I almost got Replaced at my job by AI until… @vaibhavsisinty

Ladies and gentlemen, I have a very interesting podcast today. A mirainti

is on the show. For people who do not knowiints outskill, which is one of the largest AI education platforms in the world, he runs staying ahead which is the largest AI community in the world where he

promises updates every two days where he's engaging with you about the current state of AI. And his mission, which is unique mission, is to help 100 million

people learn about AI.

influence and we are touching upon things that you know otherwise get very uncomfortable to touch upon because so if you're watching this order yourself some food go out for a walk go

to the gym put this in your ears enjoy this freeranging conversation about artificial intelligence with webin and by the way next episode I'm dropping

with a sports legend who you know who you know follow for that. Hey guys,

I forgot to say if you want to be part of the staying ahead AI community, the link is in the description.

>> How much are you into AI?

>> Sufficiently. I mean how much should I be?

>> What is sufficient?

>> Well, I mean I use it every day. Um I am >> Are you rolling?

>> Yeah, we are rolling. Okay. I am um I am reading up on at least all the non-technical stuff of it. I listen to Duaresh once in a while so I get a technical understanding of it.

>> But you are you a technical guy? No, I'm

not. I'm not technically.

>> So, you're like me. Script kitty.

>> What is the word you used?

>> Script kitty.

>> Script kitty.

>> So, have you been an engineer?

>> No, I'm not.

>> Okay. So, script kitties are those engineers >> who if they read the line of code, they probably will understand what it is because they studied some code in the past.

>> Got it.

>> But they have no idea what can do.

>> Okay.

>> Right. So, you're very good with copying and pasting.

>> All right. But you understand so much that copying and pasting of code doesn't scare you. Copy paste.

scare you. Copy paste.

a terminal window uh will not scare you but you have never worked on it so you'll figure you have the ability you have the feeling that I'll figure it out bro what a strange world we live in now

man I mean the average that must be a very scary situation. No.

>> Yeah, I think uh I mean education has to be redone completely ground up, right?

uh path traditional path that has fundamentally changed already but uh I think the world has not realized it. So

if you see China and stuff right first sixth grade say I think first grade AI because here's the reality okay if anybody knows what has happened in the

next 3 years they're lying >> nobody knows >> nobody knows >> right and yesterday I was meeting Alex Wong Alex Wong is the head of metai right we had a round table

>> uh >> and I had I was the only pseudo technical guy in that table so I had the opportunity to ask questions technical questions uh and not look like an idiot

because there's nobody better than me in that uh setup. So I asked him a question. Yeah, tell me something.

question. Yeah, tell me something.

Look, there's a PR vetted answer that you guys give when we ask you what is going to happen to the world, right?

>> Every AI researcher like there are papers being published and Elon Musk comes census, we're in singularity and uh we're talking about Dyson spheres and whatnot, right? We're talking about

whatnot, right? We're talking about sci-fi stuff that is turning into reality.

Has has the government or are you guys having a dialogue with the people on the table on what the [ __ ] will happen to this world two years from now?

>> How will economy even work? Yes. If see

we've been debating about jobs.

The way things are moving there are no jobs. there are going to be very few

jobs. there are going to be very few jobs and those few jobs will also go away because what people are saying have you read about the layers of >> why don't you explain even if I have why don't you explain us like so go through

it I >> I'll try to explain I'll connect a lot of dots okay and before I say this frankly this is me reading stuff me geeking out

on it okay right so what is the current state current state is AI is pretty good with most of the things right there is something called as uh GDP

GDP val is a benchmark which kind of measures AI versus humans on the top skills. So they have picked up 44 jobs

skills. So they have picked up 44 jobs which contributes to major chunk of GDP for us right 44 jobs AI it could be accounting nursing finance software some

44 job roles and they give tasks the same task a 10 year experienced uh person in that role does today as a good set of tasks are thrown at AI models

okay right now AI is better than better than human in 80% of the tasks and that's not the kicker H >> the kicker is they're thousand times

faster much much cheaper.

>> Oh [ __ ] >> And that is the case today.

>> And this number was 13% 18 months back.

So GPT4 which was like oh [ __ ] what is this? That was that score was some 12

this? That was that score was some 12 13%. 4.6 opus model actually yesterday

13%. 4.6 opus model actually yesterday only 4.6 dropped score 30% higher than the old model. So we are at probably 75

80% of the task AI being better at.

>> So when this happens what people were saying for the last one year is yeah take AI is very good at it but there is something that AI is not good with and that is taking decisions because AI is

not living and having this conversation.

AI is not going around. It doesn't have the human sense.

>> So you still need humans to do it. As a

result instead of 10 humans doing a job there's going to be one person doing the job with 10 different AIs. or 10

different agents. This is a narrative that you have heard. This is a concept called as orchestration.

>> You're you're a AI orchestrator.

>> The human instead of managing people, they're managing AI agents, which is great. Which made a lot of sense. Now I

great. Which made a lot of sense. Now I

ask you a question.

AI models, have you heard of Merkar?

>> No.

>> Merkar is this company which does a billion dollars in revenue today. Okay.

Panchillo all these large companies what it does is let's say you it wants uh let's say a AI company this is Spotify wants AI to get very good with

podcasting so what will it do it will get a plurer and 10 more to come and give uh you know like feedback to Spotify's AI

that is reinforcement learning a human is giving feedback to AI so that AI gets better at What happens when a lot of Praers right now AI has figured out podcasting all by

itself as a result is still winning but when Praers Joe Rogan and 50 other podcasters in the world come and give feedback to AI saying this is where you're screwing

up this is what you get better with what will happen eventually I can't fathom what the job will be >> right I can't imagine nobody has an answer to this everybody says humans

evolve but boss I just told you the speed of evolution of AI in 18 months.

>> Most of the country thinks AI is tangibility.

>> Right? Most of the world countries are wrong. In fact, India is better. By the

wrong. In fact, India is better. By the

way, India is better. I'm coming from the air summit. Of course, a lot of things they did is questionable. But

because of sure scale things can break, but I've seen some incredible companies coming out as maybe interesting.

I believe someone has to figure out the future of [ __ ] economics of how the world works, >> right? how the power law is changing,

>> right? how the power law is changing, >> right?

>> And power law right now, so far we were talking about 2080, then it became 1090, then it probably became 595. Right now,

we're probably at 0.5 and 99.5. I think

that's going to equate to 0.001 and rest of the world.

>> Right? You have to explain. Let me do that for people. So you are comfortable in Hindi max but the power law thing is basically the tools when technology gets inserted into

the struggle of capital people with existing capital make much more money and then people who don't have capital make much less money normal the rich get richer and the poor get poorer that distance will widen

because people who have early access to artificial intelligence and all its resources I also think you are Because you know I've been having a lot of conversations

about the future of work because timeline level is longterm everything is solved long-term the world autorects

population will autorect right like all that but shortterm jobs or political

boss that's very dangerous to think of like prompt engineering will be a new job.

Prompt engineering. I think prompt engineering is just added contextually pick up. I mean proper engineering is

pick up. I mean proper engineering is the simplest way of saying orchestration huh language basic of orchestration because the incentive of the artificial intelligence is to understand you

without you making too much effort like as a 20ome as a 15 something as a 35 something and you deal with people of this type all the time for a lot of

people.

Exactly. Give me a little brief on that and then >> tell me like what the hell like what the hell that's the mood of the conversation today. Huh.

today. Huh.

>> Fair. Yeah. Uh see I've been I've been doing too many things uh since the last uh 10 years. I've worked for interesting companies. I've built companies. I've

companies. I've built companies. I've

worked for companies like Uber. I've

worked for companies like Cluke. Uh did

business roles a bunch of them, right?

worked in multiple markets uh built a few companies did growth school before right now we are completely focused on outskill what we do at outskill is uh you know

teach the world AI right and while I don't think it's a long-term solution to the problem but I do believe that it's a short-term solution uh of figuring out what could be the next thing and that is

why we enable people with AI teaching AI right uh and that's the only thing that we do and yeah we teach to people in more than 100 different countries every single month.

>> We teach millions of people.

>> We have we're trying to make AI education free.

>> Uh one that we do is via distribution.

What we are doing right now, of course, tapping into your distribution to teach a lot more people about AI.

>> Uh and of course, uh there is social and then there is our zoom sessions and stuff like that. So that's what high level I do. So I basically the reason why we able to have this dialogue is because I've been geeking on this for

the last two and a half years. I've left

everything to just do this right and the reason why I did it is because first time when I used not chat GP but something called as uh code interpreter

inside of chat GPD so code interpreter back in couple of years back right after chat GP released 3.5 they released a feature called as code

interpreter chat GP I was able to dump a CSV file and I was able to say can you analyze and tell me the top insights so this was a couple of years back and when it drew me graphs and charts and gave me

insights that my data team would do in like 5 minutes I'm like >> this is 2 years back >> right >> and since then I'm saying boss what is going to happen to jobs because I saw it

write code at speed that no human can even imagine >> right right completely >> and this was two years back so since then till now I've been tracking I've been seeing I've been reading I've been

geeking on what hap what is happening in the world of AI and been playing with it on top of Right. So that puts me I believe in an

Right. So that puts me I believe in an interesting position to kind of gauge >> what could be the thing and I also happen to run a 200 member company. So I

am also seeing how business how roles are evolving inside of my company. How

my product designer >> is able to ship a full app who someone who was sitting on Figma right now shipping an app which is in production

being used by users. How a intern who joined 6 months back has built a system that is being used by half a million people in our uh of our users without a software engineer.

Right? How people different kind how marketers have become video guys who are doing creative strategy and creating visuals and creatives all by themselves while before this they were just writing

copy.

How a head of performance marketing kind of runs performance marketing via MCPs and AI agents rather than more humans.

The world as we know it is changing in front of us.

>> And what what puts me in this position and what makes me most of the people either are ignorant about it because they're too scared of

it of accepting it or two is they just don't realize what's happening. And both

of these guys have to be told boss if everything was right >> right let's say I'm wrong and I hope I'm wrong. Let's say you're wrong and I hope

wrong. Let's say you're wrong and I hope you're wrong. But it's not our thoughts.

you're wrong. But it's not our thoughts.

