

00:00:34:01 - 00:00:38:10
Sandesh
So.

00:00:38:12 - 00:00:41:21
Sandesh
Vasu thank you so much for joining us today.

00:00:41:22 - 00:00:43:08
Vasu
Thanks for having me.

00:00:43:10 - 00:00:53:03
Sandesh
I did have a question for you. This is a total Indian question. And by the way, you'd be impressed how how they see Chris can be here. Oh, yeah.

00:00:53:05 - 00:00:59:12
Chris
I haven't gone to one of the Patel matchmaking sessions yet, but you know.

00:00:59:14 - 00:01:00:23
Speaker 4
He's close.

00:01:01:01 - 00:01:03:16
Sandesh
He. He loves Indian food. He had,

00:01:03:18 - 00:01:04:01
Chris
That I.

00:01:04:01 - 00:01:10:20
Sandesh
Do. You had a, He had a chef that was cooking, him Indian food at home.

00:01:10:22 - 00:01:14:01
Chris
By chef, he means our babysitter.

00:01:14:03 - 00:01:15:18
Sandesh
Oh, that was your baby sitter?

00:01:15:20 - 00:01:19:08
Chris
Yeah, I don't I don't like you. Think I have a personal chef, dude?

00:01:19:10 - 00:01:29:17
Sandesh
Well, you know, there's no like that. There is. It's called, Russell a Russell. It's in, Chicago. Look it up, dude. These people will cook Indian food for you.

00:01:29:17 - 00:01:37:15
Chris
You got one of those here in Evanston on Dempster Street. You take him home. They're full of full, fresh cooked Indian meals.

00:01:37:16 - 00:01:41:09
Sandesh
No. Every day, every day it's homemade. And then they bring it to you.

00:01:41:11 - 00:01:48:01
Chris
It's like a meal. Well, this one, you have to go get one. Or maybe they deliver to you. I don't know, but it's like a it's an Indian meal prep thing.

00:01:48:03 - 00:01:50:11
Sandesh
Oh, I'm too lazy for that. I don't want to.

00:01:50:11 - 00:01:55:12
Chris
Go like that. But they're pretty much already done. You know, it's like you get the heated up this pretty much you.

00:01:55:12 - 00:01:59:00
Vasu
Have to get someone to actually cook it for you. So you got a choice. Well, I.

00:01:59:00 - 00:02:03:15
Chris
Mean, they cook it for you basically, but you got to eat it up. So.

00:02:03:17 - 00:02:04:03
Speaker 4
So I, so.

00:02:04:03 - 00:02:12:21
Sandesh
I just wanted to ask you, and you can answer as little or as much as you want, but, what part of India are you from originally?

00:02:12:23 - 00:02:24:14
Vasu
So genetics wise, I'm from South India. Okay. Right. So southern mama. More south to climate to Bangalore climate. I don't know if you have had a political climate there.

00:02:24:15 - 00:02:24:23
Sandesh
Okay.

00:02:24:23 - 00:02:49:11
Vasu
Yeah. It's, you know, south of Bangalore, but I grew up in multiple places because my father was getting his job, was taking them various places. And so, you know, in Maharashtra, in Bengal. So I went to college in my epic character, which is some bay, you know, eastern side. But I grew up my high school was in the western part of it, but I'm actually from South Chin, so,

00:02:49:13 - 00:02:52:22
Sandesh
What languages do you speak then? Do you speak multiple?

00:02:53:00 - 00:03:04:15
Vasu
Multiple? Yeah. I think typical in, in India. Right. Of course. Yeah. You know, I speak Kannada at home outside in Tamil, but in the school, the teacher Hindi and English.

00:03:04:17 - 00:03:04:22
Sandesh
Yeah.

00:03:04:22 - 00:03:06:07
Speaker 4
Yeah, yeah. Right. When I was in.

00:03:06:07 - 00:03:26:04
Vasu
High school then I would there's been one AP outside. My college, the a lot of locals speak Bengali. Right. So it's like this. I mean, it's like a mini Europe, isn't it? I mean, I think the head of the Europe, I don't know for multilingualism at that common, but,

00:03:26:06 - 00:03:44:16
Sandesh
Yeah, yeah, India is definitely special that way. I, I was just there last year. I've gone a lot when when we were younger with typical Patel family. Right? We, we go there for two weeks. The first week, they find someone to marry. The next week is the wedding. Then they come home.

00:03:44:18 - 00:03:46:01
Speaker 4
And it's the there's.

00:03:46:01 - 00:04:14:18
Sandesh
The standard routine. And, we, But I loved it. I was so fun. So great. And, just going back there, just seeing how much, how different India is now. It's quite incredible. The only disappointment I had on my last trip was when I went to Delhi. I was little, you know, seeing all the pollution in the air, even like when the sun, which should be just beaming, there was so much pollution.

00:04:14:20 - 00:04:19:21
Sandesh
It was. Yeah, it's kind of sucks. And so that's why a lot of people have left Delhi.

00:04:19:21 - 00:04:37:18
Vasu
Pollution and Bangalore traffic pollution still pretty bad, but the traffic was even worse. So, you know, like half a mile in like half an hour. You just walk there, you could just walk even. But the thing is, if you are like, you know, five miles away, ten miles away, you don't want to like that. So.

00:04:37:20 - 00:04:42:22
Chris
Yeah. So I'm just going to walk a block because he'll he'll sweat.

00:04:43:00 - 00:04:51:12
Chris
Which this is, which we once took an Uber around the block in Dallas because it was so hot out okay. Like how would you survive in India.

00:04:51:14 - 00:04:57:05
Sandesh
So I just want to point out I'm an American born. They see okay. So like ABC money.

00:04:57:07 - 00:04:58:07
Vasu
Okay ABC.

00:04:58:09 - 00:05:02:20
Speaker 4
You know that was oh like oh he knows it. He knows this.

00:05:02:22 - 00:05:22:22
Sandesh
Well he's he's one of my best friends. He knows an ABCd really? Well, that's for sure. And, so we're in Dallas. We're going to a customer meeting, and literally our hotel is like a 4 or 5 minute walk. It really wasn't bad. And Chris is like, dude, why? Why would Uber just let's walk?

00:05:22:22 - 00:05:24:21
Sandesh
I'm like, dude, it's 110 degrees.

00:05:25:01 - 00:05:30:00
Chris
He argued with me about this over and over. I'm like, how? How bad could it be.

00:05:30:02 - 00:05:40:01
Sandesh
In the valley? I finally I'm like, fine, we'll just walk. Then we walk in. Man, he was in there of sweat.

00:05:40:01 - 00:05:42:01
Chris
He sweated for like 45 minutes.

00:05:42:05 - 00:05:57:11
Sandesh
I'm in the meeting with the client. It's like it's a good thing that we know him really, really well. We knew him really, really, really, really well. But drenching in sweat, you know? And when you're a sales guy and you're sweating.

00:05:57:13 - 00:05:59:15
Speaker 4
That's not not a good look.

00:05:59:17 - 00:06:01:20
Vasu
Without a good, like, the other guy clerk.

00:06:01:22 - 00:06:05:23
Chris
Hey, at least you're not a waiter. That's the worst one.

00:06:06:01 - 00:06:06:07
Sandesh
Way to.

00:06:06:07 - 00:06:07:19
Chris
Sweat eating your food.

00:06:07:21 - 00:06:11:17
Sandesh
I also saw that you went to it. How was that experience?

00:06:11:19 - 00:06:13:15
Chris
Oh, man. And it grad.

00:06:13:17 - 00:06:14:10
Sandesh
Yeah.

00:06:14:12 - 00:06:39:04
Vasu
Yeah. Well, you know, that was way back when I actually had in the college in 94 said, can I teach me? And, yeah, I mean, there are 2000 people getting the 2000 seats and and of course, if you want to compare science, you got to be in the top 300 if you want, like, you know, so there's, there's a well well defined gradation, like if you're computer science and electronics, communications and then electrical engineering mechanic.

00:06:39:04 - 00:06:48:19
Vasu
And that's up there. Everybody knows that's what that's a rat race is about. And so I couldn't make the 302 computer science. So I went to the second best.

00:06:48:21 - 00:06:50:00
Speaker 4
In that, like, a.

00:06:50:00 - 00:07:09:04
Vasu
Thing is I didn't want to do any of those. I wanted to be a physicist. That's. I was telling my, you know, physics tells that kind of love and physics amount. And so I'm getting my undergraduate courses in physics. And I was in high school, but my dad, went to my physics teacher and asked him to convince me to, you know, get into the engineering.

00:07:09:06 - 00:07:17:06
Vasu
And so my, my physics teacher, who I had a, saw the world up, came to me and said, no, electronics is just an electron physics. I should go to learn it.

00:07:17:08 - 00:07:20:02
Speaker 4
Yeah. I think they're just like,

00:07:20:04 - 00:07:30:15
Vasu
You know, brainwash me into it. It's not like I regret it. Maybe. Maybe after this, I'll go back to my, love for physics and do something, because this is the age of physics. Isn't that like an hour of of atoms?

