🚨MISS AVVADDU… AI AGENTS CHANGING EVERYTHING! 😳| ft.Akshay Sharma * Uday Kiran | #podcast #ai #life

Ajouté : 2026-06-08

[00:00:00]Smartest people stuck the longest.

[00:00:02]>> Okay. Why? I mean in a specific company.

[00:00:05]>> Yes. Yes. Yes. Many people it's like >> let me take an example of myself working in Microsoft. Yeah. Stable income and suddenly a layoff happen.

[00:00:19]I'm thinking that is the best thing happened in my life. Why? Because AI was there from 1960s itself. There is a particular phase that went artificial general intelligence, artificial neural intelligence. There are many things that were come. This is one part of it which came in which is like from using the LLM's training and predicting models which is more related to generative predictive model which are pre-trained transform GPT we call it. CH GPT is like generative pre-trained transfer.

[00:00:51]Gemini everything has their own credibilities everything can perform similar kind of action but they're more confused about what to use when the best for the designers Why? Because generation images for developing images is doing great.

[00:01:45]Open it. Good.

[00:01:51]Hello everyone, welcome to classic connect withan. So before we start the podcast, I request everyone to grab a coffee, a pen and a paper because this podcast is not a podcast where you can listen in the background while you scroll your Instagram feeds or Facebook feeds. But this podcast is specifically designed for the people who want to enhance their skills in AI. So what is happening in AI? how do I enhance my skills and you'll learn lot of things that are happening in and around the world and making sure that you pause again you listen back so my sincere request is everyone to grab a cup of coffee and also grab a pen and paper with you while listening to this podcast and I'll make sure that this podcast will give you lot of insights what is happening in and around the AI how we are using the AI are we using in the correct way or how do we enhance our existing AI skills Today we have a our guest Akshai Sharma who has worked with many bluechip companies like EY and Microsoft and has started building AI products. He has wider experience on developing AI products in a manner where the customer accepts and increases the customer satisfaction. So with this let's start the podcast.

[00:03:08]>> Hey hi Axi.

[00:03:09]>> Hi there.

[00:03:10]>> First of all thank you for accepting for this podcast. I know you're very busy in building a products and tools right but even we want to enhance your knowledge to the people where they can also understand what's going on with the AI in near future and currently so first of all thank you for accepting this podcast coming way along and hopefully this will add a value to the wider audience irrespective of ages and gender >> definitely it's my pleasure to be part and it will be helpful on the people also keeping us part of exal aluminia of IIA and IM also working at Microsoft. It comes with lot of experiences. I'm happy to share my experiences and share how the AI is actually turning out and we'll be discussing in like how the overall journey like from my childhood experiences and how things are going on and differently on how I can actually value add for the viewers at the end of this podcast.

[00:04:05]>> Sure. A let's start with a simple question. We have heard people saying that especially parents We need to get settled.

[00:04:17]So don't you think that the settlement word is no more required as of now? Very good question.

[00:04:28]Correct.