We are reciprocating what we are reading. I'm I'm reading what a Elon is

reading. I'm I'm reading what a Elon is saying. I'm reading what a Sam is

saying. I'm reading what a Sam is saying. I'm I'm reading what I Yan Lakun

saying. I'm I'm reading what I Yan Lakun is saying. Right. I'm I'm saying basis

is saying. Right. I'm I'm saying basis of what I'm hearing from people. And

most of the people don't have an incentive to lie.

In fact, they have an incentive to compress the facts.

>> Yeah. Minimize it.

>> Minimize it.

>> Which they have been doing for a very long time, which is a sad part, >> right? They have been trying to say this

>> right? They have been trying to say this because nobody wants Trump to come and say that you know what let's stop it while he can't do it because China will screw

their asses right there's a competitive angle to it and there isology competitive forces that just outweigh everything else >> I think we are out of that uh question also Huh?

>> There is no like back in the day. I I

don't know if you have read this that Elon Musk and all these people signed saying we should slow down AI pace.

>> I think so. I think so. Yes.

You're right.

The game that is playing is on a different league. You see the world

different league. You see the world order is changing.

Yeah. Yeah. But you know what is very interesting about what you were just saying right now about how performance marketers have an ascension in their role. Yeah. Um somebody who was doing

role. Yeah. Um somebody who was doing growth is now doing creative also.

Somebody doing PM is now suddenly also building products.

>> So is like the short-term future of work in artificial intelligence just ascension key all tasks include okay suddenly I have a whole role as opposed to a partial sliver.

>> I think so. I believe so. I truly

believe in that because I'm seeing incredible results in that level of optimization because I believe you know everybody has to become something called as an AI generalist >> who are generalists generalists are

problem solvers right and problem solving as a generalist was confined to the skills that you had before content creator

to streamline your podcasting service right it is confined to a point before even though you are a strong generalist but AI just break those barriers you can do anything you want I was just having

dinner uh sorry lunch before I came here and I met a couple of uh kids working for a startup that I don't want to name uh and they saw me and they were excited and I asked them what do you do what do you guys do one person is a PM product

manager and the other person is a marketing manager and I saw the screens set around both of them had VS code on VS code is a ID where software engineers write code both of them had VS code on I

was like you guys have VS code on I was like yeah we are uh vip coding a product to solve uh basically this company has hundreds of YouTube channels and they're

not able to figure out what is the insight from each of the comment that's coming because they're getting tens of thousands of comments so one product manager one marketing manager look at the combination are sitting and building

a product to understand what their audiences are saying which is supposedly a software engineer and a aim engineer's job >> a marketing manager or a product manager

I was so happy to see this.

>> What they're basically doing at that point is probably giving requirement to something like a cursor or whatever and or clawed or whatever and then it is producing the code that they're then running.

>> Yes.

>> Wow.

>> Yes.

>> Right.

>> Yes.

>> It's a different language by the way altogether. It's a language that they do

altogether. It's a language that they do not speak necessarily.

>> They don't understand what's happening but it is producing digital infrastructure at this point. And the

cost of producing digital infrastructure has gone to close to zero.

>> Yeah, it's the cost of inference is going to zero. Like yeah, I was reading one more stat today as I was coming to this part. Uh Sam Alman just said that

this part. Uh Sam Alman just said that India hit 100 million monthly active users on chat GPT. 100 million

on just chat GPT right now.

>> What does that mean?

>> What does that mean? That basically

means that every 100 million people in the country today >> that is 20% of people who have access to internet have an access to a superpower.

>> That's a very interesting statistic >> that is democratization of knowledge at billion scale.

>> Sure. Right. Huh. So like everybody from a farmer to a carpenter to my caretaker

uh you know who now so my grandmother has caretaker she speaks in Canada she opens up Gemini and says auntie is

feeling like this in in Canada >> what should I take as the next step in the night when nobody's up let's say I'm not allowed >> right

>> I just taught her tap on this button say and Gemini says says it out saying okay you know what give her a massage it's for my grandma right >> give her give her a leg massage and do it like this in Canada >> in Canada >> in Canada

>> that's crazy >> yeah it's not perfect multilingual is not perfect Indian languages perfect that is where AI for Bat and Joe talks

and serving a lot of work together to build India stack but technological >> democratization you get my point I'll tell you one more huh I have Sep who's

my cook >> right uh protein shake I'm done I don't want to drink more protein shakes right I've tried every bloody protein shake in the world right

>> now I've given him protein milk and I is which is 50 g of protein done for the day >> and now he just goes to chat GPT or something turns on voice because he

doesn't know how to type WhatsApp recipe. Every 3 days he surprises me

recipe. Every 3 days he surprises me with a new recipe of protein >> of protein.

Two days back he got strawberries, raspberries or something like that and some cumin powder something.

This is like a bloody milkshake, right?

>> It's crazy.

>> Yeah. May I was having um you know a lot of our jobs are sitting down and sitting down over a long period of time will begin to hurt your neck and your back and overall you realize

>> well I have some postural discomfort >> for a year I was struggling with like going to physios and trying to understand and I did get a handle issue

we to come up with a system in my busy schedule of just stretches by the stretches and strengthening.

It's like command prompt window back in the day in 1997 before Windows came about reality

and boss I am I feel like a new man and then you should do this kind of strengthening and that kind of strengthening idea generation I feel like it's infinite my wife

actually you know when she's uh painting she will use it to understand what colors to mix to produce special effects.

>> Beautiful.

>> I don't understand the technicality of painting that well, but like Uska, it's an active. There's three people in the

an active. There's three people in the room.

>> There is the subject that she's painting. There is a painter itself and

painting. There is a painter itself and then bloody chat who she's like talking to constantly figuring out how she wants to express herself.

>> And friend at this point, it's a friend with infinite ideas. That's what it's come to

infinite ideas. That's what it's come to be. It's an assistant if you want it to

be. It's an assistant if you want it to be. It's a friend if you want it to be.

be. It's a friend if you want it to be.

It's a it's a EA if you want it to be something of that nature, right? That's

what it's acquiring as a space.

>> After having spent a lot of time also on it for personal use, I found over a long period of time. So like at a certain level of depth

after seven, eight corrections to my stretching routine, it forgot what the original use case of the stretch was. So

that taste or judgment because I look at it from my profession.

My profession operates on personality to some extent. It operates on systems and

some extent. It operates on systems and optics.

Personality, taste, and judgment. That's

what my game's about.

It by nature is a game of non-verages. M

like there are certain games you play non average wins and I feel like are >> what do you mean by non-verages wins?

As a trader, say I trade the stock market right?

>> If I have the same information as everybody else, there is nothing exceptional I will do.

>> Got it. Got it. What is your edge?

>> Right. You need that sort of edge. The

market is moving in that direction.

Artificial intelligence. That is

something that the AI is not predicting yet. You know, at least the ones that we

yet. You know, at least the ones that we have access to because they are very generalized. Maybe you're smiling at me

generalized. Maybe you're smiling at me like I'm wrong. You're smiling at me like I'm [ __ ] wrong. So tell me if I don't know properly because I don't think you can have super

intelligence democratized. I think

intelligence democratized. I think intelligence layer like everybody has it then you need something else as a competitive advantage. No am I never about the tool.

advantage. No am I never about the tool.

It's about how you use the tool.

>> Tell me more right again like you know your girlfriend uh does painting with the same paintbrush that you can try to do the same painting. No way. You can't

right? Like there's no way you'll get there.

to generalize a statement that I made but the thing with AI is what they have done so far is how can anyone in the world can use the product is what they're optimizing for right you're

right on that level everybody has the same tool if you ask it to write uh do a research on the questions that I should ask web hub like it'll give you a list of questions and it'll probably give a similar question to 10 more people who

ask how do you take it to one level x one level one level more hey you know what webhub is coming for this podcast.

Why don't you go through all the shots of mine that I've ever created around tech and AI pick up the comments basis of that frame the questions that is second layer of

better prompting or better pushing the system to think the third layer of thing is hey web has done so many podcasts he has so much content go through all of

his comments podcast of his comments podcast of his shorts and reels pull out all the top questions that has come in pull out the things that have done Well, I want to add my flavor to it and then

build a question that gets to a better layer of questioning even further. I you

can take it to even next level saying there are some big voices of AI coming in. So not just this information of what has worked well and what what webhouse audience what my

audience are asking but actually go into Reddit and go into threads of Reddit which is getting the highest amount of negativity in terms of people being scared about people need answers that

they don't have. Find me those questions as well which is datadriven.

Now the same tool that you're using can do all of this.

>> It's just that we don't know how to make them happen. It's the same paintbrush,

them happen. It's the same paintbrush, but your girlfriend figured out that she can use AI to mix two colors to give a effect that you can't even comprehend.

Right. Because you are not an artist.

>> Right.

>> Right. With AI, the m the more we stay on superficial levels. That is where it stays. You have to push it.

stays. You have to push it.

>> You have to push it. You have to ask more. You you have to say that I I don't

more. You you have to say that I I don't like this. What more can I do? Give me

like this. What more can I do? Give me

more ideas. How can I make it better?

And it does help you make it better. Now

as I say this thing coming to the point of hey it forgets context it's a technical problem every AI model has something called as a context window have you

heard of context window context window is see AI has AI is like a gaji right information most of the context windows are under

200k tokens so the okay simply saying let's say more than two books of information if you go beyond it what was said before it forgets So it's it's called context window. How

much can it remember?

Let's say I'm assuming what you would have done is you opened up a chat GP or a Gemini or a Claude window let's say Gemini in this case and you started to talk to it.

So that has gone beyond its context window.

Now there are two ways to solve this.

One is that your general tools use these are consumer grade tools that is for people like you, me and also Sundep >> right everybody should be able to use it so it by default comes with a lower context window because of cost

optimization but most of these models like anthropic and all all of them have have hit a million context window memory is already five six 10 times

better but expensive as a result open and you would have used the right API there then this problem would not have happened >> but the best solution to context window

actually is to use a project >> got it >> claw project >> right right >> uh Gemini may there's something called as Gemini gems >> right >> chat GPT has something called as projects or chat GP also has something

called as custom GPS >> just may you can write a prompt prompt is not anything you don't have to go any scientific you just talk about let's say you had a nice conversation you got it

for the first time that AI nailed it.

