In Good Company with Nicolai Tangen - Snowflake के CEO: AI एजेंट कैसे बदल देंगे काम करने का तरीका (Hindi version)

Episode Date: June 22, 2026

निकोलाई टैंगन बात करते हैं Snowflake के CEO श्रीधर रामास्वामी से। Snowflake एक डेटा प्लेटफ़ॉर्म है जो दुनिया की आधी सबसे बड़ी कंपनियों को चलाता है। इस बातचीत में वे समझते हैं कि डेटा और AI की दुनिया में असल में क्या हो रहा है।दोनों इन बातों पर गहराई से चर्चा करते हैं: Snowflake सिर्फ़ इस्तेमाल के हिसाब से पैसे लेता है (कंजम्पशन-बेस्ड प्राइसिंग), और यही बात इसे आम सॉफ़्टवेयर कंपनियों से अलग बनाती है। श्रीधर अब मानते हैं कि AI मॉडल बनाने वाली कंपनियाँ टेक इंडस्ट्री में सबसे बड़ा ख़तरा हैं—बाक़ी किसी से भी ज़्यादा। और AI एजेंट अब हर चीज़ बदल रहे हैं—डेटा पाइपलाइन से लेकर सॉफ़्टवेयर इंजीनियरिंग तक।श्रीधर अपनी कंपनी Neva के बारे में भी बात करते हैं—उसे शुरू करना और फिर उसका नाकाम होना, और इससे उन्होंने क्या सीखा। साथ ही वे उन मूल्यों के बारे में बताते हैं जिन्होंने उन्हें तमिलनाडु से टेक इंडस्ट्री के शिखर तक पहुँचाया: मेहनत, हर हाल में ढल जाना, और हार न मानना।यह दिलचस्प बातचीत ज़रूर सुनें!——Snowflake CEO: How AI Agents Will Transform the Workplace Nicolai Tangen sits down with Sridhar Ramaswamy, CEO of Snowflake, the data platform powering half the world's largest companies, to explore what's really happening at the frontier of data and AI. They dig into how Snowflake's consumption-based pricing sets it apart from traditional software models, why Sridhar now considers AI model companies a bigger competitive threat than anyone else in tech, and how AI agents are transforming everything from data pipelines to software engineering itself. Sridhar also reflects on the lessons learned from founding and failing with Neva, and shares the values of hard work, adaptability, and resilience that have shaped him from Tamil Nadu to the top of the tech industry. Tune in for an insightful conversation! Hosted on Acast. See acast.com/privacy for more information.

Hi, friends, I am Nikolai Tengen, Norwegian Sovereign Wealth Fund's CEO. And today, I'm with Shredar Ramoswami, who Snowflake's CEO is. Snowflake is a data platform where, where the world's many big companies up to work work. So, when your bank has a lot of proofed or any hospital patient data to one-as-lake is working on. Shretharne in Google in 15-sallel years before, where they're advertising business to $1.1 billion to builda and $1 billion to

up on. Then they've got to start to Google. two years back, they're so that Snowflake's CEO been able to. Now, here NBIM in the snowflake's investors be

and it's a product's a big amount of using. It's the side. Snowflac on there's a

pretty two pettabytes data is that that's about 20 lakh gigabytes data to be and we're every day

database in there's a lot of 30 lakh queries so you. So, you're very welcome

here. Thank you. Here's here came it's a I'm very excited. First, you can't how can't

how can't how much who's not in short? Yeah, so we're a data platform we're a a cloud computing platform

like AWS but our focus on data we're we're allagged data to analysis, it's analysis,

and then then those systems to take you where you're a analytic data platform platform have

your clients who are? Oh, so um, so, um, so global 2000's

close adhh accessible companies, yeah, China's out of course are our customers are. It's in

financial services, health care, advertising, and many other industries in our several customers are. So,

we can't say, that this list is a lot and we, and we, uh, we, we're 25 to

do things in different ways, but our customers this even a lot many, many-seye in the most of the

country's started. When we're going to separate from particular. It's got to let's the about us.