00:07:30:15 - 00:07:32:13
Chris
Material sciences in particular?

00:07:32:16 - 00:07:56:19
Vasu
It's now the age of atoms of software is coming to kind of a, not a, not a close, but the age of atoms are finally catching up. Right? So there's there's just nothing lasts. The last 30 years, I've tried all the things I could do with software, and now I think we got to do, making them, you know, broader, more energy, more, physics, no more mechanical stuff.

00:07:56:21 - 00:08:03:18
Vasu
And so, yeah, maybe I'll, actually, I'll go back to that. But, that has always been my dream to be to work on physics.

00:08:03:20 - 00:08:23:15
Chris
Well, I was, I was a political science major. And you know what I do too. Like, it's that scratch that itch is I like, build guitar pedals, you know, so I get to do a lot of, you know, soldering SMD stuff and, you know, like, I'm always. Yeah, I'm tinkering. So I like, I, I exercise a demon a little bit and that's what that's what Linus Torvalds has started doing that too.

00:08:23:16 - 00:08:28:23
Chris
I'm on like these weird little puddle building forums with, now it's just strange.

00:08:29:01 - 00:08:30:07
Speaker 4
It's like.

00:08:30:09 - 00:08:33:12
Vasu
Wow, you know, what was and, Chris Pratt and. That's right.

00:08:33:17 - 00:08:46:11
Chris
Well, he's he's he's I've been using the Stm32 microcontroller and that's seems to be where he's landing now too. So it's a it's a very small community on, you know, like discord. Yeah.

00:08:46:13 - 00:09:08:02
Sandesh
My, my last story and then a question for you and then we'll get started. So way back in the day when Cohesity was very small, Chris and I used to bring in startup founders and VCs and stuff like that to Chicago, and we would, schedule these dinners. Chris and I just have a very loyal customer base.

00:09:08:02 - 00:09:18:14
Sandesh
And so, you know, we getting people there was an an issue and we'd get a lot of executives and they loved it. Right. Like we're bringing the Silicon Valley to Chicago. Well, we bring Moorhead. So we have this dinner out.

00:09:18:14 - 00:09:19:14
Chris
This is a great story.

00:09:19:17 - 00:09:20:22
Sandesh
We we we do the right.

00:09:20:22 - 00:09:21:21
Chris
Now exactly where you're.

00:09:21:21 - 00:09:24:19
Speaker 4
Going. It's so good. We do.

00:09:24:19 - 00:09:45:02
Sandesh
The roundtable discussion and we talk about why secondary storage and like, you know, the iceberg. And then we're talking about all this other. And then at the end, you know, we're we're kind of trying to make casual. So I asked him the question. I said, well, most if you didn't go into technology, like what would you what would you want to do.

00:09:45:04 - 00:09:50:12
Sandesh
And his his answer was very interesting. What do you think he said, no.

00:09:50:14 - 00:09:51:00
Chris
You'll never.

00:09:51:00 - 00:09:52:14
Sandesh
Guess. You'll never guess.

00:09:52:14 - 00:09:53:06
Speaker 4
This.

00:09:53:08 - 00:09:54:10
Sandesh
So it's an.

00:09:54:13 - 00:09:56:19
Vasu
Experience to be a politician?

00:09:56:21 - 00:10:00:02
Speaker 4
No, no, no, he said, I said, I've.

00:10:00:02 - 00:10:03:08
Sandesh
Always been interested in bodybuilding, so I'd probably.

00:10:03:08 - 00:10:10:23
Speaker 4
Be cool if I'm like, what? Why so? So you're this.

00:10:10:23 - 00:10:13:16
Sandesh
Incredible genius and you would going.

00:10:13:16 - 00:10:16:19
Speaker 4
To to be a weightlifter.

00:10:16:21 - 00:10:22:08
Chris
It just seems like I like the discipline. I like their focus. You know, it's like, okay, that's strange. One.

00:10:22:13 - 00:10:28:15
Vasu
You go, I mean, yeah, you got the rest of it. Yeah. I think a discipline focus on how it be the teams.

00:10:28:17 - 00:10:36:20
Chris
I mean, like, you know, I, I admire I admire his willingness to take that very long journey that he would have to go on. Yeah. To get from where he is today to there.

00:10:36:20 - 00:10:46:15
Vasu
But I just expect that I mean, I know someone else. I asked this, very successful tech person. I said I would be a politician. I was like, flawed, like, what is that?

00:10:46:17 - 00:10:47:16
Sandesh
I know?

00:10:47:18 - 00:10:48:04
Speaker 4
Oh, God.

00:10:48:05 - 00:10:52:15
Vasu
You know, you never know what's inside. What what drives people, right?

00:10:52:17 - 00:10:59:14
Chris
I know what mine is. I would be a lottery winner. I think I would be really good at that.

00:10:59:16 - 00:11:02:17
Speaker 4
You could be that right now, Chris. Yeah. That's my alternate career.

00:11:02:21 - 00:11:07:04
Chris
You know, it's it's like keeping that on the side.

00:11:07:06 - 00:11:08:06
Speaker 4
Yeah.

00:11:08:08 - 00:11:32:12
Sandesh
So last question. This is one that, I try to ask questions that are very relevant that I, before we get started. So the one for me that's top of mind, we're seeing this debate that is happening right now about the concept of data centers in space. You have Elon Musk, that is, you know, putting his argument out there and is, you know.

00:11:32:16 - 00:11:35:19
Chris
There's your first mistake listening to Elon Musk.

00:11:35:21 - 00:12:02:17
Sandesh
Oh, Chris. And, and then you got Sam Altman. That's just like, that is just absolutely insane. And like, yeah, go ahead, spend billions and there's just no way it's going to happen. And so, it it is quite a radical idea. Yes. But what? Yeah. Since you wanted to be a physicist. Yeah. Yeah. What do you what do you think about that?

00:12:02:19 - 00:12:03:21
Speaker 4
I'm.

00:12:03:23 - 00:12:31:06
Vasu
I see the thing as a lot of people have talked about what could be done in space, but everybody has a mindset of the previous generation where getting a pound to orbit is very expensive. If you make that super cheap, the whole game changes. And with Elon Musk, he already owns the rockets. So his additional incremental cost is the cost of fuel.

00:12:31:08 - 00:12:51:12
Vasu
Right. And if you get to that low I think there's a there's there's probably something you could make on it. For example, you get more power per square feet. So there is no clouds. There's 24/7. Some problem with, you know, Earth based solar is we have days and nights, plus you have clouds and you go rain. And so your factor of efficiency.

00:12:51:12 - 00:13:13:01
Vasu
Interesting. 2 to 3 x. And so for same size of sort of 2 to 3 times the power cooling is extremely easy like that. The problem it's satellites on one side is too hot, one side is too cold because it's just a space. So you can radiate heat much better. In space, the only thing is going to be, I think, and maintaining them.

00:13:13:03 - 00:13:28:12
Vasu
Right. How are you going to like, the chip fails. How are you going to unplug it? Unplug it? But there are creative ways to do that. Like, you know, there are those satellites that have been created that have like the small right in a couple feet in length. And so you could actually de-orbit the module that is, that's faulty.

00:13:28:12 - 00:13:45:12
Vasu
And they dock another module there. So it's all possible. It's all possible. I wish that I could follow SpaceX pretty closely. Sound design by working in space. Excellent on. And so, yeah, I never know what's possible. I think these are all, the world is changing.

00:13:45:14 - 00:14:15:01
Chris
Yeah. I mean, I think there's some substantial problems with with that model, though, because, like, you know, one and like, you know, satellites don't last forever. And maintenance of satellites and keeping them in orbit and the fuel required to, to to do that. And you'd have to have a pretty considering the size weight of a data center, you have to have a pretty high orbit to which would be problematic just from just like telemetry, you know, down because you can't put something like that into a geosynchronous, you know, really effectively, you know, so they.

00:14:15:01 - 00:14:16:23
Vasu
Need to do that. No, I don't.

00:14:16:23 - 00:14:39:20
Chris
Know. I just the, the latency between the transmission getting the data because it's already, you know, 22 milliseconds round trip to low Earth, right. You know, so but you know, you'd have pretty, pretty, you know, for a data center like that, that you're gonna put that kind of money in. I just feel like you have latency challenges, you know, like orbit challenges, you know, cost of, like, keeping things up there.

00:14:39:21 - 00:14:44:12
Vasu
Satellite data, each each of the satellites being just one track unit, kind of a size.

00:14:44:14 - 00:14:50:19
Chris
Yeah. I mean, maybe if you did, like, maybe if you did disaggregated it and like, did it like yeah.

00:14:50:21 - 00:15:04:14
Vasu
I'm in my mind is disaggregation because you don't have to do the same thing that you do here. There are independent units. Like how how is the Starlink is working. Right. Very disaggregated 10,000 satellites doing the stuff. But they're pretty high speed network between them.

00:15:04:14 - 00:15:08:13
Chris
But they they do orbit constantly too. That's the other problem.