Okay, to converse, it figured out the way around it. AI nailed it. Then you

just say, look, you got all the context.

Make sure you convert everything that I've just told you into a reusable prompt that I can use inside of a project. So what will AI do is for all

project. So what will AI do is for all the conversation it has had so far, >> it will condense all of that into a prompt. You take that prompt, go to

prompt. You take that prompt, go to chat, GP projects, sidebar. Yeah, Gemini

gems. Yeah, claw project. Whichever tool

you're using, there's always a box called instructions. You just dump it

called instructions. You just dump it there.

Now, the base context is set. No matter

how many times you use that project, it always goes back to that same book.

>> Ah, got it.

>> So, the information is retained as a result context breaking. Every time you found a new insight and I do this if I'm using a project and I believe that I've

given it new instructions new insight I say can you tell me what are the new things that I just told you it'll again give you an insight you copy and paste that inside the instructions again

>> so you're building your workbook >> instruction book >> very interesting >> so this is technically called as a rag right but the simplest language for all

these commercial grade products are essentially uh like I told you >> memory window >> uh it's it's a it's a instruction and on top of that right now

memory is actually slightly different right like memory is remembering the nuance of things across >> got it context is more shortterm >> context is how much it can remember over

a course of time got it >> right uh after a point of time so memory and instructions every time you send a question it doesn't answer it goes and

reads the instructions first. It goes

and reads your memory first to remind itself what exactly it is supposed to do. So it never forgets.

do. So it never forgets.

Sometimes it does. So you just have to say, can you make sure you read the instructions again?

>> Right.

>> And it just works.

>> It's pretty funny if you prompt your artificial intelligence with uh extreme, it behaves very differently. Yeah. And

so it kind of replicates the human emotion.

And it's so funny when you're typing it, you're like and it gives it gives a different response. The there is also some kind of

response. The there is also some kind of deep research mode you can access. For

instance, GPD has reason mode, right?

Like where it where it does like analytical stuff and it explains why it's thinking. And I remember the first

it's thinking. And I remember the first time I came across that it was kind of like bro like for the first time bro I've read all the books you can think in the world right like I'm very fond for the first

time I read something and I was like this makes no sense to me this is such advanced knowledge and now you can laugh about it in hindsight the minute you're first exposed to something like this

it's fairly scary and um >> I think the short-term good news is this is a tool but >> I wonder you talked to a lot of people who who are using or beginning to use

artificial intelligence what do you observe about them like how are they using it where what is what are some things they can do to make it better etc I think

the best use of tech so far of AI has been on offloading a lot of thinking

right outsourcing thinking to AI it's like working with AI or thinking with AI you know what you want to do uh but you

want to get that get that AI's reasoning model to think for you deeper then you can because you cannot sit and read 25 articles

so people who've been outsourcing that thinking hey I'm thinking of uh you know answering this question to Praar's question

what do you think about the answer >> first is it factually correct because I might have read something else and I'm telling something So is it factually correct? Two, can you quickly get me two

correct? Two, can you quickly get me two three data points? These are my original thoughts though, right? But can you get me two three data points that validates this?

>> So this is outsourcing of thinking, right? And I see this people doing

right? And I see this people doing across for different things like voice mode is a powerful way to do it. A lot

of tech guys or in fact in my team the sub tech guys like uh they operate like tech guys. So a lot of this thinking

tech guys. So a lot of this thinking happens while building products also. So

I'm thinking of taking this approach.

What do you think? And oh yeah that's a great way to go about it. They'll say

okay implement it now.

>> Right? So that outsourcing understanding how you can outsource thinking is a very interesting thing to do. What you should not do is allow AI to do all the work. So there

are two ways of going about it. Give me

answer to this question or which is a wrong way to do. The right way to do is this is what I'm thinking to answer this question. Tell me how good am I or how

question. Tell me how good am I or how can I make it better?

Right? This is working with AI to the first option is just giving the work to AI. The moment you give work to AI, your

AI. The moment you give work to AI, your brain becomes a rot space. Right? You

stop thinking. Your ability to think goes away goes to donkeys. And there

have been researchers, right? You must

have read that MIT research where uh you know the cognitive ability of humans who use a lot of AI to outsource in a way that that they're not thinking anymore.

They're giving air to think. Their

cognitive ability is shrinking. It's

going down.

>> And it is already measurable.

>> It is I mean they have how they measure.

>> Can you look up the study?

>> Look up the study MIT study one and a half years. In fact, there's a recent study that happened on coding.

This is specifically on coding. And I'll

tell you before I go go into the coding thing. The reason why coding is a very

thing. The reason why coding is a very very any study that happens on coding is very interesting and is relevant to everyone not just software engineers is because the world on the internet runs

on code. It's just that we are not

on code. It's just that we are not surfaced to it.

>> Right. You don't write code to edit your videos but it runs on the code in the back end.

>> So anything any study on code is actually relevant to everyone who is working in the knowledge economy.

>> Right. Right now this coding study failed to remember who did it quite frankly. You can look it up again. What

frankly. You can look it up again. What

they did was uh they gave a problem statement uh to two sets of people.

were able to ship the product much faster. People who didn't have AI, they

faster. People who didn't have AI, they had to learn. Both of them had to learn and do it. These guys took the time to learn without AI and eventually ship the product. Obviously, people with AI were

product. Obviously, people with AI were much faster, which is great news, right?

But then a follow-up question on when something breaks and someone were asked a question on the system that they built. People who built with AI

built. People who built with AI struggled to solve for it. They could

not fix a bug. They could not figure out how to make the changes. But the people who did it themselves were much better at it. Though they were fast. Now in the

at it. Though they were fast. Now in the bucket of people who used code or who used AI to build it, there were two kinds of people. One who just got AI to do everything for themselves. They are

the guys who suck donkeys. But there's

another set of people who took the approach of understanding what the problem is with AI figuring out what AI is actually doing and working with it

where actually the best out of these three lots.

>> So it's not alone it's not only AI it's that AI right okay >> it's with AI >> are the people who did it. So there's a fine line that you have to draw. So when

interns and all these kids join, right?

In the orientation, I say one thing. If

you not if you are someone who will not use AI, you will not be in this company for a very long time. But if you are someone who use only AI, you'll not be good enough to be in any company in the world.

>> You have to find that middle ground.

That's very interesting and that's very well said actually. I think it is um you know I was reading this about um lawyers. I come from

lawyers. I come from >> a lot of people in my family are lawyers. So uh I almost also became a

lawyers. So uh I almost also became a lawyer.

The context is right.

Right. Spiritually

I was reading this. I think the chief justice from the chief justice Supreme Court said something. They're writing

their petitions. Lawyers are writing it in using AI and it's just producing [ __ ] >> What is that? Mankind versus

>> mercy. Mercy versus mankind. And it's

producing some [ __ ] case like case laws and whatever and inserting it. It

is uh AI did something where it quoted a judgment written by some other judge where the judge has not said anything of that nature at all. cross referencing

paper. So,

but it has never come out of his mouth and he realizes you would bake something into law that only AI imagined and

right and so the the chief justice was well obviously angry annoyed at the fact >> but it's very well said key no matter

what person of what age they are asking me about artificial intelligence or even social class like my driver sitting be like right

crazy but voice through he will find a way to access it also but it's kind of like magic and it's going to invade every part of our lives and so we have to find

a way to work with it that is not there has to be that sort of element Right. Yeah. Very interesting.

Right. Yeah. Very interesting.

>> And it's not just India. I think this is uh lawyers being very lazy on not having a paid account of a tool. I guess

because while hination is true ease where it quoted things that didn't exist, >> this is where it becomes important for you to have the knowledge of using the

right tool for the right task.

>> You cannot just throw like you cannot just be that person who will just ask a question and get an answer and use it.

>> Right.

>> Right. citations and all these used to be problems before. They're not so much of a problem anymore >> because of features like deep research and all which builds everything

>> basis of available information to notebook type tool if I had to do something with law in fact I have done a few things companies you know what

happens right so I need to be a micro lawyer myself every founder has to be >> so basically I have my notebook LM set up where I have fed in every corporate

locker uh stuff and I keep doing deep research to keep adding sources of you know citations of case laws and all of that and I query that knowledge rather

than open chat GP knowledge.

So any question any answer that comes for any question comes from this rather than the open word as a result hallucination fundamentally goes down >> wrong citations

>> to fix this >> the way I you the way I operate when it comes to I can't be wrong in this is I don't rely on one model >> okay >> answer

I will basically go to a claude >> turn on the extended thinking and turn on the research mode on and I'll say these are the facts vet it and tell me how many of them are correct, how many of them are wrong or how many of them

are directionally correct.

>> I want to have understanding of all of this and it gives me the second layer of vetting. SN AI cannot do it. It's just

vetting. SN AI cannot do it. It's just

that you don't understand how to use AI to do it in the right way.

>> That's it.

>> Right? Because I have seen this screw-ups being done not just by lawyers man like Deoid Consulting has published a report for US I think for Fed, right?

Which was which had all wrong citations.

They were fined. Deoid was fined for this.

>> So I was sitting with a guy. I mean I turned around and he was making this fancy deck. I didn't know who he was.

fancy deck. I didn't know who he was.

And he was switching between uh chat GPT and deck chat GP. Copy paste copy paste copy paste.

He was like I'm a VP in this consulting firm and I'm going for a presentation for a client and hence I'm doing but you're using AI bro for everything. Yeah

that's what we do. Then why do people pay you? and he turns around and like

pay you? and he turns around and like look man I don't want to say this but the reality is that people pay us because they don't know how to use AI anymore >> really that's that's what he throws around and goes back to his work

>> wow it almost sounds like a and say the producer is using AI and say the consumer is using AI like you know the funny instance is I am texting somebody for something and I'm using AI to text

them and they're texting me back using chat GPD prompt and so eventually computer computers you're just like performing, right? Um,

>> orchestrating.

>> Orchestrating is a big word at that point, boss. Like, you know, I think one

point, boss. Like, you know, I think one of the mistakes that freshers use, freshers do when they're using artificial intelligence is they use it in communication. And I think it's such

in communication. And I think it's such an instant distrust signal.

You'll tell me these are my thoughts. I

just used AI to compress them. You're

making a fundamental social mistake here. I will not be able to trust your

here. I will not be able to trust your intelligence because you mastered it behind AI very early like at least or >> I think it goes back to the theme of supplementing what you know

>> adding AI into the workflow as opposed to AI becoming the workflow itself at least in these realms. Right.