Absolutely. This very important concept is the most easy is the easy is the way, you think, we've always

computers how to use or used to have. I say, I say that I've been past 10 years

when any company had been any company That box in storage, compute, and memory of a fixed matter, which was fixed were fixed, and, and, you'd have to be the box with

five years to work to work to work. So, if you'd have got a more memory would have been, then it could be just in a lot of if you'd have got a CPU power then, then, it would have because you'd have already

had already, you'd have already your phone is a box-y-as-the-chot-box is, and a little box to a box and a box to other box to take a data to be a quite much much more. for it was a different integration project to be it. So, here he's the Snowflake

made use of this poor thought of this poor thought you can't you can't even storage and the robust compute power systems as much which can't

share in we can't share we, we've made compute and storage to each other other's different we're allocked. We've said that

you need, and just any power should be, if you need, if you need a if you're a deep-analetic job can't Because NBIM's

in the business for an investment for a lot of idea, which you're you're going to do you need to a weekend in a

hour, analysis, can't, and then, you can't know, and then, how to apply to apply to do.

This is Snowflack model's magic. Jaddu. Yeah, most software companies per seat charge but you you're not the

people use. This is the way is. Because it's the pricing seeded value creation with aligned It's a good thing is.

This is a good that we're a demand to make a because we have we have $13,000 so we

we're we can't we can't and you and you can a more stable and better model this is the

this way it's a you have to you have to you have got you're you're going to use and how

you're going to you're going to you're the when they need they need. Zadda-tar-a-triks' not-haping. As you know,

transactional workloads in the day and night's between a lot of banking transactions not are not so many things. So,

there's many patterns are that first from the first-a-of-a- priceing. Because we're we're very high high-faitty pricing

because we can't we can't manage and amortize can't. This is our company pretty-to-value-reation with aligned you're at least

payment but we we're revenue that he recognizes when you have when you and your team Snowflake's we're going. We're going to

this part are the AI's in a lot because you're using as a pay-as-as- pay-to- pay-to- Why?

Shreder, now, why you know, because they're they're going to whole thing. Look, my mind that's the software industry

in the very high-of-a- work-one- for us-mere- people for all- here, now this thing all-to-saf-to-

soft-to-to-to-to- that software's cost economics is very quickly is a lot of and if I'm I'm asking, I'm I'm going to I'm at afti-mort's

like a need to take away from and the fact that the fact of the software is to a quite a kind of craftsmanship he has. Some people

be a hard-mahed software binae it's a lot of different systems with integrate and you're the best best programmers you can't

as a concert pianist like, as a lot of many things are a lot of a lot of a lot of you're in a

thing you're a lot of your then you're a great musician and then you're still a lot of still that's still that's

but that the software the models what they represent is the software development as industrialization as we've

never never never seen. And so if if it's the big reason is that I'm I'm the model

companies and their building agents to my sub-the-up competition so, in all looking at

we've been made a whole coding-asement system front-door are going to go and the

short-of-the- software engineers but it's the effect it's the impact of going to but what?

But Amazon and Microsoft Jetson, who have competing products have? Yeah, no offense.

But, when the thing about pure data platforms comes, we always top of the

charts have, and looky, the companies that companies operate operate are you never underestimate

not to underestimate not really because they have much as difficult problems to solve to solve to solve for many people

can't be really, but I'm looking at I'm looking at this is this thing that this is

the main the main and in a way to a new thing a new thing that they're not survive

but thrive be, this is my number one challenge is. Okay, so AI's back to

a way to make the how many how do you? Um, okay, so the part is if I'm

AI and Snowflake like a software company, so basically our core on two things are we're doing, we're a

great software that's which my engineering team's and we're and we're you like customers to our software

bakesh and we're doing it so much more so much power focused to be this part it's fast

how to make. Sales side per our sellers Now, now, these tools who are information instant access

they're in the phone a sales agent are used to be our solution engineers just in 30 minutes in Nikolai for a customized demo can make,

which will be your bank's like it's like you can't be a good results to their ability their ability really shan-dard-of-

and software engineering if you're talking, so coding is pretty different I'm just, I'm just a meeting

from here we're talking about we're talking how we're how we're more more people to speak-driven coding development

in the event in the event in which you want to produce to want to, that's the way you're just an English language spec and then the process you automate like code's first version to write, and it's the same the other things. This is the software engineering

today is the way the same year ago had been the first year. We're the most big challenge is that our team's engineer this type of work to seek and these new possibilities to put upna and their fide. But we've got these superstars

which are which a average software engineer tolame 50 to 100 than a lot productive are. Okay,

so my other question is the AI's coming how much how much different different.