00:15:08:15 - 00:15:26:12
Vasu
I know about this because yeah, I learned from my friend at the space. Yeah. They concept things fail that the orbit it and actually had been to the factory where they make those satellites. Yeah that there's if there's some there's an error there it is the orbit and it just burn. I go on to another one. The thing is the space satellites are so cheap they're just like $30,000.

00:15:26:12 - 00:15:36:13
Vasu
So like $60,000, something like that. So they can do as many. I mean, GPUs are not going to be that cheap, but, you never know. There are clever ways to do stuff and,

00:15:36:15 - 00:15:50:22
Chris
It, it strikes me that like, it's if you wanted the cooling, you could just go to, you know, like some Arctic location where you could, you know, like, get it and then you could, you could, you could put solar arrays in space and beam the power down to.

00:15:50:22 - 00:16:00:22
Vasu
I think it's a yeah, it's, it's a, it's an energy constantly available energy. And clearly it's the energy from motivating energy thing that is trying to. Yeah. Nobody knows like.

00:16:00:22 - 00:16:01:21
Speaker 4
I don't know for.

00:16:01:21 - 00:16:04:10
Vasu
Sure, but it's it's very intriguing.

00:16:04:12 - 00:16:10:03
Chris
Probably not in my lifetime kind of situation. But I'm old, so you know, I'm not going to live that long.

00:16:10:05 - 00:16:12:06
Vasu
And I, I want to be surprised.

00:16:12:10 - 00:16:41:11
Sandesh
I yeah, it's been great chatting with you with the, the chats that we've had so far have already been amazing. So, it's great to meet you. And, for me and Chris, too, we've we knew Mohamed from way back in the day when Cohesity. Nobody knew of Cohesity. And so we've seen, this baby grow. So it's just really special for Chris and I to continue that relationship, with the company, which we saw when they were small.

00:16:41:12 - 00:16:58:05
Sandesh
Right. And then they're still trying to figure out messaging, you know, and now look at where you guys are. So it's it's just super cool. So thank you so much for being a part of this. Maybe we start by, a little bit about you and your journey and how you got to Cohesity and, we'll go from there.

00:16:58:07 - 00:17:19:00
Vasu
Thanks for there's. I mean, you guys are fun, to chat with and, it's a pleasure. It's not a good day. Well, you know, I started my career as an engineer. I like to build stuff. Then I realized I have joy talking to customers more. I became a product manager, and in the process of it, I realized, you know, building products that customers love is is a lot of fun.

00:17:19:00 - 00:17:39:16
Vasu
And I want to do a startup company and, ended up starting a database in-memory database startup that Oracle acquired way back. Back, you know, that taught me about what it takes to sell to enterprises and the companies around the world, the difference in various markets and so on. And the journey took me to Rubrik, where I was a part of the journey to be a CEO.

00:17:39:18 - 00:17:55:21
Vasu
And now at Cohesity, but Sanjay did a pretty tremendous job of acquiring Veritas, making the number one market player. I feel fairly privileged. And to, to be in this position to direct products at Cohesity.

00:17:55:23 - 00:18:12:01
Sandesh
Yeah. So speaking of Cohesity, maybe you want to give us the before we get into the product in the tech. I want to give us a little bit of how did Cohesity start, you know, how did you guys get to where you are today? And then we'll get more into the actual product in in value side.

00:18:12:03 - 00:18:29:21
Vasu
Yeah, yeah. Some days in the, you know, market for so long. Right. And on what it is a guy that anything that he covered is $1 billion product that you know and you can see the trace the lineage from Google file system that helped create the Nutanix. That helped create and to Cohesity. But you could see the progression.

00:18:29:21 - 00:18:55:09
Vasu
One is about building a repository for the world's data. The second one is building the same thing for, compute and applications. And it can extend like a stellar architecture. And now the same thing to do it for the bulk of the data in the, in the enterprise. And that's what Cohesity was about. It was founded just to where all the world's data in a more efficient format and bring applications to it.

00:18:55:11 - 00:19:15:15
Vasu
And we've come a long way, as you say, in helping customers do it. You know, 70% of the fortune 500, customers and and a huge stop it footprint. Yeah. We have come a long way since the more its vision of building a platform that'll hold all the world's data and bring applications to it.

00:19:15:17 - 00:19:23:20
Chris
And I think I think what's cool is, you know, we're in this world where AI is such an, like the the only topic that people seem to want to.

00:19:23:20 - 00:19:25:22
Vasu
Talk about anymore.

00:19:26:00 - 00:19:33:13
Chris
And I'm really curious to hear, you know, what your AI journey has been like and, and where you where you've gone with that.

00:19:33:15 - 00:19:58:22
Vasu
Yeah. So, you know, data is the fuel for AI. And we were always built to be a platform that holds all on data. And there are many uses for this. And we how this uses data to make a company like our customers bet on cyber resilience. The same data can power AI applications. You know, for the longest time analytics applications were all driven by structured data, right?

00:19:58:22 - 00:20:20:21
Vasu
So tables and customer lesson orders and so on, financial numbers and so on. For the first time, gen AI bringing the whole unstructured data into the fold of analysis and bulk of the data that we store is unstructured. And so this is our opportunity to help customers with all the unstructured data they have, and put it to good use with AI.

00:20:20:21 - 00:20:40:11
Vasu
And that's one of the fundamental, and that's not the only place that you play in customer journey, but that's really one of the fundamental rights to play that we have all the world's that pretty much the bulk of the world's data, hundreds of exabytes and getting that, usable quickly without a lot of, extra investment is really what our key market crop is now.

00:20:40:13 - 00:21:04:17
Sandesh
I think, what you're highlighting there is how data is so critical to AI, and where I see so many AI solutions and startups really get challenged is if they don't get access to data, then there's really not much value that they can create. And just by that premise, some of these startups are just not going to make it just because they're never going to get access to that data.

00:21:04:17 - 00:21:25:12
Sandesh
And here you guys are sitting on piles and piles of data. So it really gives you this competitive advantage to actually do something that can add a ton of value to clients. And so as we get into that, in our last conversation, you mentioned the term AI resilience. Can you give us an, an idea of like, what does that mean?

00:21:25:14 - 00:21:46:19
Vasu
Yeah. So the business that most customers pay cohesity today is to be resilient. And what does it mean? But it means less than a business attack or a natural disaster data center, you know, shuts down or something like that, you know, to bring back the business, right? You want to bring back that business the way it was.

00:21:46:20 - 00:22:07:18
Vasu
And that's the business we were in from day one. So now traditionally called backup and recovery. And that has become way more complicated with cyberattacks, but either a cyberattack or natural disaster. Our job is to, you know, make a business resilient by bringing them back. And the only way we do that is by storing all the data that happened in the past.

00:22:07:18 - 00:22:24:12
Vasu
We have a full capture of how the business was every day in the last year, say take 90 days or 100 days. So that's that years in the past. So we have a full record of that. And so we can bring the business back in action. And that's what customers pay us for. With cyberattacks has become even more complicated.

00:22:24:12 - 00:22:49:12
Vasu
Bringing a business back means you've got to bring about business in a clean way. And not reintroduce the bad, malware or the bad actors. Log in and you've got to clean up all of that. And so, again, putting a lot of effort into making that and making that better. And cleaner and faster turns out, the same applications that we had run the running the business are going to be run by agents in the future, and it's already happening.

00:22:49:12 - 00:23:13:19
Vasu
A lot of critical components are being, run by agents. So our next step is to make the agents also resilient, right? Business has to come back UPS is run by agents or run by applications. You have to bring them back. And so, one of the pillars, what we do is to extend the same resilience to AI, which means capturing an agent state.

00:23:13:19 - 00:23:31:05
Vasu
It's memory like what it's doing, what state data working on. It's a data that is, tools that it's working on, all of them. So if something happens, you want to reset an agent state or that agent dies or something. Natural disaster and, well, data center dies, we can bring it back up.

00:23:31:07 - 00:23:36:09
Chris
Or if your agent even completely starts hallucinating and you don't know why, you got to roll it back, right?

00:23:36:12 - 00:23:45:08
Vasu
In fact, that that is the second part of it, because agents are not can be perfect. And so they do like what I call shenanigans. Right?

00:23:45:08 - 00:23:46:20
Speaker 4
So, you know.

00:23:46:22 - 00:23:49:14
Chris
That's one of my favorite words by the way this.

00:23:49:18 - 00:23:50:12
Speaker 4
Right.

00:23:50:14 - 00:24:07:09
Vasu
So because we have the full capture of the enterprise infrastructure, we can also bring it back to a state where agents didn't show up. They didn't mess up. Right. So look at this problem. I mean, agents are going to do a bulk of the work, but who is going to check in on them? They go to do so much stuff.

00:24:07:11 - 00:24:30:08
Vasu
It's not human beings that are going to check in on them. It's going to be other agents that are checking on them. And what do they do if they need to restore something back again? I just need to think to. But they cannot solve it unless they have a copy of the data in the past. And so the same value that we've been building this company into this $2 billion, you know, run rate applies to the AI infrastructure itself.