>> Fair. I think I think I think we're so early to these things that I think our judgments will also evolve.

>> That's a good point.

>> Right. Uh Mdash does annoy me. Right? Uh

when I see emails, when I get cold emails and all, but I also realize that without that mdash, if the person would have written the email, I hope that it was his knowledge or her knowledge and they used it and not just said write a

code email to web hub.

>> Right?

>> But I also do the same >> now I'm judging that person based of the fact that that person forgot to do a chain prompt saying make sure you remove mdash.

>> Right?

>> Now the way I judge that person right now is like okay the person how do I treat this person? Like the person is using AI, it's a good job. I want I want to work with more people who use AI.

>> But then the person is using fundamental AI which my grandmother uses. So I'm not so happy with you.

>> So I'm okay with you using AI but just using it in the right way.

>> Right. Right.

>> Right. Uh andically treated which mimics my style of writing. Okay.

My bloody content is generated using AI.

I generate 100 million views per month on just content which is generated using AI.

>> All right.

>> Now, if anybody says that, bro, you didn't put effort AI created. I'm like,

are you [ __ ] mad?

>> Do you think 100 million people will watch my content if I put no effort?

>> All the thoughts are mine.

>> It's me thinking. It's just that I didn't have the time to sit and shoot content.

>> Right. Right. So it's the same feeling that I have before I judge someone saying that hey you use AI today.

>> I see what you mean. Um you know I was talking to Vun yesterday >> and I was likeun the first two places where AI will or rather I I'll say it

like this the first point of contact people will have with artificial intelligence is content. Not just in the sense or tweets but

if I look at the world as an evolution of media cycle we went from radio waves to television waves which were gatecapped by production houses to then

democratized >> but it is still curated content it is still content makes that then Himmanu consumes or consumes now we are in personalized content era where whatever I talk to it gives me personalized

responses at the end of the day it's content it's low-level entertainment mano even the way Chad GBT responds and says oh by the way do you also want me to answer these three questions it's

this continuation of content right um and we are already in the era where all content is personalized >> my guess is most people in this case will be using

artificial intelligence for low-level anxiety management uncertainty uncertainty macro level and I'm like hey boss uncertainty boss uncertainty

Is that your read also in in people like the primary purpose eventually becomes this pseudotherapy kind of a zone for people? It is a huge use case, right?

people? It is a huge use case, right?

>> Huge use case.

>> Huge.

>> And it has become a entry barrier. I

mean, it has become an entry point for a lot of people to use AI, right? Uh

because uncertaintity, questions of uncert uncertaintity is something there was nobody to answer them and now they have somebody to answer, >> right? Like it's crazy, dude. My dad

>> right? Like it's crazy, dude. My dad

recently got diagnosed with cancer.

>> Oh, I'm sorry, bro.

>> Yeah. But he's doing well right now, right? Rice up coach it's all good right

right? Rice up coach it's all good right now and he is so educated on this right now purely because of AI every bloody report that comes in goes inside of it if he feels back pain before he tells me

he tells AI >> hey these are uh because AI knows what has happened and he I've given him the project of all his files inside of it so it has context right and it has become

his go-to thing and AI in a way has been quite nice right it is more positive than negative >> for most of the cases Right? Because if

you go to a doctor, what is happening with hospitals in the in the country, right? My grandmom is feeling a little

right? My grandmom is feeling a little unwell. So we take her to hospital and

unwell. So we take her to hospital and the doctor says admit her, right? Just

admit her because that's how they can make money. Right? Same happens with

make money. Right? Same happens with this. But an AI is able to take a step

this. But an AI is able to take a step back, look at your reports and able to gauge you based on data, based on reports, based on reality like the scientific movements that have happened,

based on all the data it has. As a

result, it is able to guide him and tell him, "Look, you just had a therapy session a couple of days back that could lead to inflammation and that could lead

to back pain. So, it's fine stretch and he did those stretches. 3 4

days later, the pain went away and he built that trust. You know what I mean?

With now if I say something, dad, you're wrong." He says, "No, look what Chad [ __ ]

wrong." He says, "No, look what Chad [ __ ] said.

It was right before so it's right now also I don't trust you son >> right >> I trust this damn thing more then I have to debate more I have to ask him show me your conversation show what you asked it

this is beautiful but if there is something wrong it will put those red flags as well which has done it for my grandmom >> right it put those red flags at 2:00 in the night I saw some symptoms kind of

worsening on my grandmom and I asked AI it said boss this is not looking good just call an ambulance and take her to ICU and we did

or else what would I have done because instinctively first thing to our head is not that take her to the hospital ICU is not something that is commonly thrown at in India right we don't even know what the entry point to

ICU is stage exactly because for us in movies that we have seen ICU yes you're for us. So the only thing that I could

for us. So the only thing that I could think of at that point of time is let's see what happens in the morning.

>> We will take her to the hospital and I asked AI the same thing. Can't I take her to hospital at 9:00 in the morning?

>> And it said no. AI said 8 hours back you said this is the condition. 8 hours

later you said this is the condition and it asked me to get some test done of a sodium test done. I got the test done or sodium low ta >> right which also doctor recommended but

three hours later when I got access to a doctor on pract >> right >> but I had recommended before that so I did a test on the beauty of the country that we live in man >> I can get a test done at any hour in

hours without an appointment >> right >> got the test done send the report back AI said look this is what it is the symptoms are this it could be this this this and this

>> it's not worth taking a risk considering how old she is and if you delay it till the morning, you don't want things to get worse now.

So if you can afford to if you can just take her to ICU that's the best thing you can do as a grandson being nice as well.

>> And there's a moral push to >> Yeah, there's a moral push to it always.

Always >> right.

>> So interesting.

>> And I'm like what my dad was like I told my dad this is I was like my dad was like wait till morning like she's talking and all but like like no let's take her and we took her.

>> Um my grandfather has a very unique eye condition. um very rare and anytime we

condition. um very rare and anytime we get any tests done even when he's sick like at this point my father will be like and once I started doing that trust now

I get blood test done every 3 months just to be on top of my health right >> everything to the tea like I spend 5 6,000 rupees every 3 months test my body every organ right

>> and not only was I able to be like there is a delta here for which I would have paid a doctor What are supplements that I can also consider? Acha. Why are certain just

consider? Acha. Why are certain just vitamin D is a big one, right? All of us are deficient in vitamin D. All of us.

Now, here is something I learned. No

matter how much oral vitamin D I took, it's just not cutting it.

>> It will not move the needle beyond a point.

>> No. But the minute I found liposomaal spray that goes under the tongue.

>> Interesting, >> bro. My vitamin D went from 10 12 at a

>> bro. My vitamin D went from 10 12 at a certain point, which is very low to 50 in 3 months. which you will not find without injections but I just found the right thing that works for my body I

guess and this is an experimentation tracker you if you don't have on the health side >> I would have needed an expert boss for this kind of stuff right and so being able to understand your body so quickly

um I started dealing with some amount of health anxiety as a function of just growing up suddenly start thinking oh ailments right um family history um for

instance sleep apnea is a big one family and having a very light sleep is a proper thing and I started developing early symptoms of sleep apnea. Sleep

apnea complicated unspoken. I bet people listening to this podcast don't know what sleep because obstructive sleep apnea or obstructive sleep apnea or central

nervous system sleep apnea bro. Suddenly

I know everything like like [ __ ] research like you know um and it also helps me not go into these spirals because if you remember

right like Google did not have that kind of precision until it has moved to LLM based search functions and it's very interesting because you are also right about the fact where the AI is

encouraging you you go do the stupid [ __ ] go it'll tell you where you're wrong. Also, I don't think it does that

wrong. Also, I don't think it does that job perfectly. I think the beauty I

job perfectly. I think the beauty I found of deep AI usage in some of the people I know use it a lot. Their sense

of judgment is very retained.

They'll be like, "No, no, it got it wrong." So, that sense of judgment is

wrong." So, that sense of judgment is very retained. But overall, what I'm

very retained. But overall, what I'm trying to think about here is that strong to I said, "Vun, what do you think about the psychological cost of everybody being in their own bubble talking to their own AI all the time?" M

he said whether you like to believe it or not we are all low psychosis patients already that level and it's micro it's not it's not diagnostic

but that's a very interesting POV too that we all have more pronounced individuality because this stuff is telling us you're right this confirmation you're confirmed go ahead do this Right. That's a great idea.

>> That's a great idea. I think it's a life-changing idea you've come across.

Prairie >> there's a word for this. I forgot the I forgot the word. I'm terrible with these remembering of these words. This this

characteristic of AI has a word. Can't

remember it.

>> Okay. Sophensy.

>> Secoensy. Yes. Yes.

>> Right.

>> AI is offensic.

>> Seopantic. So interesting. key AI and the most ideal engagement tactic that these companies who spent hundreds of billions of

dollars had was chu right there is a social clue in there. There's

a social clue in there. We've known this from without AI also.

This boot booting, right? AI confirms it because it's optimized for the opposite.

AI got it because of the society.

>> Sure. Right. Right. In the sense that it still is the meta, the most effective tactical, right? Short at anything. So

tactical, right? Short at anything. So

interesting. I've only mostly use GPT some Grock and Grock's kind of unhinged.

But are anthropic and claude also psychopentic?

>> Everything is because they're fundamentally the way they are trained >> the core material is that. But there's a way to make it better by putting in instructions. All these apps have

instructions. All these apps have instructions uh core instruction that you can set up for whole of your product. The whole of chat define

product. The whole of chat define instructions.

I always say be the opposite of being nice. Tell me why I'm wrong.

nice. Tell me why I'm wrong.

>> Right? Don't don't try to be nice. Don't

try to convince me. Tell me the facts.

So you write an instruction like that and you evolve over time. It get better.

It gets much much better.

>> Got it. Got it. So how much of your life at this point is covered with AI as an educator?

>> All of it.

>> Most of it or all of it? All of it. And

there's no difference between most and all. 90% like you were saying

all. 90% like you were saying >> anything that can be done with AI is being done with AI. Is all of it according to Got it.

>> Now can I sit and uh teach a Zoom session with 25,000 people live on it?