[00:04:34]overall process flow mainly when we looking into how the overall transition is happening I feel that AI world has its own merits and demerits but how do you feel a settle right now job security it's not there and if you see financially it's very tough for the >> exactly so how do you find settlement so I feel important settlement what do you mean by settlement correct day in day Sole it's not word itself is not settled if I'm correctly understand yeah settlement overall process got what I can learn what I can do it how can I exceed But man, you played a perfect playbook man. So in our entire generation or in our entire friendship or the friends that we have, right? You played a perfect playbook like every parent would have wanted that like getting a good score in intermediate then becoming an IM grad working in a big companies like Microsoft and EI every parent want that they aspire that their children to become like that but my only question is when did you realize that this playbook was outdated that's actually right I think thoughtful question even I I was thinking about it so would I mainly I feel that when we started I personally failed that when I started man I still remember my fifth class or sixth class grade my mother was a principal and my brother is like a topper of the class I am from the bottom I'll be first or second >> so I still remember my 100 marks in quarterly or half a year should be in single digit like nine marks 11 marks sort of thing so >> one thing my mother believed me saying that he understands max okay >> so he can survive sort of thing and at one point of time she also was thinking actually let's uh redo one class or two clangas the director at that point felt that akshai knows max so he'll survive so I had still discussion on seventh class during my board exam that if you don't do anything also it's fine I'll keep a kirana general kirana stores we'll invest money and we'll we'll sort it out so I feel that it's more like a parenting so never that settlement came into our discussion with my parent man that comes with also she's being a teacher principal all together she understands me better so always that pressure aspect did not kill. So that pressure actually made me to perform better in 10th. Then I never looked it back saying that what should I do properly because after 10th I felt like engineering would be like 10th and inter I have I was planning to do I was growing. So I felt that at the early stages it's always to give a free hand you never know what you required. So that is where I felt that my career transition happened from like 10th to inter then after I pursued two years for IM Kodi code. It was a enriching experience. Then it was not direct transition. I directly after MBA it was like complete different world. I had to learn so many things because you come with a lot of like you see all the grads from IATS NITS performing like crazy and you find you are like okay fine I'm from a local college here >> like it's a vi but it's a local institute in the city and then directly go it was a struggle then it felt like every day I was learning like saying that okay I need to get into better college >> then I felt like I need to get into better company so at that point of time I went into sales in Lent and I was not liking the sales like what should I do?

[00:08:24]But I my one word I understood from my director is that every day is something you need to learn. Oh so that is the reason where I learned sales then went into consulting EI then product management in Microsoft then I felt right now I'm doing a course in IAT and in the postgraduation programming statement and that is also the mainly related to mainly on the generative agentic I want to get into the top 1 percentage. How are you managing all these things? Like one side you are working, one side you are studying. So I hope in this particular generation or in this era both working along with studying is a very big task because you have lot of things coming into picture like AI implementations or some other way everything is evolved right. So how are you basically managing these two things? So majorly when actually I was looking into this different aspects. I was just saying that I I said to this like many people my friends when I have conversation they say that what is my I want to early retire like more like uh what is early retirement plan like fire like fire process and they they have this acronyms like faster retirement process and they they try to say me actually I want to retire at this age 40 years or 60 years and I always have this discussion and at that point of time I was saying that I am at the place where I wanted to be. Got it. That is the reason where I always try to say that okay nothing came as easy to me and you want to stay in. So every day every day I want to keep 10% to my not getting any pleasure from family like you're not spending time with family parents.

[00:10:23]>> Yeah. Yeah. parents, everyone have to manage it. So I'm learning to manage better but you compromise sometimes at every stage of life.

[00:10:56]studies next maybe next two years.

[00:11:01]So you mean to say the pressure is inevitable but it acts according to the situation and how you will take it pressure is everyone mask like I was discussing I was seeing a few of the podcast from Alon Musk also with Kama.