00:24:30:10 - 00:24:49:20
Vasu
And this is apart from making that a better, like, leverage and giving out the entire data, access to the data that I access with Z. So this whole, you know, set of functions that make a business resilient and it's running AI and it's making the AI itself resilient and making the business retaliate against that of space.

00:24:49:22 - 00:24:52:08
Vasu
This protect all these resilience.

00:24:52:10 - 00:25:16:12
Chris
You know, that there's some like really unique challenges to protecting AI, right? Because, you know, one you have like what what do you say. What's the important part to say? You know, like you don't probably want to, like continuously back up and learn, but the weights might be really important and somebody may tamper with weights. So you need to have from a security perspective, you may need to have that right.

00:25:16:14 - 00:25:37:07
Chris
Some of this data is is transient. So like, you know, like point in time kind of matters, you know, for the state, you know, piece of it and you know, you know, you're going to have drifts and all sorts of things. I mean, like, it's just there's a lot of, a lot more. It almost seems like there's a lot more granularity to what you got to look at.

00:25:37:07 - 00:26:02:16
Chris
And I think right now the people, people's perspective is like, well, you've got the foundational model and you just use that and, you know, like if you something happens, you just replace the foundational model and kind of go on. So I don't even think, like a lot of people are really considering backing up their AI, you know, or how they or like even conceiving of how to do it right or protecting it either, you know, right.

00:26:02:18 - 00:26:23:14
Vasu
Way back when, when clouds were new, nobody was back in the cloud either. Right, right, right, right. And way back when when, you know, pre nine, 11 data centers were not backed up. And so there was always a reckoning that that makes people up. I think 911 was a big reckoning for data centers to have a darker side.

00:26:23:16 - 00:26:44:13
Vasu
And cyber attacks have been a reckoning for many, many enterprises to make sure they, they can come back. But that reckoning hasn't happened in AI. But it's it's definitely in the minds of, you know, folks like us and some, some, you know, forward thinking CISOs and CIO on how to make it happen. Even today, the awareness about that, you should backup all data.

00:26:44:15 - 00:27:04:11
Vasu
You know, maybe no SaaS. And their email is not fully understood by all customers, but it's going to happen. It's going to happen. It's necessary. You could say it's this pretty common sense that we got to do it, and we're Cohesity has been doing really well is that you don't have to do a lot of thinking about what to backup and be too smart about it.

00:27:04:13 - 00:27:25:05
Vasu
We'll make sure it's pretty efficiently backed up. Customers like throwing, you know, hundreds of, terabytes petabytes and sometimes exabytes of data. And we can store it pretty efficiently, like many of the services, videos and consumer in the back end, we have to use the platform either backing up or even having the first copy of it. And so that's the key, right?

00:27:25:08 - 00:27:41:07
Vasu
If you have to think too much about it, then it becomes complex. But if you can make it efficient to backup everything, including the weights and including the temporary state, like for databases, every transaction sometimes like every every five minutes. But you've got to find the right balance, right? How much you're going to pay for it.

00:27:41:07 - 00:27:54:00
Vasu
But, that's what it is. I mean, if you can rewind or you can get back to where it was, do you really have an authoritative data? Right. Because you could raise it right. You could lose it so well.

00:27:54:00 - 00:28:19:12
Chris
And the question is like at some point, like, can you get back to that state because like, you know, I look at like, well, first of all, like the market's shifting so rapidly. I was just listening to somebody talk about, you know, like in January, you know, because like with the release of opus 46 and from an anthropic, you know, we've got these a gigantic, you know, coding, you know, operations that are running as long as two weeks.

00:28:19:18 - 00:28:42:05
Chris
Whereas in January, you know, there was like it was miraculous that you could run one for seven hours and then that's just that's like in two months, you know, that level of changes happen. And like when you work in these all alarms, you know, sometimes I will find that it it runs differently in the morning than it runs in the afternoon because like they're upgrading them so fast and changing them so fast.

00:28:42:07 - 00:28:52:01
Chris
So getting back to that state, there's value in getting back to that state. But like how how can you do that. You know, like what what what are the tracks.

00:28:52:03 - 00:28:56:08
Vasu
Well getting back is not always the solution right?

00:28:56:11 - 00:28:57:02
Chris
Right. That's true.

00:28:57:02 - 00:29:18:01
Vasu
Agent takes an order at the wrong price. Your answer is not they want to get back to yesterday's state and forget about the order. Right. So they've got to go out and find it. So. So not all, not all remediation of an error is just going back in time and continuing, you know, and I'd be able to help in all those situations.

00:29:18:01 - 00:29:37:17
Vasu
And we're pretty honest and upfront about it. Yeah. Having the data helps because you can go investigate what went on. What was the state before. What was that later especially in cyber attack scenario, we can we can go back and trace everything that happens. Remediation is a more complex problem. And yeah, you know, to do it at scale, you probably need intelligence and maybe some other agents.

00:29:37:17 - 00:29:59:10
Vasu
Hopefully they don't have this in different ways. And then the original agent had this in it. But it's it's a more complex problem. And that is why we don't believe we will, you know, be able to end all up getting our everything right. So we have to work with many, many, many other partners to make it happen, like detecting an agent, whether it made an error, how easy that are difficult, and then very difficult.

00:29:59:15 - 00:30:17:23
Vasu
Very difficult because you can use human beings to do all of that. You have to actually have another agent, and that needs to be complementarity. At the rate where the other agent makes an error should not be the same place the original agent makes an error right? You got to make sure there's high probability of detecting the error and figuring out how to fix that.

00:30:17:23 - 00:30:41:07
Vasu
And so our goal here is to partner with all all, all other, partners like Facebook to provide data, people that provide agent security. It could be a service. Now that is building a whole agent control tower. It could be Datadog which trying to debug performance, you know, stuff with applications could be Google is billing all security center on.

00:30:41:08 - 00:31:02:17
Vasu
But we have vendors, our partners, suppliers doing a whole bunch of job on agent permissions. And so it's a lot of, parties need to work with. But have we all figured out how to detect errors? No, I think they figured it out. Yeah. And remediation also is going to take a lot more touches. But we get that in the early days.

00:31:02:17 - 00:31:13:01
Vasu
And we have something that would be indispensable in going and remediating it, investigating and remediating it. But just a full snapshot of what what happened in the past.

00:31:13:03 - 00:31:14:08
Speaker 4
You know.

00:31:14:09 - 00:31:56:01
Chris
When I think about this, sometimes I, it's it's it's more because getting an an, an AI back to a state to reproduce outcomes, you know, like it's it's intentionally variable, right? I mean, you like, even if you rerun the exact same state and everything two times, you're going to get different answers. Yeah, I almost wonder is like, there's going to be like a big business in preserving output, you know, like, and maybe that's maybe that's really the answer is like sort of preserving that chain of output or and having some, you know, you know, like the sense of what the, the inputs were to, to retrieve that output or, you know, some

00:31:56:01 - 00:32:04:16
Chris
of the, the behavior that, that went into that. But, you know, maybe that's, that's a big valuable piece and that's a lot of data. There's a lot of data.

00:32:04:16 - 00:32:26:07
Vasu
Yeah. Yeah. That's what we are good at handling a lot of data. But really with you totally with you on that, I think, without having a full idea of what happened and what are the variations possible, how would you ever, be able to investigate or fine tune it for a better outcome in the future? And there is going to be benchmarks that is going to be, log of these outputs, synthetic data.

00:32:26:11 - 00:32:46:14
Vasu
Right. So the interaction produces, a set of data. And where once you judge whether it's good or bad, you can actually use it for training too, and that's for sure. So yes, and I can definitely say in the foreseeable future, I mean, you never want to forget what happened. Like even with human interactions, pretty much every customer that was called the to call today those columns are being recorded.

00:32:46:15 - 00:33:01:17
Vasu
Right. So the capturing that has as a value in itself, definitely going forward, it has more value than the cost it is to store those bytes. And we are here to make sure that the cost is very low. So there's data is going to be that.

00:33:01:19 - 00:33:04:10
Chris
Yeah. And a lot of it.

00:33:04:12 - 00:33:28:06
Sandesh
So if I was to like try to simplify in a sentence of how how does Cohesity solve problems. By leveraging AI. I look at the core value proposition that yes, you store data and you help customers recover that data. And what I think is unique here is like, you can use AI for so many different use cases in there.

00:33:28:07 - 00:33:43:18
Sandesh
Yeah. You can help people first prevent the attack. Yeah. And then and and then if there is an attack now you can use AI to recover quicker. Is that is that kind of a fair way to simplify it a little bit.

00:33:43:20 - 00:34:02:16
Vasu
Yeah, 100%, 100%. In fact this prevention recovery process, we call it five steps of cyber resilience. And each of those steps we can actually use that use AI. Right. First is to make sure you have protecting on your protecting all the data. But the big problem is how do you know you're protecting all the data. You're kind to be able to scan it.