It can't be done with AI today.

>> So I do it.

>> Got it.

>> But can I make a 20inut video on YouTube without me shooting it? Yes, it can be done because an editing team it's not live. So my editing team can edit it out

live. So my editing team can edit it out nicely even if my AI clone does it.

>> So that is AI.

>> Got it. So you're teaching the live stuff is you're teaching it yourself and a lot of the video stuff you can sort of offload. I see what

offload. I see what >> and this is and this is example of content creation, right? like you pick up anything uh that there is >> it's mostly AI first at this point of

time >> from starting from hey like what's happening on my Slack uh again like I said 200 employees in a company on a Monday Slack is a [ __ ] mess there'll be thousand messages >> right

>> there is and I'm tagged on everything because cing webHub is important >> right every team thinks CCing web is important so everybody CCs me for something that I don't need to don't need my attention of also right

>> so the way I use Slack is fundamentally different. My AI reads my Slack first

different. My AI reads my Slack first and tells me what I should read, then I read those things. Same goes with emails. If a unknown number calls me on

emails. If a unknown number calls me on my phone, I don't pick up. It's my AI agent that picks up my phone, talks, understands, and if it thinks it's worth it, then it does a offloading to me. If

it's a known number, then it comes to me directly. If it's a number which is

directly. If it's a number which is saved, >> that's crazy. WebHub, that's crazy. If I

call you with an unknown number, I'm going to be speaking My phone's charging outside thankfully. I'm not excited to

outside thankfully. I'm not excited to speak. Bro, I had this experience. I've

speak. Bro, I had this experience. I've

said this on the podcast many times, but this was kind of shocking for me. I was

in uh Vegas about four or five months ago. Get into the hotel. It's dead of

ago. Get into the hotel. It's dead of the night. The room is smelling. So, I'm

the night. The room is smelling. So, I'm

like, boss, somebody needs to come clean this.

>> Call uh reception and I speak to this woman for pleasant woman for 2 minutes.

I am like explaining, listen, it's been a long day. I need somebody to come in urgently and clean this. What is the smell like? I tell her what the smell is

smell like? I tell her what the smell is like. Where do you think it's coming

like. Where do you think it's coming from? I tell her where it's coming from.

from? I tell her where it's coming from.

Okay, thank you so much. I'll send

someone shortly. And as soon as I'm about to hang up, I realize this was not a real person.

>> I just spent two or three minutes on the phone with a bloody AI >> and something in the end like something that humans don't naturally say gave it away.

And I turned to I'm like this is wowing.

like you know the the feeling when you encounter it in a lab it's very different because you can say this is a uniqueness in the wild is very different and this is going to take up all the service sector jobs >> oh bo is dead

>> bo customer service this is not me saying by again the keyboard barriers right but the reality is it's not me saying look at vinod kosla's conversation now

if you doubt like if you question the credibility of vin kosa then I don't know who you are and where you come from so I'll I'll not get into that debate but >> Vinod Kosa yesterday I was watching a

podcast uh with Mint because he's in India right now >> right said BO is dead IT sector is dead financial sector is dead

>> every large uh >> pattern matching system >> every large company in the country that operates on the economy of internet is fundamentally dead

>> it is still on to a point right now because adoption is slower but adoption will happen in 3 years and it will be dead in 3 years of like that is a question that nobody has

an answer to >> right no but like from your education POV when you're teaching people like what are you trying to do to help them because >> there has to be despair right like despair confusion

but there is some place to go >> so that's individual level so that is what I recommend everyone I say long see there is going to be dystopian world >> and then there's going to be utopia is

what I believe and a lot of people believe and the reason why I believe this is because whenever this job loss happens and all of this happens right the world takes time to tweak itself to figure out what what needs to be done

but the world always figures it out right and this phase is going to be dystopian and it is very likely to happen or it's happening already because

of speed now in this dystopian space who are the people who are going to win people who are going to win are people are going to orchestrate things while or I don't believe orchestration is a

long-term win game because long-term we don't know what it's going to be nobody knows but if you want to win the next five six years because that's all you can optimize for at

like orchestration orchestration right I think the orchestration part will kick in in a year from now okay >> because it has only happened to code to an extent right now because there's a lot of building phase figuring out

that's happening >> tech is there >> adoption is taking time right so I think 5 years I believe 5 years is where orchestrators will win so optimize for

these five years >> and so what you then eventually end up teaching a lot of your students is orchestrator >> how to orchestrate a so that is what I say become a generalist >> right what is a generalist

>> when you're thrown a problem at you >> you need to figure out a way to solve it and you cannot say I'm not a software engineer you cannot say I'm not a product manager you cannot say I'm not a designer >> it's your bloody problem to solve it to

use AI to solve it. And how do you do that? When you have access to the right

that? When you have access to the right AI tools, when you know what is AI capable of, which AI is capable of what?

How to build agents, how to wipe code, how to build rags. And for any of these things, you don't need to be a developer.

>> So anyone in the world who can use internet who has common sense will be able to do it. It's an adoption problem.

>> So get out of that silo that you're living in saying subchangasi. It is not the case. M

the case. M >> come out and put in the effort is what we say.

>> Just in case the dystopia and the utopia doesn't happen, well, you're still fine.

>> Right. Right.

>> In case it happens, well, you have a winning hand.

>> How do I make you that one out of 10 who will survive this game is what we optimize for because I can only increase your odds. I can't make you win win.

your odds. I can't make you win win.

>> Yeah. Of course. Right. Right. Right.

Right. Right. So what you're teaching basically when students come to you PEL is tool use and then you are going into more advanced situations like VIP coding agent banana and and this kind of stuff

and then you're building a collection and is that the offering key then by the time you leave this educational situation >> you have it you get the context

>> see again we we don't say that everybody who goes through this will win people who execute are the people who will win Right.

So what do we do? We basically say you have to become an AI generalist. Here is

a five-step road map on there are five levels to it. Level one you learn basics of tool use. Which AI tool to use?

Where? Where do you use JGB? Where do

you use cloud? Where do you use Gemini?

Where do you use a notebook LM? Where do

you use NA10? Where do you use a replet versus a cursor versus anti-gravity? All

the tools stack that you have build your own AI toolkit. That is level number one. Right? You get surface to tools,

one. Right? You get surface to tools, right tools in the right place. That is

basic. Level two is doing the same beyond text. Get your hang around

beyond text. Get your hang around images, videos, audio and everything around it. Because as a generalist, your

around it. Because as a generalist, your game is not confined to only one set. It

could be any problem that you that could be thrown at you.

>> So get a hang of that part. That is the foundation. On level three, we say look,

foundation. On level three, we say look, you have figured out how to prompt. You

have figured out how to use AI tools and everything. In level three, you should

everything. In level three, you should get into the building zone. How can you automate things >> that you do on everyday level which you don't want to do? Let's say you don't want to check your emails. Let's say you don't want to respond to the calls.

Let's say you don't feel like making a call to anyone. You don't want to order pizza. It doesn't matter what it is.

pizza. It doesn't matter what it is.

There are 10 things that you do, seven things that you don't enjoy. How do you automate it? Well, there you learn

automate it? Well, there you learn things around MCPS, model context protocol or there you learn around how to build projects, how to build rag systems, right? things like that or how

systems, right? things like that or how to build basic NAT automations or basic Zapier or make automations which can automate things for you >> right and you can do all of this without

code you don't need to be a developer for any of this right that is level three once you have understood automation automation still needs you to

come in and say hey do this for me how can you automate it without you saying anything is an agent >> right >> it does end to-end task >> so In level four where we say learn how

to use agentic software and also learn how to build your own AI agents. How can

you use a tool like Lindy? How can you use a tool like NA10 again to build agents or Opel right now or any of the agent tools out there? There's a problem that you want AI to do. How can you build an agent for that? Today the

worlds of openclass opened up which is absolutely bonkers.

>> Right.

>> Actually tell me a little more about that because that looked a little bonkers when I first saw. I

>> I'll show you also.

>> Sure. Sure. Sure. Huh. Right.

>> Right. That's level four and level five is finally wipe coding.

>> How can you productize stuff that can work beyond you as well?

>> So these are the five levels we basically say people to cover. What we

can do in our sessions at scale with tens of thousands of people on them is we directionally tell them key we teach them a few and we give them a road map.

Now people who actually win are not the ones that take our learnings and say thank you so much this is helpful.

People who win are eventually who learn from what we say, take the road map and continue their learning.

>> They are the people who will win. But

that's what we do on a high level.

>> That's very interesting.

>> Yeah. So this is what we feel is a full stack AI first problem solver can do and we've seen enough and our employees are our guinea pigs to start with. They've

been doing incredible work right now.

Right.

>> Right. And uh yeah >> my question see the adoption problem is effectively a trust problem right or it's a lowest

hanging fruit kind of problem like if somebody saw the benefit they could get >> with the lowest lowest amount of effort the likelihood of them adopting it increases quite a bit.

>> Not every time.

>> Aa that's not the case every time.

>> Not every I'll tell you a case when it's not every time but yeah go on.

>> Right. What I'm trying to understand is for most people that are watching either they work a job or they have some kind of a freelance self ownership kind of a thing, right? What's the easiest

thing, right? What's the easiest possible thing in your experience they can do to get ROI on that experiment?

>> I think uh the easiest thing that you can do right now is take a paper out, >> write down 10 things that you do every day which takes 80% of your time, right?

and just take a picture of that and upload that inside of Gemini and say hey based on today's technology and please turn on the pro mode on Gemini it's free anyways so or chat GPD or whatever

whichever AI tool you use upload that picture and say look these are the things that I do which takes up 80% of my time based on the technology that exists today in AI what are the things

that I can automate you will see 80% of the things that you write no matter who you are and what you do 70 to 80% for most of Chad Gibb or Gemini will give you an answer saying these are the things that

you can automate and this is what you need to do take one at a time you don't need me you don't need this podcast you have see the beauty of the world that we living in is content is democratized

>> right >> knowledge is everywhere >> right >> we are just not seeking for it >> right right so you just do this isn't this is this very hard to do not really

>> just do this and the moment you get point number one and you get a descript description you're like oh this is interesting I want to go deep into it copy the whole thing paste it back into chat GPT and say give me all the bloody

resources out there which can help me to teach this in fact it's better if you go to Gemini because Gemini has access to YouTube >> right this is what it is make me a list

of uh YouTube videos so that I can learn this and automate this you'll get a you'll get a bunch of YouTube videos and when you're trying to implement something if you get stuck

somewhere go to Google AI Studio You can share your screen, say that, hey, this is how my screen looks like. I

got stuck. I'm not able to set this up.