[00:11:14]>> Yeah. So he's he has like if someone says he's busy you have to see life of Elon Musk managing three unicorn companies more than three companies and he's just trying to say he worked till 9:00 >> and 9 to 10:30 spend the time with his family he made sure his kid sleeps and at 10:15 he just after sleeping he took just a small face wash he came there after and he attended 3 hours of podcast with Nikl Kawa and at 1 2 he just sn with the same happiness and so do you feel that the busiest person in the world if he's able to manage the time with his family and also extraord things I feel we could do that it's more like how do you pursue like sometimes it's without anything also you get pressurized I think in this AI world I personally say that use AI to increase your productivity if you're doing some manual work like how I see this complete scenario is that when you are actually doing a manual work think in this way like if you are doing today's you are organizing emails got >> you can have a agent who or maybe anything that can summarize what happened yesterday and give you highlight what are the three important things that you need to do today. Now listen it might be Jira tickets. Oh it might be like it will able to scrap all the Jira tickets and tell you which one you need to concentrate because sometimes it's like 100 Jira ticket you'll get correct some information for your information few are there you require to act. Got it. Now you need not go into 100 and check the AI will summarize it. For example you can send your agent on behalf of funus zoom meetings. Oh, that also it will take summary and it will see if you have anyone talked about you for example a time has just told me it will let me know saying that this need to give to that so wherever possible like I personally want to like take one week >> and see which part of your life you could automate >> and with which tool and if you don't know the tool don't be don't be shy about it talk to any chatbot or you talk to your friend you try to understand that one hour how can I increase my productivity when you lose that like for example your 8 hours work if you're doing in 7 hours you have that 1 hour extra plan on that always try to see it's not about being hectic I personally say this in one thing when my manager also used to say to actually don't be busy hecticness busy will not give you any be productive like in that 8 hours what was actually useful attending a 1 hour long meeting and you have like 5 minutes take away is it worth it I'm not telling that all meetings you can skip but if you have parallel meetings you can just ask them for your information you just latest summary what happened you can get that summar so as you spoke about AI yes and Alon Musk now my biggest question is what is that changed in 20126 compared to 1985 and 2015 so let me keep it like three phases that way it will be so 1985 it's like very old time let's say 2000 below and we have around 21 approximately we have GPT coming in popular it started 19 2019 by 2122 we are able to see the actual thing M >> and currently if you see before 2000 the overall things changed in such a way that we are doing lot of changes in terms of how we work like it came recently people were moving to IT or from agriculture and this is a point where people are just looking into like this was something people were talk like I was going through few of the materials in 2019 85 your income is X >> okay >> your house is like like annual income is X your house is like 3x got it like twice the year in now The income has rised 10 times. Your income has rised 10x. But house purchase has became like 200 or 100x in many in Mumbai it is almost like 200x your income like for example if your if your income is like five lakhs >> houses became around two crit parity has increased and all. So the way the overall things have become because your house becomes 40%age of your complete uh amount. Yes. So you understand your take home is decreasing.

[00:15:10]So why I'm just saying that everything is same with AI coming into picture what happened now you have to work more.

[00:15:17]Exactly. And at the basic level jobs if you say like for example teleer let's say AI agent is able to talk maybe it's not doing 100% right but it's able to take your time and job and AI agents are replacing interns rather than managers right because what previously interns used to do or freshers immediately used to documentation.

[00:15:37]>> Got it. It's taking care. It's able to arrange meetings >> as a PA. You say personal assistance or maybe you want to talk it. So that also it was doing it. So you mean to say that AI is a new intern.

[00:15:48]>> That's what AI is a >> new intern >> new intern for you and personal assistant. I'd say that rather than intense like personal assistant coming.

[00:15:54]>> Got it. So my next question is uh do you think that our parents has actually given us a stability or incorporate incorporated the fear? Personally I feel that uh it depends on parenting and all I don't want to generalize things but many of times I've seen many people fear mongering is always there >> so in 10th if you don't get 90% yeah you're not for exactly >> and you can't do anything later >> and day in day out you get comparison even with me also I observe that everyone feels compared at every stage like inter if you don't get into IAT you are lo like you are not fit talented talented guy and if you're in IAT you are already being tanned as a talented guy but I've seen many people without doing miracles with funding and other thing you don't give that opportunity at all then IM comes into like doing IM had its own credibility saying okay they are like okay they are good for business or not they are able to do it now comes why at every stage you get compared I'll tell you inside and thing also if within your IM or IAT because I've done I'm doing the both on things. So as aluminina I can say that within IIT also you feel like you're incompetent because there are 100 equal people and you find to be best in that >> rat race going on at every stage. So until unless you don't take a piece that a >> this is me my own gen >> okay >> and you are able to take and I feel I am a perfect example that that like after I am I was in the bottom 20% of salis and because of any number of reason it can be like maybe the things that I worked on did not worked out and placement is a gamble >> yeah yeah >> we don't know anything and I had three layoffs in many companies I had to struggle with him But one thing kept me motivating is that there are people who are ready to help you. It's more like you have to reach out like if you don't tell talk >> to ask them help they'll not so I just wanted to tell you a story which my mentor said >> small one like he said that there is a prison.

[00:18:02]>> Okay.

[00:18:02]>> And in the prison on the first 15th floor there was a person who was locked and he does not know why he was locked.