Two things. First, AI, AI, you're just accessible bansans. Snowflak

intelligence is this is the kind of agentic interface is, where you're you can't ask,

how you're doing, I'm the best the bestment to be able is that's going to down. I'm going to

tell you. Is it a way it's a way that's the way the way of the way? I'm a

problem. All the questions, which were first, analysts, to be the other people are you

and it's the same means that data's fast programmable access of the way the next

the future to be a very much dramatically change. The other other way that's

our many customers are far far are the things like native coding agents. It's a Aksur not

that I'm a customer to go and I can't, I'm a case I'm a lot of times a or two-guna-

or two-guna- to do you have every time. This is we have a personal team's and we're even the

value we're at the same the company who are the or it's using, or it's using, both,

both, they're using are very very good, we're now, we're a lot of a thing,

which we have you can't connect to you? Can you, can't be, can't be short in our audience

can tell us that's what MCP can't be how much it's a lot of this is a MCP

MCP's METE MENTL control protocol. This is just a way that which can

any language model or coding agent can talk and it can't and it's and we

want to our customers snowflick intelligence like products use that's data

to make sure as aeasa'aupd-crapability also our focus interoperability on the tools that's the first you have been so that's no-flex

in the agent any other agent can't or interface to remotely access you use you use

this is MCP is a mccableable layer is this is a agent's how can't how can't how can't this is a big

question and I think that agents' shabd to be often people people pretty much people really really the same

in any way the way they're in the way are you are the same model and some code

mail are you have done there's not much kind of kind of and they need to how to

how to use for example if you say, write a little program for me then he know,

that's a file how to make, and then how to make a and then then the

code to try to call to call to kind of can't call to call but it's

but it's the time you ask you know what you do you're doing a agent's not that the portfolio

tool to call to call and then how how to how to and then the result of how to

how much and so that's all the level of work at even you you're

you're you're about but you're general concept very powerful is it

this coding agents actually actually the kind of any type of any line not want to

he's not a great benefit will be able to get. He's a every document access will be able to

access to and he's all the use very easy easy to can't be a easy to be a example.

If I you know that you access to you know, you know, any question any

any other agent can a plan can't make a and you can't execute to

you can't make you help can't gaps are and dachok-purned systems

kind of together are together in a lot of this is a way to this is a big root of

a lot of it's a way. It's a big root would be a big root would be

but I'm that this problem every day easy is a easy to the

data's problem's applying to apply to apply to apply to a

actual First, if a complex data set just a additional column would have to, then it was a data pipeline in badlau could have and this work for a programmer

can easily a half a half afto leascq, because he was he had to go and make a lot of things. Now, we have skills like things

are you know, which you can you you can't use the language in these this whole process from from end to end to

automate to automate to do. Some person can just can't work start can't But, okay, so GDPR

is, is it is the sameate or Dhinga is the same as or deeming? I mean, our personal data protection laws go.

Look, snowflake's the fide is the big hissas that we have been better

governance features offer to offer to the companies. You're make sure that the

newmos being a extra burden is. Look, I'm a American and every

If any other, any other people criticise citesites are and it's a un-chained consequences be put in. But, GDPR's many ways

are the same from a lot of time to be a good or bad? I think it's quite a lot of this is a much much-en-en

consequences are coming but it's it's also that's not so much thought to be made-you-

and it's every point-of-a- people are deep-sose-sue consequences are? Yeah, I first you

I'm just a positive points I'm a consumer as a company can't say you can't you know, you know,

you know, you know, you know, this is a great new thing this is a big big positive point.