Tell me what needs to be done. There's

an AI tutor on the other end which will literally tell you, click on this button, tap on that button, fix this up for me, fix this, do this, and it'll fix things for you.

>> If you're slightly advanced user, I would not even do it. I will just message it to my Jerry, which is my executive assistant, which will figure it out itself by opening its own browser and doing everything for me. And Jerry

is your tech >> executive assistant which is open claw right now.

>> All right.

>> Which is open claw right now. I just

switched it to openclaw. Before this it was on n >> but n now.

>> And where is your assistant? Is it on?

Oh, on slack. Okay. Okay. Okay. Okay.

Oh, so aa >> interesting.

>> And it can it has agency enough to go outside Slack and wherever else as well.

>> Yeah, I've given it full control.

>> And you've used what device for that?

>> What do you mean?

>> Like it must be ported on a device now.

Yeah, >> so it's it's on my Mac.

>> Ah, it's on your Mac. It's on my Mac Mini. A lot of people for OpenClaw take

Mini. A lot of people for OpenClaw take a different Mac Mini. I've been a little >> braver here and I said I'll put it on my main main Mac Mini. Let's see what happens. But right now the Mac Mini runs

happens. But right now the Mac Mini runs all the time >> and sometimes I just make sure Open Team Viewer to see if I'm away to see it's not doing anything nasty inside.

>> That's crazy.

>> But yeah, I've trusted a little more than what I should. I I was like cancel next Twitter cancel.

>> Uh but that's that's crazy bro your PC is operating on its own all the time.

>> Yeah. Not all the time only when so most of things are automated basis of what I want like Slack is my biggest time monger so that is automated. every 3

hours it sends me a message saying hey I check Slack and uh you know these are the things that I found nothing to worry about you can continue your work that's what it says unless there is something important that I have to attend right it

checks my emails right if I have to talk to someone and I'm too lazy or they're not picking up my phone I call you you don't pick I offload it to my saying keep calling Plucker and pass this message till it act till plucker

actually responds back you can see Jerry has just messaged me something it's proactive by the way it messages me proactively so So it has sent me a message. Oh, Slack scan. I told you,

message. Oh, Slack scan. I told you, right? It does a Slack scan. Don't read

right? It does a Slack scan. Don't read

things out.

>> No, no. Aa

>> every 3 hours it's proactively sends me message.

>> So it's basically given all the important details that you must attend to in your Slack in one message so that you have a compression of information.

You don't have to read through.

>> This is one thing. This is one thing.

This is one thing. So yesterday one of the use case was I was in the India air summit >> and uh I wanted someone uh to kind of uh talk to and we were supposed to meet and

I was realizing that >> you know he will call me I will not pick up and then I'll call him he will not pick up >> because you're both no like because we are in the India summit and it's a mess

and whatever. So I just said till this

and whatever. So I just said till this person picks up keep calling him and tell him that I'm at the meta booth till he acknowledges the message keep calling him.

>> He got the call seven times.

>> H >> he got the call seven times from my AI and it outright says it sounds like me >> but my AI outright says that hey I'm web AI executive assistant. So I'm not

trying to be Webhub like >> right right right >> and it says hey webhub is in the meta booth he would love to see you there and he'll be there for the next couple of hours >> right >> right first time he picked up he could

not understand something or he said something and he cut down because he thought it's me and then it again called back to him I think it called like five six times and then finally and it was going from my international number because it doesn't go from my number

>> right >> right so it finally picks up and says webhub this is webhub AI executive assistant please come into the meta booth and that guy was like, "Yeah, I'll be there." And it stopped calling.

be there." And it stopped calling.

>> So this is these are very small things if you think about it.

>> Sure.

>> But uh it can do almost everything.

That's what I'm trying to say. Let's say

uh before your podcast, right? Before we

do this, I was like look like uh what are the things that are not spoken about? Are there things that I should

about? Are there things that I should speak about based on the things that have happened? So it did a research for

have happened? So it did a research for me and gave me a bunch of things like for instance that uh like the uh the GDP index that I told you GDP the GDP val

had a big difference right now that happened yesterday so I was out >> I was not aware so it pinged me saying bro GDP val numbers have gone up by 300 points >> you should talk about it

>> and so your project knows what you've spoken of in all your media appearances >> not the project this is uh so these assistants these agent Agents are 100

times or 20 30 times more complicated than or more advanced than a project.

>> Okay.

>> Right. So,

>> so you have an agent that goes out and collects all your media presence and what you've said.

>> I have done it once and stored it in the memory.

>> You stored it in the memory and now what happens is that it takes the proper set of all things you can say versus the things you've said already and then tells you what the differential is.

>> Yes. And it also has told me what I should not say in my podcast. Like for

instance what cancer like there's like listen if you're saying if you're quoting something quoting a number make sure you quote that it's not yours right okay

>> so proactively before this also I used to say it but it'll come in the end or it'll come in the beginning and I know what you mean 30 second clip right you optimize for the space Right.

Right. Right.

>> Right. And you you don't think of your editor is not thinking of that oh this clip will get canceled out.

>> Right.

>> You have to think about it. And I'm not Samman that my policy team is sitting behind my ass saying that I want to see every clip that goes out.

>> Right.

>> And a content will go out without that little nuance of context >> and it'll cause all the trouble.

>> It'll cause all the trouble. It has

caused all the trouble so far.

>> Oh, you've had that?

>> Yeah. I I mean not that bother it affects me, >> but I'm like my girlfriend was like just just be a little reassuring.

>> All right. Just say that it's not you, it's Vinod Kosa saying and say it three times.

>> Yeah. Right.

>> Just say that uh just outright say if you're quoting based on GDP value.

You're thinking that you're quoting based on GDP val. But just know just remember it and say it.

>> And these are all nice little instructions that have come in from my PR.

>> Give me some uh >> which is an AI.

>> Yeah.

>> Yeah. At this point I'm not certain if I'm sitting with you at all. Uh but bro, you actually do live in the future. Like

frankly uh I have used artificial intelligence. I know some people who use

intelligence. I know some people who use it to quite an extent but I think you're really maxing. You're AI maxing.

really maxing. You're AI maxing.

>> I I I would like to I think I'm AI maxing for who I am as a non-software engineer. So I do a lot of things that I

engineer. So I do a lot of things that I possibly can.

>> But I think a lot more can be done. If I

was an engineer I would have been able to do a lot more.

>> Right.

>> Uh but uh I get my software engineers to help me. Is there is there any value in

help me. Is there is there any value in the future world for degrees and education in this case because I would suppose so right >> I don't think there's value for degrees and certificates even today what is valuable is the experience that you go

through right >> so I went into four years of college right I didn't go through a great college like like not an IIT or anything like that's what I mean not by a very well-known college you went to Colombia if I'm not wrong

>> right I don't know your reflection of that but I've always heard you speak education of very highly right >> I'm a huge fan >> I know I M >> I know but I'll tell you that I don't

dis education I don't say anyone not to go into education or not to do education not because or I never say not to go through this experience of education

purely because it teaches you more than the books >> sure >> books have not taught me anything even before a era forget about a era I don't think my 10th 11th 12th has helped me

with anything that I've done so far I think school education is different I don't know how it has helped me And I'm sure it has helped me in some form of fashion that I can't understand.

Engineering is when I'm capable enough to know what how it has helped me and has not helped me in any way of what I have studied. Right? Maybe the only

have studied. Right? Maybe the only thing that has helped me is code because software engineer I mean computer science engineering doesn't mean I've written code >> right >> that's the only trait that has given me but that that didn't need four years of

engineering but what it gave me is exposure. What it gave me is how to

exposure. What it gave me is how to manage people. What it taught me is how

manage people. What it taught me is how I can be with myself independently. What

it taught me is that I can take bets.

What it taught me is how to make friends. What it taught me, what it

friends. What it taught me, what it built me is a network of incredible people that I can still rely on.

>> Right. Those things cannot happen without a college. It teaches you life.

>> Sure.

>> It's not so much about the books. It

never works.

>> Now, I don't This is education in India.

Right.

>> Right. Right. Right. It's gamified and I get the frustration.

The minute that thing happens, once you're mature, you look back, you're like, but that's the that's >> and also one interesting thing that I've seen with education is that it's the

correlation of your grades to your outcomes in life.

>> It's really not true. Huh? It's

variable. I know some really backbenchers who've done phenomenal work and some front benchers who've gone backbencher territory. Right. Yeah. Yes.

backbencher territory. Right. Yeah. Yes.

Yes.

>> I mean again I'm not quoting see most of the things that we speak about education goes into dusbin when you're talking about the top 2% or top 1% like like

there is no way to kind of even debate that hey if you are going through an IIT I am MIT Colombia Stern whatever Harvard Stanford there the game is different.

I'm not even I don't think we will ever be able to have this debate saying that it's not worth it. M

>> I think it's always going to be worth it because of pure density of incredible people.

>> Mhm.

>> That's it. Period. Right. I think that is invaluable according to me. But

anything below that, when you go to the 5% 10% below 10%.

>> I still think everybody should go through the whatever that percent is, you should still go through it. But you

should be clear about what you're getting out of it.

>> Yeah. Education. I think for India Gorov actually Goravjal changed my view when I heard him speak on a podcast and he explained it's like there is only one game that promises social mobility in

India degrees becomes a permission slip for you to be like I can apply to a multinational corporation in Gura or in Bangalore get

that job socially move up and whether people like it or not because these the tech crowd tends to be very philosophical whether people like it or not fundamentally Life is about status

and so if a person sees I as a father see acceleration there is no way I will not tell them to go to college everything else you know if you've ever been friends with athletes in school

badminton they make this very clear distinction at a certain point >> and the reason is because even though sports top 1%

your life is set beyond any >> but the odds that you will get that very low >> but high if you're able to manage a degree and get through to even >> a simple company that does a simple job

and you are a simple accountant in that company your social status guaranteed accelerates more >> he's right and that is true like if you want to get into a TCS Vipro Cognizant Accenture blah blah blah all of that they have that tick tick box that you

need to have a degree >> right but they will not exist 3 years from now in the way they do if they continue to operate with this framework of degree being the Why? Why did they

have to have that? You have to go back to the genesis of why these things existed? Because in a country like ours,

existed? Because in a country like ours, there are millions of people who want a job, >> right?