[00:18:08]>> Oh for 15 years he stayed there. God and next day like 15th year after that aster also came inside and he's asking him why you are here he said no I don't know you did not ask anytime he said no I didn't ask like the log is closed and there was a person who was staying there I just asked him saying did you ask him any time like anything on the top he said no then why didn't you ask oh I said that should I ask you should be proactive on asking well he never ask then he asked >> okay >> he said I also don't uh then the inst can you remove the lock he said yeah I'm fine in that and he opened it >> oh >> the so it's always a just question so every viewer should keep in mind that ask you need to ask we need to ask right question >> right questions to the right people at the right time that's it so we always feel like mentors are there people are there you always telling who will help me.

[00:19:11]>> Correct.

[00:19:12]>> You did you ask anything?

[00:19:13]>> Okay, got it. And many maybe there are many people who are ready to help you.

[00:19:18]So you should go and ask and this helped me every place like I never I like for Microsoft referral >> I ping 10 one person referred it and I got through it.

[00:19:27]>> Understood.

[00:19:28]>> E I reached out to 105 aluminum and they gave me the reference and I never said I'm like saying that okay fine I'm exerson they said that I'll help >> understood >> and they end me interview. So it's asking a right question at right time.

[00:19:42]>> Don't shy away from asking question.

[00:19:44]Always don't think that who will be ready to help. Maybe you never know.

[00:19:47]Just raising a hand there might be 10 people who are interested to ask you saying that what what was the issue I can help you out.

[00:19:54]>> Got it. So there I saw a transition that is happening from our parent generation to our generation. Like previously it was like graduate from a good college, get a job, marry, then have a house.

[00:20:07]That was the old tradition according to you. What is the new tradition? I personally feel what is the current tradition also like if you talk to your parents on >> the new formula but things have changed like personally say that you talk about things >> like where you are double income no kids >> you have single income no kids are lot of other people also I was trying to talk and that is a place where you feel that >> like world is changing now I ask my like for example about kids and everything many people are saying why you're delaying the kids like they are like let me enjoy my life God was it's been a while that I have been struggling and everything. So everything every generation has a different way of approach. Right now I personally feel the structure where you have complete your like get a job marry then have a kids and then get back into like job and struggle it is still there >> but you have your own pace. So if someone tags you saying that 25 to 30 is the time to marry the norm subs now 30 to 35 you can take it you can take a decision if you want to purchase a house at 30 and stuck in with mine or you want to be in saying that I'll stay in 40 and purchase a house because you know where you stay >> correct >> that is a thing so I personally say the structure is need to be decided by one got it >> that is how you actually grow and personally my life also I've seen this one like if to chase about that timelines you're always already you miss the timel and you'll feel sad about it rather than that plan your times >> plan proper make sure that what you're doing if you're happy with it you can do it and I can say my example of my friends who are in the Europ >> like they're happy with what they like okay even they are aged and everything they're happy they think when things come to them for marriage they'll marry when they like why don't you purchase they're like I want to travel all over the having a house and restricting my savings there and everything at one place I wanted to do it in 50s so why will I do it in 30s >> previously the 30s is like to settle down correct correct >> now like they are like I want to roam around the world I don't want to spend that is a way of life someone is there they wanted to settle down at 30s and they want to roam the world in the 40s and 50s that is also fair so uh you have worked with top jeans right so why do you think that always the smartest people in in a room are taking time to do something or taking time to what I say like the smart always stuck somewhere.

[00:22:36]>> When you say stuck somewhere, do you feel like like chicken?

[00:22:40]>> Correct. Uh so what I mean is like so the smartest people stuck the longest.

[00:22:45]>> Okay.

[00:22:45]>> Why? I mean >> in a specific company.

[00:22:48]>> Yes. Yes. Yes. So I feel this is something has been in the past also.

[00:22:53]>> Twitter CEO one he's been out of the company maybe you can say kicked out of company. what he has done. Got it. Yes.

[00:23:02]Yes. And in a similar way I feel that there is lot of things where I've seen many people it's like struggle of having stability.

[00:23:10]>> Got it.

[00:23:10]>> They don't want to move from the stability. Let me take an example of myself >> working in Microsoft. Yeah. Stable income world renowned company and everyone saving lives and everything. So you never want to get into outcome like you never want to see opportunities outside.