00:34:02:16 - 00:34:18:03
Vasu
And even now and customers don't know that, hey, somebody has made a copy of the bunch of data in S3, or that there's a bunch of files of SharePoint that are not being matched up. And so we can use AI to figure out what to protect and also what not to protect. Me what to delete. Like there's a lot of redundant data.

00:34:18:06 - 00:34:37:06
Vasu
Most customers don't know what is in this big, you know, 100 terabyte filer, right? They don't know what is it? Better help them understand that. Delete what? They don't need is that self is a big deal and I helps them. Classification. Making sure backups are always recoverable means that you want to make sure that is insufficient copies.

00:34:37:06 - 00:34:56:09
Vasu
Interesting critical data. They don't have copies. They're not bolted. They don't have a slide before we can detect that it AI and help customers detecting malware when that happens, how that happens if you can expose it early on, we can prevent an attack. And that is if you have a whole bunch of tech that's working on using AI.

00:34:56:11 - 00:35:15:22
Vasu
And then there are short steps practicing recovery. So you don't want to be stuck, you know, when the fire happens that you're figuring out where the exit is, you've actually done a fire drill and you know exactly where to go. So practicing that we actually create automated blueprints, for applications. Here's the most important application. This is how we recover.

00:35:15:22 - 00:35:34:14
Vasu
We help them practice that. And then that's like, you know, constantly scanning and making sure your your data is still following the policy and, and that you don't have any violations of it as it's happening. Right. So data tends to proliferate entropy. You know, it's not too bad. Things become chaotic. And you've got to call arena then.

00:35:34:14 - 00:35:46:21
Vasu
And so all of these five steps you have are in fact, not only that, I mean, how we work is also AI, right? So yeah, we do. We now started using AI every day.

00:35:46:23 - 00:35:47:18
Sandesh
Yeah.

00:35:47:20 - 00:36:15:02
Chris
So you know, I, you know, like preventing and detecting AI based attacks is a really tricky space to be stepping into. And yes, you know, like recently we had, you know, like to kind of big incidents. I think one was sort of the, the anthropic issue where they had, like a pretty much autonomous set of agents that were, launching attacks from there.

00:36:15:06 - 00:36:48:04
Chris
I think they attacked like 30 different companies and, and anthropic eventually identified it. But, you know, like each each task was sort of discreet agent that, you know, didn't look particularly malicious. But when you aggregated the agents, they became rather malicious. And then we saw recently, malware that's being introduced where it's utilizing public alarms to increase its polymorphism and rewrite code and, and, add features that it doesn't have when it detects certain potential vulnerabilities and things like that.

00:36:48:06 - 00:37:13:21
Chris
So, you know, like the, the, the quantity and vector of these attacks has just like gone through the roof. And then you got a, you've got like, you know, poison data poisoning and things like that that get extraordinarily sophisticated. And I can imagine really difficult to identify what I mean. Like, what do you what do you think is going to be the sort of state of the art of like how how you approach that problem?

00:37:13:23 - 00:37:32:21
Vasu
So, Chris, you know, there are so many security companies at any point in time. And just because it is such a big beast, like an elephant, it's like everybody grabbing a different parts of it, and solving a specific problem. Yeah. I don't think we can expect, one company to be able to figure all of this out for sure.

00:37:32:21 - 00:37:33:14
Sandesh
Sure.

00:37:33:16 - 00:37:53:21
Vasu
Right. And so we have to work with an open interfaces with, other security companies. So as they get better at solving one problem, like data poisoning, figuring out data poisoning, it's not easy injection. It's not easy. Yeah. Verifying if an agent is doing the right thing. Detecting when it makes a mistake is not easy. How do you know that?

00:37:53:21 - 00:38:15:05
Vasu
It's not easy, right. All of them. And and so what we are trying to do is to build an open interface that any of them can actually leverage our product, intelligently and, and, have it as a part of remediation, have the point of detection. If an agent wants to investigate what happened in the past with an agent, they can actually query us and figure out, you know, what happened in the past.

00:38:15:06 - 00:38:33:23
Vasu
If they wanted to figure out how to remediate it, they can query our, you know, system, a record. So we are working with a Databricks, for example, and exposing all these data that we are collecting into the, into the Databricks, data lake. And so custom customers could build agents that can query these things. So we want to be open.

00:38:34:01 - 00:38:55:16
Vasu
We know we want we are helping customers aggressively and by storing all the data. But we cannot be closed minded and saying, oh, we're going to monetize this data, and I'm going to put a whole bunch of roadblocks with that. I think that kind of slows down the progress. Instead, we are building and shaping server into a product where any of these agents or any of these partners can actually query what's going on.

00:38:55:18 - 00:39:08:06
Vasu
We will proactively go load up their catalogs, like in Databricks, for example, or snowflake and Palantir will load up those catalogs good knowledge of what we have. And so it just makes it easier for customers to.

00:39:08:06 - 00:39:09:09
Speaker 4
Access.

00:39:09:11 - 00:39:13:19
Vasu
And, you know, get going with this whole evolution.

00:39:13:21 - 00:39:14:04
Speaker 4
You know.

00:39:14:04 - 00:39:29:11
Chris
You have all this, this data and you're doing all these analytics on it now. I mean, do you vision maybe a future where you could. Yeah. Because one of the, one of the big challenges with AI is like how credible is this answer. And starting to like pin credibility, you know, because it's going to be a sliding scale.

00:39:29:11 - 00:39:32:09
Chris
I imagine when you talk about security, like this is kind of not.

00:39:32:09 - 00:39:33:12
Speaker 4
Secure, but it could.

00:39:33:12 - 00:39:57:02
Chris
Be, you know, like this is very not secure. This is okay, you know, like but but I mean that but that's a big problem for everything in terms of like responses and things like that. So if you've got if you're starting to see like the credibility of models or of agents sort of declining in your kind of evaluation of the security challenges of that, that maybe something you could feed back into to the whole ecosystem.

00:39:57:02 - 00:39:58:05
Chris
Two right.

00:39:58:07 - 00:40:18:21
Vasu
Right now, where I expect much of this use of this edge attack is going to be where it's possible to verify. Right? So it's a problem of there are there are things that you can verify easily and things you can't verify easily. Right. For example, an agent creates a support ticket is impossible to verify that it's an actual real customer.

00:40:18:23 - 00:40:37:20
Vasu
It's probably possible to verify they have a licenses there, and there's probably a valid input, that came in. So there are things that you can verify. And I expect most of those real action oriented, work oriented towards, to be verifiable to start with. But then you go to this knowledge work and like what we do day to day.

00:40:37:20 - 00:40:49:06
Vasu
Yeah. It's like, you know, I cannot say this no impact, but then it's a personal responsibility that kind of it's like the Tesla FSD right now. I mean, it'll do everything for you, but you're still liable.

00:40:49:07 - 00:40:51:14
Speaker 4
Right? Yeah. Yeah. You know, if.

00:40:51:16 - 00:40:53:11
Vasu
You're your insurance and and,

00:40:53:14 - 00:40:55:02
Speaker 4
Probably you have a.

00:40:55:02 - 00:41:19:15
Vasu
Duty if it makes a mistake. Right. And so we are kind of in that mode right now. But when you have an automated system and if I running a business anyway, if an agent is going to take, take orders, I better have a good verification system and so there is no really room for creativity there. Right? So the you know, I expect most of those things to be there, but detecting when something goes wrong, I think is more important.

00:41:19:17 - 00:41:38:11
Vasu
That's about like ambiguity of whether that answer is correct or not. But I mean, there are places where it is going to make a call. And this is where I'm like, I'm on the side of your dumb audience. I'm like, check it out. Where enterprises will start letting those those like making decisions of what kind of decisions.

00:41:38:13 - 00:41:50:17
Vasu
Yeah. Yeah. So there's a lot to be done there. And definitely I'm not the expert on figuring out how to, Yeah, judge of an agent and things like you're not that today.

00:41:50:19 - 00:41:52:20
Chris
So that's a big problem to solve.

00:41:52:20 - 00:41:53:22
Sandesh
Huge problem.

00:41:54:00 - 00:42:11:17
Vasu
It's a huge problem. And we use open benchmarks right now that are benchmarks. The problem is benchmarks only do certain things. Like if you have an outside of the benchmark, you don't know and you know in any of them today might miss score higher. But if you challenge it, they say, are you sure? Or maybe you're wrong. It probably will backtrack.

00:42:11:17 - 00:42:18:11
Vasu
And so I don't think the I don't know how much the benchmark to be able to catch such, you know what flopping behavior. Yeah.

00:42:18:13 - 00:42:40:14
Sandesh
Yeah. I love that comment you made about Tesla, by the way, because, I have a Tesla and I put on FSD and as you would say, there's shenanigans. It's, it's not perfect. You know, like there are times where, like, I've been with my son and we're both thinking like, does it see that there's a train coming?

00:42:40:14 - 00:42:46:02
Sandesh
Does it see it? It sees it, it sees it, it sees the train. It's. We're getting closer and closer.

00:42:46:02 - 00:42:47:12
Chris
We get to that barrier, right?