>> And you cannot have such a wide funnel.

You cannot interview so many people. So,

how do you reduce the funnel of your inflow? By putting barriers,

inflow? By putting barriers, >> right? Huh?

>> right? Huh?

>> So, this became a barrier. Right.

>> Right. Right. Right.

>> It's a terrible barrier.

>> Qualification criteria.

>> It's a terrible terrible barrier. Right.

Now if you think about these companies the reason why people generalized careers to top five IT companies was because they were were the biggest

hirers in the country but in the last 12 months do you know how many people have how many net employees have these companies added top five companies in India together >> I have no idea but I I don't think it's

a big number like zero negative >> 17 satra so like topflow satra net net that's why like outflow plus inflow >> right is net

>> 17 top five ID companies I will not name them you know what they are yeah >> these were the mass recruiters right right >> right what are you optimizing for boss

you are not living in a world of tick boxes anymore >> so if look like will I still say you should go and get a go into education I I will still say you should go into education because when you're that

young, you don't know what you want to do.

>> Right?

>> So this gives you direction in some form or fashion.

>> Right?

>> You go through the four years of experience without your parents around.

In fact, I recommend everyone to go to a hostel.

>> Right.

>> Right. Because the world that you live in is different. You should be as far away from your parents and your comfort zone as possible because you learn life.

But you should optimize your time.

Doesn't matter which college you are in.

Content, information, knowledge is democratized. Everything is out there

democratized. Everything is out there for free.

So play your Counter Strike, watch your Netflix, you have four years to chill, but do something with your time.

>> Build things because this is the era of builders. Everybody in the world can

builders. Everybody in the world can build right now. Use your time when you're in college to build stuff. Fail.

Look, look, I think for anyone to succeed, they have to fail at least 50 times. Micro, macro, big, small failures

times. Micro, macro, big, small failures all together. If you're not failed

all together. If you're not failed enough, there is no way you'll succeed.

Even if you succeed, that will turn into failure.

So increase your failures when you're in college.

>> So by the time you come out of college, you have seen what failure looks like.

It doesn't hurt you and your odds of winning just goes up.

>> You know what my view on education particularly from my experiences student and CS CA charted accountant? I

was a charted accountant student, right?

Um and >> had I not gotten into an Ivy League university, I'd have probably finished, right?

But complain day one say or for whatever reason I was a little more first principal. I was like

principal. I was like how can you punish me for calculation when a calculator has been invented? Why

would I be punished for arithmetic that I can do in my head? Right? And so these exams that value how you pass or fail will get gified along incentive plus

disincentive to negative right.

>> I found a lot of beauty in reading a wide variety of subjects that opened my mind for the first time.

>> What did you study?

>> So my core education was in psychology and economics but I did the 3D route where I sat in religion film right. So I studied physics, I said

right. So I studied physics, I said biology which I would have never studied. My accounting student, right? I

studied. My accounting student, right? I

studied uh religion, anthropology, I studied mathematics, bro. I took

advanced courses in mathematics I would have never taken.

>> You got the ROI, man.

>> So I I was like and um they offer some very interesting classes. Liberal arts

side to help you learn how to think that end.

>> Why I'm mentioning this is in hindsight, I wish I'd started building sooner. See,

I started building in college. I I did and even still I wish I was like five years younger because five years compounding is insanity but the partal if you and I sit and compound on the

thing we're building it's insanity what can happen and you're right you take 50 failures no matter how smart you are no matter how well positioned you are you take 50 failures that no most people don't see it happen silently before

you're like right and so life is such a long fight in any which case you get also very bored of exploring in the way that college students spend for the first four or five years and then you want to do something

>> because you have nothing to lose during college. No, right.

college. No, right.

>> That is the beauty.

>> I think it's the best bloody time.

>> In so many ways, it is. Right. There is

when you meet older people, they'd be reminiscing about college and you're in college. You're like a fight. Like, you

college. You're like a fight. Like, you

know, and then you were older and you're like, man, they were so right.

Now you're the older people, >> right? Now you're the older people.

>> right? Now you're the older people.

You've hit 30 like you're status. And

you look back, you're like truly if I could tell somebody in college or school what they should do is they should enjoy it more.

>> Yeah.

>> Right. Such a value.

>> I had written a note on this. I was like I worked too much during my college.

>> So you >> I had three startups during my college.

>> Oh wow.

>> Right. It failed. I started one more that failed. I started one more. I was

that failed. I started one more. I was

bloody working all the time not going to classes and all.

>> So I I like when after four years when I had to write a note on like somebody asked me this question in one of the talks and I had to think hm I was like I wish man I'd gone out for

the dinner with my friends because now I don't have their time then I did.

>> That's very interesting. I sometimes

think the opposite. Yeah, I wish I started early and this is the nature of humanity.

We're never happy, right?

When I see my colleagues who are now successful because they've worked in media or tech and then they're spending money having fun.

I had the exact opposite experience. I

had so much fun.

Like if my wife ever asks me, "Where do you want to go travel?" I'm like, "I've seen whatever I wanted to see. Right

now, I just want to build." And so I exhausted that bit where no extra dinner with my friends is going to make me happy. I'm sick of my friends.

happy. I'm sick of my friends.

However, in the 7 years that I lived abroad, I didn't live close to my family. So every extra dinner with my

family. So every extra dinner with my family makes me happy. And this is human nature. Now, whatever you don't have,

nature. Now, whatever you don't have, >> everybody's trying to find a balance, >> right? Whatever you don't have. Yes.

>> right? Whatever you don't have. Yes.

>> Right. But um it is true I also graduated in 22 >> college that is the year Chad GPD hits and so my

education is pre GPD and I had to write that essay and I swear to God bro I don't think anything is written it completely right I'm very skeptical of

the fact especially writing for writing sake I don't think you should delegate that.

I think you should write it. I I think I I I'll tell you why it has become 10 times better >> for writing using AI while you write it yourself because there are two things of writing, right? One is that how you

writing, right? One is that how you write, how you craft it, that's a skill that very few people have. The thing

that I think everybody, the reason why everybody should write is because that pushes you to think harder. Yes.

>> Right. That thinking harder can still happen. In fact, the thinking harder can

happen. In fact, the thinking harder can become 10x better with AI.

That's what I'm saying, right? Have if

you were writing that essay, did you speak with 10 different friends?

>> Well, you go to 10 different sources, you read 10 different things, you come up with opinions, you synthesize them.

>> Once you had opinion, did you put that opinion in front of a friend saying, "Yeah, what do you think of this? What

is >> It depends on which context we're talking about."

talking about." I I'll go back to an example that everybody would relate to or at least most of them would like let's say I have to put a slack message like because you took the example of Slack message right

>> right I got I get AI to do it >> but it starts with a voice note into my claude or chat GPT or Gemini or open claw or whatever it starts with a voice

note saying look this is what I want to say this is how I'm thinking these are my raw thoughts.

help me organize this >> and also as I as you organize this also tell me what are the downsides of this will this affect a set of people that it

should not affect should I smoothen it out in a few places >> right >> so that because you are doing let's like right now if I put a tweet out it still

goes out right now but that tweet is PR vetted to a point where my thoughts are going out but it's not offending people also Right?

>> So I'm like every time I put something out, I write it. I put my voice note out and I say that look, I'm going to put this on Twitter.

>> Will this create problems or will this bother a few people? Will this lead to negativity? If that is the case, why?

negativity? If that is the case, why?

>> And how can I soothen it out while putting my word out?

>> Oh, right. So it does help. I mean, unless

right. So it does help. I mean, unless you are someone who's like, "No, I want to put it raw, which is fine." No, I meant more in the sense of if you're

given base writing work but I in my head is whatever I it's emotions it's some ideas but nebulous once you put it on paper this is a type

or you type it which is what paper means what really happens is thought organizing or that is what I do as well I do it on voice note and you do it in typing like it's thinking

>> actually voice notes also like most of my AI consumption is voice where I speak to it and it sends me text back, right?

That's the best way because I can otherwise once I start speaking, it organizes itself in my head.

>> I mean, quite frankly, I use a tool called whisper flow. I don't know if you've heard of it.

>> I have heard of it. I don't know what it does.

>> It's magic, >> right? It's voice to text. So on my

>> right? It's voice to text. So on my computer, I just put a function tab and spacebar. I have my earphones plugged in

spacebar. I have my earphones plugged in and I'm walking and I'm talking and it takes all that information that I'm talking, condenses that into nicer format and just spaces it.

>> That's crazy. Yeah,

>> that's crazy.

>> No, but this is very popular right now.

It's I think it's a no-brainer, >> right? And

>> right? And >> for us who have the like on a phone, I'm sure you use a lot of voice notes.

>> Sure.

>> Right. It's very instinctive on a computer as well, but we never used it because dictation is terrible.

>> The tech of dictation has been terrible.

Right.

>> But uh companies like whisperflow and all are coming which are working purely on small language models which are optimized for transcription and it's beautiful. It's beautiful. So I would go

beautiful. It's beautiful. So I would go on my LinkedIn let's say I've got a message from someone saying hey I was supposed to get this email from your team and I've not got it. I get a lot of those messages because of the scale that

we operate at email.

So it's the same standard reply I have to give every time.

>> So I just put function space a function button and I just say support response >> and it just autofills the whole thing.

Do you remember the text auto right now it's happening via voice >> my LinkedIn URL fills up my LinkedIn URL actually https/lin.com/in

actually https/lin.com/in same goes with if I say my social handles all my social handles it autofills that so little little things but it saves up time and I think when it

comes to context a lot of times also this happens with me I don't know if it happens with you that as I type I forget what I'm typing so I have some thoughts

that I want to type out but as I start First point, second point, third point.

But with voice because it's so fresh and words per second, words per minute that you can type is like 30 40 30 40 words per person. But voice is 250.

per person. But voice is 250.

>> So it's six times more efficient.

>> Right. Right. Right.