[00:23:26]>> Understand? Because that way what will happen you know many times you'll be like okay why should I move the stability there might be something ups and downs but I personally feel as a life also like sinus it goes up it should come down the flip and if you try to maintain that statement at one point or time you see and this is a place where I feel that many spark people everyone want stability and they don't want to take any risk at the top position no and that is a place where many times we observe that they stuck for longer understood and let me give My example as I said like when I got like at a point where I thought like Microsoft is my dream come I I'm going to be at the top most place and suddenly a layoff happened >> and things were like okay I have to look outside Microsoft right now and then that is the best thing happened in my life why because I had to struggle I know there is a low that was happening but I'm pretty much sure that when I'm actually upskilling myself I'm in a better position in salaries >> better position in learning and that kept me more growing because at one point of time I kicked me out. Got it.

[00:24:30]Uh so now coming to the the actual part AI. So in this fascinating world AI is one of the fascinating thing that people from all ages from kids who are from 7 to 80 years to people around 70 80 years old. So they speaking about AI. So when they hear about AI they some feel it's a mix of curiosity and some feel overwhelmed but no one has a clarity what is AI. Yes.

[00:24:57]>> So from your understanding could you please elaborate what is AI in a functional manner not in a complete technical aspect. Okay. This is something like I would definitely say that right now and we are saying like I was remembering this one when digitalization start postcoid right everyone is doing UPA >> UPA became a craze of word you scan QR code you're able to send money correct and the adoption was very fast and things were going up now I'm telling you with meta launching in WhatsApp and there is a chat GP coming in pictures and people are already using it >> but for simple audience like artificial efficient intelligence like the AI research is more like something you can advocate work to like for example more than what you think you can do it like human intelligence is one thing artificial is like you on the top of human intelligence you are trying to keep extra layer on the top so for many people AI is like charged okay fine they are trying to say that okay AI chated started in charge but the major thing which I wanted to AI was there from 1960s itself.

[00:26:06]>> So things were there building on the top. It's like everyone were building at 1965. There was a particular phase that went artificial general intelligence or artificial neural intelligence. There are many things that were come. This is one part of it which came in which is like from using the LLM's training and predicting models which is more related to generative predictive model which are pre-trained transform GPT we call it.

[00:26:30]Chad GPT is like generative pre and transfer.

[00:26:33]>> Can you elaborate more on that?

[00:26:34]>> Yeah. So here mainly what I wanted to say I'll take you over artificial intelligence then we get a layer artificial artificial artificial general.

[00:27:07]What do we call them? So artificial three layers. Three layers.

[00:27:32]Got it.

[00:27:34]intelligence almost equivalent games.

[00:27:43]So first step it won the game. So that time it was a breakthrough moment for complete saying that humanelligence.

[00:28:18]So multiple things have speed.

[00:28:26]So our point of time transformer simple example Sunrises in space.

[00:29:03]Okay.

[00:29:05]Sunrises.

[00:29:10]Okay. Predicting. So it's not intelligent. It predicts better. Oh, understood. So changes.

[00:29:22]So the first stage of AI is predictive model where we predefine some inputs and when someone gives as a external person I give some input then it will search in that nearby thing and it will give you okay view. So mainly it's like large language models LLM models. Okay.

[00:29:39]>> So there are different types of LLM models. This large language models is like there are small language model where large amount of data if you give large model if you give small amount of data and you train it it's a small it's a small model. Okay. So you have to understand which will be helpful for you understand.

[00:30:08]Okay.

[00:30:13]So you have to understand where exactly that model is reacting because you don't give it proper okay what you are manifesting. So it's like simple a model that you want and what is actor things or else model act as what senior developer senior product or a person like it might be simple that I want to do my finan Okay.

[00:31:09]For example, it's like a kid.

[00:31:37]So you have to understand how you trade.

[00:31:39]So um can you just help me with what is I mean basically the AI and the phases of the AI where it all started and where we are reading.