00:42:47:17 - 00:42:50:18
Speaker 4
Right. Yeah, yeah. It's crazy.

00:42:50:20 - 00:43:09:09
Vasu
That January I started subscribing to FSD. I was a big skeptic. Look, I converted, because it took so much train out of my commute. At the same time, I'm actually watching it like a hype, like a. No, it's not like a I'm looking elsewhere, but right. Cognitively, it's easier to just look at it and make sure it's doing right.

00:43:09:09 - 00:43:12:11
Vasu
It's like watching your son drive or something like that.

00:43:12:13 - 00:43:17:10
Speaker 4
Right? Yeah, fine. By the way, I just got my son just got his, his, his.

00:43:17:10 - 00:43:19:17
Chris
License this month, so, you know.

00:43:19:18 - 00:43:20:04
Speaker 4
Yeah.

00:43:20:05 - 00:43:24:04
Chris
No, Camden. I'm not. I'm kidding. I trust your driving.

00:43:24:06 - 00:43:26:14
Speaker 4
Yeah, yeah, I but.

00:43:26:14 - 00:43:48:19
Vasu
It is stressful than having to navigate the traffic and stop and go traffic. Especially now it's become really tough in the Bay area. So I think it's worth that $99 a month for that piece where I can talk to people and, you know, while watching what it's doing, it's taking the stress off. And then so I still want it to be unsupervised, where, you know, it'll pay for the mistakes it makes.

00:43:48:21 - 00:43:55:20
Vasu
Right now, I'm kind of okay. They're going to be in the middle where I pay for its mistakes, but I. I still pay for it.

00:43:55:22 - 00:43:56:14
Speaker 4


00:43:56:16 - 00:43:58:15
Vasu
Take some some strain off.

00:43:58:17 - 00:44:18:06
Chris
Well, it's, you know, it's, it's like when the elevator was introduced, you know, they used to have an elevator man in the in there, and you take you up and take you down, and you just told them the floor, and then they started putting, you know, the buttons, and you could push through your floor. And they still kept the guy in there because everybody was like, I don't want to get stuck in a box, you know?

00:44:18:08 - 00:44:23:13
Chris
And then when they got rid of the guy, they had to put that big red stop button in there. That was the, you know, the thing. So like.

00:44:23:15 - 00:44:25:08
Vasu
But, you know.

00:44:25:10 - 00:44:47:15
Chris
In some ways, like I will probably see that autonomy. And, you know, I was operating at a higher level of safety that most people do, you know, but it's it's like it's still there's that like you say, that discomfort factor where you got to keep your eyes on on everything. And I think a high is very much like that across the board.

00:44:47:17 - 00:44:47:23
Speaker 4
Yeah.

00:44:47:23 - 00:45:04:10
Vasu
Well, I mean, if you look at commercial airplanes that fly themselves most of the time. Yeah. Right. And and the real problem is when they get up and it's a startle effect letting fly pilots have had a tough time figuring out what to do. And it gives up because it's the worst time for somebody to take control out.

00:45:04:10 - 00:45:20:02
Vasu
Right? Yeah. And so yeah, it's a slippery slope. And, let's see how we figure it out. Let it clear out. I mean, my son, I watch a lot of, airplane, accidents on YouTube. You know, I know.

00:45:20:03 - 00:45:22:13
Chris
The challenge you're watching. I watch those, too.

00:45:22:17 - 00:45:24:16
Speaker 4
Yeah, I know I in.

00:45:24:18 - 00:45:31:22
Vasu
The at the worst time it tells you to take over is really a bad time for a human being to jump in and figure out what's going on.

00:45:32:00 - 00:45:32:13
Sandesh
Yeah.

00:45:32:15 - 00:45:33:04
Vasu
And, and.

00:45:33:06 - 00:45:45:04
Chris
And quite frankly, a lot of the accidents that, you know, you see, are because the humans didn't trust the automated systems, took it off the automated systems and then crashed. Yes. That's a lot of, a lot of that, that too.

00:45:45:06 - 00:45:56:14
Vasu
There are times when the automation gives up. It's very hard to figure out sometimes. Automation too is perfectly fine, but then it's hard for humans to trust. Yeah, yeah. This is the human condition, isn't that?

00:45:56:16 - 00:45:57:10
Speaker 4
Well, I.

00:45:57:11 - 00:46:21:18
Chris
Like to say it's it's, we we adapt to. It's much slower than AI evolves. Yeah, like AI gets better. I mean, like, you know, just talking about the, you know, January to to March or not even March. February. I says one month, you know, like how much it's changed you know, like it it's growing so fast is that sort of Malthusian curve, you know, like, yeah, it's growing exponentially and our ability to adapt is linear.

00:46:21:20 - 00:46:43:13
Vasu
What, what I've not really gotten over, is the intention behind it? Yeah. Right. And so alignment is right. Right. And I mean, with us. Yeah. So when people make a machine like, you know, in this case, I know the engineers are human beings and they want human beings who live. I don't think they're. And so they're going to take an extra measure of safety.

00:46:43:13 - 00:47:07:23
Vasu
Every decision is going to be thought through because we trust human beings to kind of take care of each other. But an AI takes over. Is the intention the same? It's definitely not intrinsic. It is being taught something. But what is the real intention? I don't know. And so I really especially for those critical things that I'm right now, I'm not sold out that kind of there's that agents do the stuff and we just take the outcome of it.

00:47:07:23 - 00:47:28:21
Vasu
I think to be able to verify that it matters. It's going to be critical. And that's why I think this is going to be very relevant. And in those areas where things that like, you know, low cost and thumbs up are going lower, in fact, I think it may maybe. Okay. Thanks for, like, visit this story that's been going around and acts like radiologist.

00:47:28:21 - 00:47:43:07
Vasu
Right. So more, you know, more AI and there are more radiologists being hired because you want the radiologist to make the automation and tell me what to do. I will help them do that. But I don't know. I don't know if that's shared, but why.

00:47:43:08 - 00:47:58:10
Chris
It's sort of like how how long lived is that, you know, is that a temporary solution? I mean, you know, we we've we saw that I forget what the name of it, but there's, there's that new website where it's like, hey, you can be, human to help AI, you know, act in the.

00:47:58:10 - 00:48:00:17
Speaker 4
Physical world so you can you.

00:48:00:17 - 00:48:12:20
Chris
Can job. It's like a gig economy thing. Yeah, yeah, it's like, that's wild. Now. I mean, I guess we'll have, you know, like, our monkey hands come out and do things for the AI. And that's what our employment future looks like.

00:48:12:22 - 00:48:17:05
Vasu
The Tesla robot or another robot comes in to do that cartoon.

00:48:17:07 - 00:48:20:22
Chris
Yeah, right. And robotics is having an explosion too.

00:48:20:22 - 00:48:23:20
Vasu
Yeah. It's resurgence. Yeah. Huge huge explosion.

00:48:23:22 - 00:48:24:08
Chris
This. Wow.

00:48:24:08 - 00:48:30:05
Sandesh
Boy, have we got no way from Cohesity, haven't we? But have we? I love it.

00:48:30:07 - 00:48:33:11
Speaker 4
But have we? Right. Because each one of those.

00:48:33:11 - 00:48:55:01
Vasu
Generates data, and it's a robotics factory with a robot centipede recipe. And I feel like they would need a way to be restoring them. And so I feel like Cohesity has a future as a one that provides resilience layer to AI agent tech. Yeah, robots are software. It's robots. Are they just they have hands, but they are run by software, right?

00:48:55:01 - 00:49:03:09
Vasu
And people have created artificial human looking figures for a long time. But it's a software. That's the difference between that and the robot we're talking about.

00:49:03:11 - 00:49:21:06
Chris
Well, and there's a lot of telemetry from these platforms, especially ones that are going to run on the edge like cars and robots and things like that. And, and that's a, that's that becomes a really difficult challenge because like, if you're talking about multiple copies of all that telemetry data, that is probably going to be too heavy. Yes.

00:49:21:08 - 00:49:26:11
Chris
You know, for for what you need to do. So like how do you manage that? You know, like in protecting that.

00:49:26:11 - 00:49:46:01
Vasu
And again, coming back to that's what considerations. Right. So you can put massive amounts of data in the store very efficiently. But it can get back to any piece of the data pretty fast. And and you can do analysis and tagging cataloging on that. So you know exactly what's where. And that can be across the cloud. That can be across on premises or SaaS applications.

00:49:46:02 - 00:49:58:05
Vasu
I mean, we are pretty excited about the AI because it it's because of all of this, because all of these competitors who doesn't need resilience, all of us need incidents. Yeah, of course, of course. The agents and I need resilience.

00:49:58:07 - 00:50:25:09
Sandesh
So I, I want to double click into what you said earlier about it. Tell me if I'm wrong. You're building an ecosystem of partners that will ultimately help Cohesity better protect and serve. Your clients. Yeah. Which you're terming AI resilience. Can you help? Like in plain English? Just help. Like if there's a customer that's going to be watching this, like what does that mean?