It's a company, even my Google AdS team to be this thing to we can't that we're

that we're a lot of that's a other thing. Other side, the prompts of the divarons that you

are you're on you're on a point on you just on you're just on yes on click and I'm

I think it's a unchah consequence is. I think it's it's not a money for business money is much money to make sure

it's the most the best of the big companies because they're having to be the same, and they're having to be

money and then money is there money. But yet the new company to start from from the new company to start

to the same kind of thing is this way to this is a lot of things are the space in our data centers are not.

I'm, I'm, I'm such I'm I'm a right person not I'm a lot. But

math as it is a lot of it's a okay, so quantum computing

snowflake to what will unlock is? This is a good question.

I think like, like, uh, first, quantum computing be

big AI models like a big security risk and and it is a

data sure sure it's a high priority is this this things to look at all the things

to look at all the other ways to bring in a lot of different ways to come and we're confident we're in new techniques and the other of the other I think that

quantum computing in the world core infrastructure being being about about about you're about you're a

but you're young but you you're even you're even you're doing after you've got you've got to you're a

So, my mean, I mean, I'm in there a piece that's a realistic and very romantic is that I

think I'm think that I'm a positive thing. And Niva started with this such a faith was

that a better search engine to make a not just for people not just a big, but a shan-dard

business also and you the much you're you're one of one of

one of the one of you are, that's especially consumer products that's until they don't even

they're not the way the way they're more than the experience not at a level I know that the search

better is but I'm there's a concrete plan not that the search and information about my

knowledge is my know how much it's the end up that's it failed because we

we've made a search engine that was a little better than and his privacy be better

behted you. Behhack, you're much you seekeed that privacy is like a

privacy is like the same, which we're about we're talking about we're doing, but in the actually we're

practice not and we're really, we're not, we're not there's a team in the team in

the team in we're a great technology and after he's back and after that so that

thing's some very great came to be that I'm I'm I'm really that I'm

really this thing things in a learning experience or learning experiences in Europe in they're not

so not? So, very good. This is the after you go to snowflake in so you are the

CEO been made, the share price this thing not quite quite you know, you know,

how you know, I'm the snowflake of the day of things? Yeah,

my mind is that I know, I'm a snowflack CEO a such a much time

and a very critical situation when a when the company's growth such in

deepy to be reading I was in the day the CEO we made the year we're going to the next market consensus

from the full of the percentage points came back. This is the market in the real garrat the real reason.

Beshack, this part to help not get to a person who never a CEO's role not been never had been he was a

legendary Frank Sluteman he's heghy was in my to me, to see, many ways to happen to come in

and then uphance is a lot of the way to be the work. I'm just, I'm just a company for the

company's that's the last year to the same month, and then we were back we've got to focus to

start to be that. After, that's the thing's the thing is the thing people,

people are talking about people are but you if you're in the software business in,

so the question is, that you are you're in a such much people are the people are really we're doing,

we're doing the only focus, we're this is the sort of products made, you know, you'll see that you can't see that

this product to look at people in people in quite excitement and how much it's a product

that they're that they're making, and this thing can't make a the thing for the company

is common. before in time software engineering was the matter of

the problem to help to help to help to help to

help and so the details at the attention to take and

they're in the same type of apply to apply to apply to

apply ability you you have to have been able can be

this is a discipline is that the the wrongs not even can't be

a comma if you're not going to the compiler he's not going to I'm just a

adder and you're he's not and you need to be this whole picture is much better than

I'm think that modern AI driven software engineer is a conceptual is a

now he's now he's he's he's he's he's they're not they're using

they're they're using that's what's the problem to solve problems help are and them to help to help them and make

the product's context in the right and upy and upy are they're these new and better tools

use use and they're trying to help are really these things this is this whole every day,

every half every half a and I think that a year this is a different a school will be