[00:31:47]>> So very interesting question as I said that AI was around 1945 odd it started on 1968 came little bit onto the face model. Got it.

[00:31:57]>> And then we saw that IBM moment where everyone taken serious about a >> got it. But if I can say like for a normal person >> like how it is like there are three ways for a normal email like for example like us. What are the three waves that we have? First wave is like more like a chat GPT like model some Gemini or someone talking about cloud where you give something you answer respond like I said the prediction. So you ask something like for example it has like initially when it came >> it has a issue with data um having recency you can't say like for example yesterday there is a match between SRH and RCB you can't say that who won the match it does not know because it has a cut off that 2022 is a cutff date so later on it little understood and they got the live data using rag models I'm using from terminology but you'll understand that but mainly on a retrieval argument generation using some models finetuning it was responding better. That was phase one, first wave which started in 2022, >> 2023, 24.

[00:32:58]>> Now current phase wave two which is going on. It's more related to it is agentic AI. When I say agentic AI now it comes into the picture where we are trying to talk what you need like I want to book a ticket. Got it. It will go to the website. It will book it on behalf of you and it will ask for the confirmation of it. Or you can say that I want to book a ticket from India to Melbourne. when the rates are below 50,000 now it will trigger that I can say that schedule an interview with this five people checking their availability so it will check availability with you you tell the time it will do well so it is asking like a personal assistant agentic AI is like it is able to think understand now the wave three which is going to start from 2027 we are like in the beta phase sort of thing where it's like the reasoning models are this reasoning models have its own importance why now it and act also. Now it's in a beta state but some 2027 the things it's like it can act by thinking on itself like agentic sust like right now it's happening website design it's a beta phase but in 2027 you can see one agent talking to another agent employing it oh and making sure it work now for example how human beings hire in freelancing company we work another place and get the work done now I will like agents are hiring designing agent Oh, the agent will hire editing agent like for example let's say one agent is there it will hire multiple people and one will develop video one will develop editing one will deop music one will combine oh so it's getting that level so agentic session so it will ask on them and it's like a super intelligence so artificial super intelligence is the place where I say that the topmost where you are discussing about more than human oh like it can think better any taken missing.

[00:34:52]>> So you mean to say that in near future instead of a company means currently the company means CEO and you have certain hierarchy you have employees. Yeah. So you mean to say that in near future you have only one CEO at all agents?

[00:35:06]>> Let's see it's in a different way that will be very farfetched. Okay.

[00:35:10]>> But I say future like Mckenzie said that there are 11,000 like I'm not ex that number this X number of employees or Y number of agents working in them.

[00:35:19]>> Oh you see their reports coming up. So people will start talking in that. So they consider agent as employed. So now the work you might say like TC saying I have four lakh employees to they say that we have one lakh employees three lakhs each >> like for example let's say today you're talking about founding team like 11 people starting lovable like cursor light of company so lean companies are coming like previously unicorn when you say like for example in TCS in the forces we pro when we talking about unicorn thousands lacks of it and within other Microsoft let's say now what you're thinking like anthropic which you're talking like number of employers they have like it's becoming very so why it's happening because they hiring agents agents are doing some other work like testing from the so things are changing that's interesting place when I also say about like when we talking about go first So you have to invest your money to make sure you are up is performing better than you.

[00:36:46]they don't want to but think like four years then he is like keeping 18 20 hours of time 15 years you learned the previous models 10 years model now you have to perform better only experience will not fetch you manager is not just a manager now that he has to perform better you have to understand upscale on the future also they can kind of companies can predict what going to happen. So what do you do? What is the game plan if this goal push is moving here and there? So what is the game plan? How do what do you do? So as we were discussing it's emotional part also >> like settlement process what will happen that you have to understand that it's never going to you'll not be happy but if you invest time on your learn you know it and your settlement it's it's more like a psychological you can take like for example your financial loan build your emergency fund prepared for that you will not set Got it. Understand?