00:50:25:11 - 00:50:27:19
Sandesh
What does that mean to them?

00:50:27:21 - 00:50:59:00
Vasu
Yeah. So a customer needs to run businesses with the AI and they need to be they need to be able to do all that we talked about, which is to monitor AI, to have a platform to run that AI. They need to be able to detect errors, then be able to correct errors in a governance. Them like who has access to what some of the fundamental problems on the business and our goal is in each of these areas to work with the best partners.

00:50:59:02 - 00:51:22:18
Vasu
For example, use detecting what permissions that are available for each agent and what tooling it can access. ServiceNow is figuring out what are the various agents that can access and what kind of workshops are possible. They helping create agents to validate a workflow outcome in ServiceNow by Datadog is figuring out that agents are functioning properly and they're getting the right performance characteristics.

00:51:22:20 - 00:51:48:03
Vasu
Google is building in our tools that do similar things. Each Microsoft agent 365 actually monitors all actions on top of email or SharePoint or other assets, and figuring out which agent can access what, permissions. And so this is the kind of ecosystem partners. And when anything's failed, we can actually help them recover. We can also help them to investigate what went wrong.

00:51:48:03 - 00:52:10:03
Vasu
When something goes wrong, and to be able to bring back, you know, what they actually needed in the first place, right? So that's why it's very important for us to be working with all these partners that are helping our customers. Know you entering the data site, gather the data to analyze the data, to build agents, to be able to operate agents, manage, monitor them, govern them.

00:52:10:05 - 00:52:11:20
Vasu
And that's where that's where we flood.

00:52:11:20 - 00:52:39:23
Sandesh
So I think that's a great idea, because what I'm seeing in the field is time to value can be very, very challenging. And so many customers are going to say to me, I got tools for this and that and that. I just got way too many tools. Right. But what you're saying is like, let me connect into what you have and leverage the value that you're already getting from people that already are doing a great job with you for this thing.

00:52:40:00 - 00:52:54:13
Sandesh
And if I'm understanding it correctly, Cohesity provides a one pane of glass or one control center where now you can get the whole tech ecosystem connected into Cohesity. Is that accurate?

00:52:54:13 - 00:53:00:08
Vasu
Yeah. Also, it doesn't always have to be, a user interface. It could be an.

00:53:00:08 - 00:53:02:02
Chris
Agent interface.

00:53:02:04 - 00:53:23:12
Vasu
To which you can access the whole thing. So yes, we always will have human beings looking at the console and looking at what's going on and make sure they have completed necessarily, and that they can recover and operate. In the face of any, you know, accidents or disasters or attacks. The same time, many of these functions are being performed by agents, too.

00:53:23:12 - 00:53:29:22
Vasu
So, yeah, I will provide it. We are providing Vincent the interface so they can do that in a secure way as well.

00:53:30:00 - 00:53:49:23
Chris
Do you think, you know, one of the challenges is like, as we go forward, I mean, because we saw this big dip in the stock market the other day because, you know, was it 200, 200 lines of, markdown that, you know, like, made up this whole legal and now it's like, oh, so that that industry just went away, right.

00:53:50:01 - 00:54:12:11
Chris
You know, like it's it's it looking like there will be, ways for people to just build agents on the fly. Agents have built agents on the fly to solve business problems, which kind of undercuts a little bit that the software industry, which is, I think what that that DEP was, was looking at. Right. How do you.

00:54:12:15 - 00:54:12:22
Speaker 4
How do.

00:54:12:22 - 00:54:30:05
Chris
You protect a world where things come into existence, you know, that are core critical items for a company, and you have to keep track of that, you know, like do all that kind of discovery, too. And what do you envision that future looks like?

00:54:30:07 - 00:54:50:02
Vasu
Yeah. So it's it's it's basically a, again, a generic way to provide resilience, which is we need to be able to constantly discover what's going on. A part of this is already happening in the cloud. Previously in the data center, you knew where the vCenter was. We knew where the databases were being, and they would just configure our customers.

00:54:50:02 - 00:55:12:02
Vasu
We just use those for just configure these things and say you pick a backup, but if you're in the cloud, it doesn't happen like that. There are multiple people that are joining mean creating, new data data entities. Right. So they creating S3 buckets, they create new applications. One of our largest customers staying in the cloud for them.

00:55:12:02 - 00:55:31:21
Vasu
There is not going to be any control. They don't have to get any permissions before they go. Create a new application for anybody can create applications. Then we are responsible for protecting them. So the automatic data discovery. So it's already happening. The future of the enterprises, we automatically discover what's happening and be able to protect it and provide resilience.

00:55:31:23 - 00:55:52:18
Vasu
We don't to wait for somebody to configure to tell us something that's going on a new. And that also helps us verify is everything being protected? You know, our people, I are being governed or not. All of those are being, you know, automated. Right. So yeah, the future is automation is already happening right now. And in this AI resilience.

00:55:52:20 - 00:56:02:01
Vasu
Yeah. Nobody was going to tell us. Yeah. He doesn't nature and his this is what you got to do. It's going to be that we are constantly discovering and figuring out what to do.

00:56:02:03 - 00:56:14:21
Chris
And I would imagine you're going to apply some intelligence around that to say, hey, I just detected this agent that kind of emerged. Is this a security threat or is this something you need to back up because it's really important kind of thing, right?

00:56:14:21 - 00:56:34:04
Vasu
Right. Correct. Or this is redundant or this should not be there. Yeah. For example, today with our partnership with Ciara, once you detect that are some German driver's license in a file and automatically put in a bill that you cannot restore it to, non-German are not you entity is on that exact same figure.

00:56:34:06 - 00:56:53:06
Chris
Yeah. For GDPR this has got to be a huge boon right. And for for things like I mean, we've always talked about, you know, leaking data and exfiltration and things like that and, you know, just data provenance issues and things like that. I mean, this is this is like taking that to a like a whole nother level at that point, right?

00:56:53:07 - 00:57:06:13
Vasu
Yes. Because the way data can be manipulated and it can be free. And what actions it can take is so much more. Richard. Imagine taking, you know, app environment.

00:57:06:14 - 00:57:14:19
Sandesh
Yeah. So, we've covered a lot. I think we're, I can talk about this for serious. This is a free world.

00:57:14:19 - 00:57:25:02
Chris
And, I mean, and and it is super important that, you know, like, I know, you know, companies are out there trying to get ahead of this problem because I think there's like I said, it's moving so fast.

00:57:25:02 - 00:57:25:19
Sandesh
So fast.

00:57:25:19 - 00:57:30:03
Vasu
Moving so fast, and who don't know who and when and who it lose, I mean. Right. Yeah.

00:57:30:05 - 00:57:42:05
Chris
Which makes it really hard for folks like you because, like, what do you develop towards. What do you you know, who do you partner with, who you, you know, like that's, you know, I, I don't envy your position. I that's really tricky world you're operating in.

00:57:42:07 - 00:57:59:09
Sandesh
I really, really, really appreciate the building of that tech ecosystem because I think partnerships and relationships are going to be so important in the future. You know, sometimes we get so siloed in our this is my company, this is what we do. And we're just going to do this and we're going to sell this, you know, like yeah, if we can.

00:57:59:11 - 00:58:26:20
Sandesh
You know, you mentioned the human like the human side of it. That is very personal. I know to Chris. And for me, definitely, I really worry about that. Like I love technology, I love it, but there's also unintended consequences of technology. Yeah. You know, I always talk about well, the iPhone right. There was social media. But then the iPhone created this whole mental health issue, which we didn't really see coming.

00:58:26:22 - 00:58:40:18
Sandesh
So. So what is AI going to bring to us? If you know there's agents talking to each other instead of humans? And this is, hey, we're on this earth, you know, without getting too philosophical, but like, yeah, we're here to be happy.

00:58:40:18 - 00:58:43:10
Chris
Discussion that doesn't end philosophically, so just go.

00:58:43:10 - 00:59:04:19
Sandesh
Yeah, yeah. But I think to me just comes down to humanity. You know, we're we're here on this earth, you know, for a reason. And what makes it and we want to be happy. What makes us happy is humans, right. Because we other humans and our relationships and everything else. But I worry like I don't want AI to take away that secret sauce of ours.

00:59:04:19 - 00:59:29:05
Sandesh
Yes. You know, it's it's going to have an impact. But we as a society, we there's we can control this, you know, we we need to. And I think this is why it's so important that we do talk about this. You know, I have three kids. You know, I'm so worried about them. You know, I'm just so worried not about how much money they're going to make or their careers or what they major in or whatever.

00:59:29:09 - 00:59:41:07
Sandesh
It's like, just don't forget to be human, you know, like, just make sure you care about the relationships and the people in your life, you know, because that that's going to give you the happiness, you know.

00:59:41:07 - 00:59:43:06
Chris
And learn to prompt engineer.

00:59:43:08 - 00:59:46:04
Speaker 4
Yeah. Product engineering so that they're, you know, they would.