I'll give you a great example I'm a my little bit 24 year's he's a year he's he's a

year old university from a software engineering he puth he was he's really he's

a system's programmer who's a very technical technicality details he's very much

much much so so three or five milliseicic latency way streaming systems

design design to his specialty and now he's he's a lab in he can't

he's he said he said he dad school and And then I've what I had that's the

any of the work in the work in a work in a work in a thing to be able to learn to be so much

you know, it's a lot of things are you are going to be a lot of that's a way, what is, what is the

thing is what is that's a thing is the first we've applied in the first in the first did you know, was a lot

specialised had been, I mean, I mean, my product to make, and use the using in the use of the steps

are to reach to, to your customers to them from each every one of the same in a very specialized discipline been made sure. You know, you have software engineers,

there are, product managers and designers and product marketing managers and technical program managers and this whole stack is always better that is all the same all the time till it

is that you have a perfect product and your target is to every scale optimize and to people to people to get to software to show but it

doesn't work not really when you are going to you're you're going to you're

a new product discuss how we've got this half the this week. This is the communication is

a way to make a way to you know, you know, you know, you're either horizontal are always always

change and evolve are you are going to you know you're a email machine can't I'm a

I'm doing to do you know I'm in life to be but I'm but sure you all the

information is all the I'm I'm much to be much to be much to make a

much better than I'm a good to and also I'm good principles to understand and

somemaging or team's important decisions to make help to do that I'm so I'm

very information consume I'm this is this is to make and you need to

you can't you that you're that are going to and so I'm

so I'm so I'm that I'm doing that it's going to it's a and a good and then you

and then you're making to help I don't think that new ideas do, new ideas, or anything is just CEO's the responsibility is the only thing the whole thing

where you can where your role is a facilitator you are the team's when you're saying we're going to this way we need to go we need we need

because it's a this is our future should be surecute this is all this this is support about it. What? The corporate culture is the important part

I think that I'm the most important things but first I'm a baseline from the baseline of the same. Like, each other's for civility, each other's between, and, at least, and, after, all of course, are the other's way of course.

This are the best-in-a-bordant baseline should be. At least, at any company in which I'm a hissna who I'm a hissah and I'm any other I don't even I don't even see-as-as-heaval-in-hue-cath-hought-haping. Company in any

any way-hack-law, or upman-jinnocir-cannic bortarer, I don't even be able to be acceptable. And I think that this-tri-set-cure is very important. And, this other part I'm the most of the same

open culture where things are the merit adhard to be where people or any other to be glottet or contradict-curned to before to think not not put,

where we're open and we're all all together we're all right, and where we meet and we're very quickly

to get to do you know, that's the we're going to do you know, so we're to execute can't,

but that open and honest communication and teamwork is a thing is that I'm very critical. Um, a lot of

you know, and you're in this is in all the world in all the all the world they're looking he's,

he's safe, say, say to say, I'm in the way, you're in the

way of the way of the so it's a kind of a car that's I'm a I'm in I'm in

Bengluru too, I'm in a place in a place in until I college not I'm a

law and middle class in a middle-class for the time in our four people's just a living room and

that was a big room to get a to this was a intellectual curiosity. My parents' from any college didn't go any college he did. They were just high school

put up, but they always he had very very education for both that they that they were a better

life and we to not do and we're to be a so, the good things that knowledgeable

to be know, smart to you're also that you're being a good that's a kardy mehnet we're

a better life and a better future to make a better future to and they're these

these qualities I'm now I'm I'm still I'm and I'm also I'm

you know that I'm my parents were very very clear very much were I'm

I'm a college where my parents me didn't want to make sure

to him to send to be too much to be and he's I'm,

I'm a lot of children I'm going to be a degree then it's a question of the you're going to you're going to

you're going to so when I'm being a lot of there's a little bit of the whole body how do you work is

I think it a very interesting subject is this is a topic is that it's a thing that is a lot of

complicated and the other side is it's a different too much as it's a different

and I'm that's much a much satisfying experience would be a Thank you, what's the way of you know what I've said. The things, what I've said. And I'm this one more

joddhought and that is that really minding really matter really makes. What you do, it's a much to be a matter making, to keep and to keep and to keep and

and new circumstances and to do not really and the third thing that I'm going, that is, the assapalta to and it's upbering the ability is

very important is it's very important you have new, new things are trying to, but with you should be able to you're in the some of the

in the way of you're in a bad you're in any way to you're not quite or any failure to say you are the way

you're going to whatever you have, I think that you're really important to so just these things

just that you just need to hard, hard, you're not there's a better way to get a better job,

you're not much, thank you so much. Thank you. Thank you. Thank you so much, Nicolai. You know,