>> So your context window, it helps your context window as well.

>> But yeah, anyways, whisper is quite cool. So any of those tools, a bunch of

cool. So any of those tools, a bunch of them, >> any of these tools are quite cool.

>> Nice. Um, what else are you thinking about these days?

>> What else am I thinking about these days?

Yeah, I barely think about anything except for AI these days quite frankly. uh again

right like I told you the the reason why we spoke so much about health is because grandmom father so that is one I don't >> but that's it work and a huh

>> work work and family and uh that's pretty much >> tends to be like that for men after a certain how old are you now >> 30 31 huh 31 as of >> like your birthday's already been this year

>> September >> 30 31 running into 32 >> [ __ ] you >> well man I just turn 30 myself.

That's why I stopped at 30 and then I was like you question please please context window. That's crazy, bro. This

context window. That's crazy, bro. This

has been a good chat, man. We should do this again sometime.

>> Yeah. How long have we been talking?

>> 1 hour 40 minutes.

>> 1 hour 40 minutes.

>> Nice.

>> Yeah. Did you Did you feel like the time passed?

>> Yeah, it was quick.

>> Yeah. Good stuff, man. Uh it's a lot of information, frankly. I'm just realizing

information, frankly. I'm just realizing how how much there is to do with AI.

Like, you know, um even even myself who uses so much and I know so many people who use so much like listening to you, I'm like, "Fuck, >> what are the top things that you use AI for?

>> Mostly thinking.

>> Um most of my job is either strategy or research right?

Both of which are purely intellectual jobs. M and so it's mostly this this is

jobs. M and so it's mostly this this is what I'm thinking what's wrong with this this is what I'm thinking what's wrong with this if we do this run a simulation and tell me right >> have you come across this uh a Chinese

company they launched a AI model which simulates human decision making so so let's say you have to take a decision and you don't know what the

repercussions are going to be let's say I have to take a decision in my company that 200 people I don't know what the repercussions will be because everybody have everybody has their own kind of uh uh you No repercussions on any decision

that you take, right? Ah, this one >> simile. Huh?

>> simile. Huh?

>> Simile.

Yeah, it simulates human behavior. So,

it basically runs a simulation ecosystem with a lot of AI humans, takes those sessions in to see what will be the impact of it and eventually gives you a

report that you know based on this human simulation what you're thinking is right or wrong. And say if you want to

or wrong. And say if you want to simulate hey like I mean this is too big of a simulation to be done but let's say can universal basic income work in India >> right so it does the multi-order human simulation >> simulation

>> oh [ __ ] yoga free yoga free >> I was thinking that this is the only thing that might not be able to work because someone has to simulate the simulation experiments every country every government does it

>> right but it needs such a massive infra they give it to a Mckenzie PWC or someone to be able to do it >> right right >> right but Now there's a Chinese company which is doing it.

>> Yeah, >> I saw a video. I don't know if I don't know why it's not uh maybe this is not the actual website. I don't know. I saw

a video or just look for this video.

It's it's wild but again you said simulation so it >> so generally that is what I'm doing. So

for for me I found more utility in being able to express and then there is a bouncing partner otherwise you would not have had um research fairly simple on any of these

but beyond very little like you know management so to some extent you can think of setting up systems manag but very little what else

>> so you're not using in your content you're not using AI in in your work work per se.

>> I mean, yeah, sure. But it's not me directly using it. Then my organization is using it. They're bringing up scripts for instance. Say I we cracked this

for instance. Say I we cracked this system of creating a 50 million view funnel on Instagram which is completely except I have to record. I I I think the betting is in two direction. Either you

bet on the direction that all content will be AI and you do AI avatars or AI morphs like that's becoming very popular or you realize key at the edges

personality will win. So power and if you stand out then people will value your personality even more and so from a

career POV we pick the personality route reasons um I think if people know it's AI value alignment >> no by the way just to tell you I have not picked a side here

>> right >> I create content using AI because that's the only way I can create content >> and that's also your business like it's very aligned with your value you or Vun both value is >> actually I'm not thinking on those lines

at all. So my direction thinking was if

at all. So my direction thinking was if you remember all of us started content creation at the same time in some form of fashion during pandemic we all met and Raji you know we've had that conversation Raj's podcast as well about

you right >> you guys you went into education you found your way in content you're extremely good at it right so is Raj is phenomenal at it and people took their

directions I took the business building route and because of that >> I was not able to create content at all while we all started with content I see right I didn't have the time to do it or

this was never my PR.

>> So AI enabled me to put my thoughts out there in a format that works for people.

>> I see for scaling purposes we've thought avatar because you know and this is like way before and we were like hypothetically running it case if you're building if you're

becoming an influencer which is not the influencer of the old time influencer as in somebody was influenced which is a huge thing by the way correct.

emotions, people will not align themselves with your values and therefore you don't build any influence.

You may build an audience and you know in content creation as a profession I think one of the missing differences is what influence do you have versus what audience do you have and they're not the same anymore right or uh influence building which is which has a 10 year

long ROI is not the same as audience building because that has only platform ROI so we went over this discussion I was at network school in Malaysia and I was talking to a bunch of people about this because you know it's such a vibrant tech environment and that's when

we were having this conversation key AI avatar side you can actually get a lot of information across if you're doing a news channel AI is best because you need speed use case different but if you're doing a

podcast and all my questions are done by AI for web hub then I'm not different from almost anybody else except I learned a bunch of new things about how you can prompt yourself better um I do

think however um so we were designing a titling testiment we arrive at some kind of an optimization strategy that strategy

would taken us a day or two to build in terms of check mark.

What I've realized about content is generally it's very what's the word you're playing with the average IQ of people respond for. If you give AI enough

respond for. If you give AI enough context, it'll set the strategy that can be executed on its own. What I learned today is I can probably also automate this.

>> Yeah. So that AI checks itself what is happening after 24 hours instead of somebody manual doing it >> and that's the evolution I'm realizing from this conversation but use so a lot of it is just text usage we haven't done

agentic or wipe coding usage because it's not being very important but today I'm realizing processes automate that's very interesting >> no bro it's I think that is a real

arbitrage >> I think you have an angle in there in that I think you can be hired by organizations to improve their workflows in this too. I don't know if that's something >> that is not some I if if I start taking

that up that could become such a large business because we get insane number of inbounds >> right >> but I what I've realized over the course of time is if you try to do 25 different things you'll be good at nothing

>> right >> so I've ch I'm trying I've learned it the hard way that you should learn how to say no >> right even though it costs you on short term >> but I think it's a lot of money bro

maybe more than what you make because organizational enterprise customers That's what I'm trying to say, right?

It's like right now I have not been optimizing for money at all.

>> Okay?

>> Because I know if I have to make money, there are 10 other things that I would do right now which will make me a lot more money with where I sit today.

>> Like for instance, >> I can't put it on camera. Okay.

>> Right. I can't put it on camera.

>> So you have some like AI hacks.

>> What do you mean? Like I mean I'm just joking but what I mean is like you know some of those guys were using it on poly market and they've created these strategies. I

mean I meant something like that but I was joking. Huh?

was joking. Huh?

>> I know I know what you're saying. I

think uh >> No, not for this conversation.

>> I think you're smiling. I think you're smiling. I think you're smiling. There's

smiling. I think you're smiling. There's

something you're hiding. Huh?

>> No. No. Uh but I I like what I truly feel is uh you know anything that you believe or that you want to get done today can be

done.

You just have to double click on each of them and say that I want to go deeper. I

want to go deeper. I want to go deeper.

So uh for the titling experiment that you did right because we have >> so if you see our content ecosystem that I've built out I don't know if you follow my content ecosystem at all.

Everything is AI. Like I said, Instagram, we've got to one and a half million. Uh, and we were at like 150K 16

million. Uh, and we were at like 150K 16 months back, 15 months back. And the

space is not easy. Education is not easy at all.

>> And average views for us is like 600K, 700K per video. YouTube we started same 300k subs 70 80 100k views on an average

on AI content which is again globally not many people are able to drive all this has happened because of agentic systems so the the title test that you're talking about right there that's

a very interesting experiment that we did in fact we are running basis of that >> you know vid IQ right >> I have yeah >> vid IQ has outlier scores and on every topic you can see what topics win and

outlier scores and all I told you right browsing a browsing agents >> browsing agents are AI using a computer just like you >> they are called browsing agents and

there's a benchmark called as OS world right now and AI today can browse websites that you have never browsed before let's say you come across a new website >> you figure out how to browse it you

figure out how to use a video vid IQ >> right >> AI can figure that out better than you by using a computer by opening its own computer and doing it so even if There is no API AI can use that website like a human.

>> Got it.

>> So we have given the you know uh browser access to this. So it scrolls through every single day a chron opens a browser puts the 10 15 20 keywords that we

create content on searches for it just like a human on opens different browser for so it's called agent swarm. So a

agent a browser use open for a specific keyword goes through all in the last 24 hours. looks at outlier scores and

hours. looks at outlier scores and basically does this analysis of title, topic and duration and thumbnail

parallelly for us.

So I'm looking up my AI is looking up content >> and how well is that done for you?

>> What do you mean how well?

>> Like are you able to measure if it's done well for you or not?

>> And it's working for us.

>> Sure. Working for us but how much like do you know the difference that it has caused? So uh so far I mean it this

caused? So uh so far I mean it this basically dictates dictates our content strategy.

>> So it dictates from the very start.

>> Right. Right. Got it. Got it. Got it.

>> I mean you same goes with uh Instagram.

>> Right. Got it. Like AI is basically watching I don't know tens of thousands of pieces of content. So the thing is if a AI content on any topic has gone out on the internet in last 24 hours I have

a data point on that.

>> Very interesting.

>> At least of the ones that we are finding. So we eventually end up finding

finding. So we eventually end up finding newer profiles >> and the newer profiles gets added. So

after hundreds of scrolls then it's very hard but I'm tracking you for life my bro.

>> Oh wow. Damn. You're dangerous.

>> No like this. This is I'm basically saying all this is possible and you can apply things like this no matter what you do content because you said >> right. Right. No. Of course. Of course,

>> right. Right. No. Of course. Of course,

of course, of course, of course.

>> But I think what you guys are doing is better than 99% of the people already, >> right?

>> Kudos.

>> Thank you.