00:59:46:04 - 01:00:04:20
Vasu
Topia they would topia science. Big science fiction fact. And then there is answers for all of these. We don't know if it will be true or not like you. What Star Trek could say. There will be a time where there's no hunger and nobody needs to work just because they have to. But they can do what they want to, because they're curious.

01:00:04:21 - 01:00:27:00
Vasu
And keeping that, clarity of the purpose is probably the most important thing. Like, what if you have enough to live, what would you do? Right? I think that's where the currently the education system is not focusing on the sorry but like agree. Great. Something about that because so much of education system is teaching people how to do stuff, but then how to do it's easy.

01:00:27:02 - 01:00:48:20
Vasu
It's about what to do and why. Because what important and to me the education system. I know I've been following and I struggled with both of my sons. The why is almost never there. It's like less than 5% of the education system is why. Yeah, 95 is what. But the what is the least important when we actually know why?

01:00:48:22 - 01:01:05:19
Vasu
And you know, what is the meaning of life? And you know, well, what we could do if resources were not at the limit. I think we will we can we can live there more fulfilling lives. But today, without having a busy being busy, a lot of people would struggle to have a fulfilling life because we don't know what to them.

01:01:05:20 - 01:01:16:15
Vasu
It's a creativity, right? If the energy was free, if the resources are free, what we should do, I think to do a lot more. But I'm hopeful. Yeah, but the transition, I don't know how the transition is going to be.

01:01:16:15 - 01:01:17:07
Chris
It's going to be wrong.

01:01:17:07 - 01:01:23:15
Vasu
But yeah, if I am, energy is going to be. Yeah. But anyway, it's going to be a pattern. It's definitely going to make everything cheaper.

01:01:23:16 - 01:01:42:22
Chris
Well I do think we're at a kind of crossroads right now. And yeah, we can choose to build for ourselves a heaven, you know, that that that scarcity free economy where we can pursue our own interests or we're going to build our our hellscape where we've got this new robber baron world of, you know, oligarchs owning and operating all the machines.

01:01:42:23 - 01:01:46:02
Chris
And, you know, you know, where do we fit into that?

01:01:46:04 - 01:01:50:14
Vasu
You know? Yeah. But I mean, without like, there is no money to be made if everybody didn't have any money.

01:01:50:14 - 01:01:54:01
Speaker 4
So it has to it has to sustain somehow.

01:01:54:03 - 01:02:14:07
Vasu
Right. And so I think even the robber barons ultimately will need the rest of the folks to make them robber barons and then to rob. But, I mean, I, I'm not getting to that side, but I feel like even at the midpoint, I think it's going to create more abundance than not only thing. Yeah, I do completely.

01:02:14:07 - 01:02:26:02
Vasu
I'm with you on how the transition is going to be, because there is going to be a transition where we all are adjusting and settling with a new reality. And, I don't know, it's going to be five years, ten years, 15 years, whatever. Yeah.

01:02:26:02 - 01:02:31:15
Chris
I'm not it's going to be shorter than you think because all I've learned for me, I.

01:02:31:16 - 01:02:32:16
Speaker 4
I hope so for the.

01:02:32:17 - 01:02:35:21
Vasu
Even though topia, it's I hope that now.

01:02:35:23 - 01:02:43:07
Chris
I, I'm optimistic about I, I mean it's from some of the things I've seen and you know some of the things I'm just I have to be optimistic, you know. Yeah.

01:02:43:12 - 01:03:06:16
Sandesh
But you got to get in there. You know, I think, the amount of people, the stat that I saw was, you know, I think, OpenAI AI or ChatGPT, whatever, they have 809 hundred, you know, people using the platform, but less than 10% actually subscribe to the $20 a month, you know, subscription. And to me, it's like, wow, you, you know, and whether it's clod or grok, I don't care.

01:03:06:18 - 01:03:09:22
Sandesh
Get in there, start. You have to, you know, and.

01:03:09:22 - 01:03:12:20
Vasu
We cannot ignore it. Like we cannot ignore it.

01:03:12:22 - 01:03:34:05
Sandesh
Of there's so many people ignoring it. You know, this is where I think we're in this time of this world where it. I honestly think it's so easy to be great. These days. You just gotta try. Like, if you try and you go learn what you need to learn, you know, the the world is going to open up for you in many, many ways.

01:03:34:05 - 01:03:51:19
Sandesh
But I think so many people are just like, I'm just going to wait until, you know, this happens or I'll see what my company does with AI or I'll, you know, I'll figure it out, you know, or I'm only going to be I hear this one all the time. Hey, I just need five more years in this business, and then I'm out of here, you know?

01:03:51:21 - 01:03:52:21
Speaker 4
And it's like.

01:03:52:23 - 01:03:59:21
Sandesh
You know, like, okay, well, you you do have children, and you do have, you know, others that you love and care for. You're not gonna.

01:03:59:23 - 01:04:01:11
Chris
Die in five years. So you're going to do it.

01:04:01:11 - 01:04:17:00
Sandesh
For that set. I'll do it for them and do it for yourself. Yeah. So before we wrap it up and I just give you a chance to, give us a last message. Is there anything that you want to tell us that you shouldn't?

01:04:17:02 - 01:04:27:11
Speaker 4
I wouldn't, but that's the right answer, all right? I had to try. I at least had to try, you know, and.

01:04:27:13 - 01:04:49:14
Vasu
I'd say by this time this airs, we would have had our user conference out there. And I would definitely urge people to go check out what you mean by resilience and what you're planning to do. And there and I think a lot of exciting stuff coming from Felicity and I, I couldn't be more excited to, you know, be in the thick of it and providing infrastructure and support for the new, brave new world of AI.

01:04:49:16 - 01:04:52:04
Chris
I can't wait to see what you announce.

01:04:52:06 - 01:05:09:00
Sandesh
It's amazing. Vasu, thank you so much. It's been so great to meet you and. And, I feel like you got to be a repeat offender here. We got, we got to keep this going. Really sincerely appreciate your time. I know you're a busy guy, and, wish you all the luck.

01:05:09:01 - 01:05:13:03
Chris
Yeah, we just got to have these fellas official AI conversations all the time. I love it.

01:05:13:05 - 01:05:13:17
Sandesh
Love it.

01:05:13:17 - 01:05:33:21
Vasu
I can spend the whole day doing that, and I enjoy it. There's something that is big about real, real serious interest of mine, because taking care to the point of everybody had a good grounding of why we do what we do. Yeah, not just babysit. What the how a part of it. And yeah, I have grown a lot in the last few years, you know, but it's.

01:05:33:22 - 01:05:38:18
Chris
Such a good observation about that. The why piece of it. I so agree with what you said there.

01:05:38:20 - 01:05:56:16
Sandesh
Yeah. For sure. And that's why I, like I tell you so Chris has multiple podcasts. But one of the things that I love about doing this is I get to meet people like you, and then I get to have these kinds of conversations. You know, I love meeting Chris, and I have this little tagline is we want to do cool shit with cool people.

01:05:56:18 - 01:06:06:19
Sandesh
And like, this is cool shit. Like these conversations are no doubt interesting, right? And and when will the Future people club.

01:06:06:21 - 01:06:08:05
Speaker 4
Yeah. Yeah.

01:06:08:07 - 01:06:14:12
Sandesh
And and when and when I don't know, I don't know if, if that that's really true. That we're.

01:06:14:12 - 01:06:16:04
Speaker 4
Cool. No.

01:06:16:06 - 01:06:18:07
Sandesh
We're good at what I cool. But, but though I guess we.

01:06:18:09 - 01:06:19:18
Chris
We talk to all people.

01:06:19:20 - 01:06:20:21
Speaker 4
Yeah, but I think.

01:06:20:21 - 01:06:42:20
Sandesh
It's really interesting when the future is unknown. You know, I get very excited about the uncertainty and the unknown. It's, you know, of course there's going to be risk and everything, but I love debating things and learning from other people because people change my mind constantly. One week I'm thinking a certain way. Next week I talk to Chris and he sends me.

01:06:42:22 - 01:06:57:11
Sandesh
He sends me a lot of stuff, and then I'll read it and I'll be like, oh man. So maybe, maybe I'm completely wrong, you know? And what sucks is I'm here creating content and expressing my opinions, and I'm going to be wrong.

01:06:57:13 - 01:06:58:22
Speaker 4
You know, like this.

01:06:58:22 - 01:07:07:07
Sandesh
Just like that's an inevitability. And it's good to be out there, but that's that's just part of it, you know? I'd rather go and try and be wrong.

01:07:07:08 - 01:07:17:17
Vasu
That's how I get better. Like they make mistakes. Yeah, that's how I make get better, but made mistakes and figuring out absolutely wrong. I mean, you guys are so fun to chat with it. I enjoy it, but, Yeah.

01:07:17:21 - 01:07:18:12
Speaker 4
Well.

01:07:18:13 - 01:07:19:06
Chris
Thanks so much.

01:07:19:10 - 01:07:23:01
Sandesh
Thank you so much. You're all the best of us, too, and we'll talk to you soon.

01:07:23:03 - 01:07:24:18
Vasu
Thanks. Donation question for sure. It.