[00:38:01]>> Then work towards like first question you have to say that if I'm not settled what uh >> it might be spend time on learning or you'll spend such on like getting on advancement in job and today one job will be removed from it might happen such a way that a company removes job but that does not fine like after two layoffs what I understand is that if some companies reming I saw many people getting 200 PS and moving to another country. So you should be in that way like settlement in a company is not going to be permanent. So when you say settlement is not permanent right reason you see many top-notch companies they are laying off the employers right so from employee perspective I'll ask you a simple question so getting laid off is a bad decision or it's a bad preparation I I think it's a very interesting question like I would like to tell this example also for you like you said about bad decision and bad prep >> like there are two things first aspect is employer you and you and tell you. So this was from my experience.

[00:39:06]>> So during my Microsoft tenure and we went to school.

[00:39:09]>> So we were like everyone were working.

[00:39:12]>> Yes.

[00:39:12]>> We were like in a mind of settled and >> there was one person who was able to see that there is uncertaintity >> coming on because of few AI investment that were happening on >> and things were moving forward.

[00:39:26]>> He took his advancement.

[00:39:29]>> He started preparing for uh his growth.

[00:39:32]He made sure that he was prepared for uncertaintity and he asked a question and he a right decision of moving from our company and within one month the decision came on and it's not that we did not see and he was able to see the same >> so you observe things not just in within your team but overall dynamics things are moving uncertain it's like preparing for the worst so he when I I talked to him I asked him that didn't you predict it >> he said that I was preparing Right. So similarly from employee perspective see employees are you have to understand they run by stake like mainly from stakeholders.

[00:40:11]>> Correct.

[00:40:12]>> And they are investors at the back end.

[00:40:13]>> Exactly.

[00:40:14]>> And they have a huge pressure to perform. M so that is a place where for them it's layoff like I've been other side of it also working with EAC and where I've consulted many companies and when we want to increase their profits and beta and everything we suggest that because that is a easier bet if you are able to get improve your financials and it's not because of anything else but at the end of the day your cost is higher in the resources now you have to see like movie is the best thing you have to improve so that is like not emotionally from them. It's more like you are like they say that you are a number in their financial balance sheet.

[00:40:52]>> So you are treated in that you are not ashi you're not you they or you are not some other person it's like one employee number so they they don't want that employee number that's something they decide. So for them it's about the larger company benefit not individual.

[00:41:08]So that is a place where I feel that as you said laying off has become a norm because people are expecting that investor are expecting that coming with AI you have to have productivity rise and the best way to show productivity rise is removing people. Okay, AI. Do you think AI is a bubble? It can bash any time because still somewhere I see people need human touch that is needed now and then but is AI overhyped?

[00:41:39]I think you are rightly pointed AI is overhyped with but okay you have to understand Zimbabwe like 2000 but there will be some companies which will survive. they will be next big companies like anthropic >> open when people are might over thinking like okay within two three years there will be completely agents working the question you ask there will be one CEO and agents working below correct >> that is a bubble thing that is capital but there is a mix and match there will be equal agents equal employees tomorrow it might be like you is having a agent or personal assistant >> working with so you have two people things because you'll be having one agent with you, you'll perform better.

[00:42:23]>> Understood? So agentic part AI and tomorrow it might be tomorrow anything comes up. It's a bubble as such but you have to understand in this bubble few companies survive you they're going to do it for simple example to add it they are investing too much into it correct you are not able to get that much profit because the expectations from investors or maybe expectations on company is too early like for example if I give you a charge subscription or openex subscription and expect your productivity to increase 50% so >> like you have to be reasonable right you have to give a learning curve for a developer also if you ask that okay tomorrow is he's working 18 hours a day and I give him a charge license or a clone license and asking to automate we'll say that see I can use at one extent >> and everything I can't ask like claw then I will not be there right >> yes >> so coming to the uh in question recently I saw metag or any big gains they are laying of extreme number of people like thousands like so as you said a is a bubble again it may bl right your point of view cost is it just they're saying because of a they're laying off people or something they are working in a back end for cost reduction to implement this in a >> very right question that you are saying like so you have to understand this is a cycle it's not a decision like in so when Microsoft of started it when meta started it I mean Google and Amazon started it it's a sinosur >> Oracle laid off crazy yeah recently and reason why they were saying that they have a huge demand of building their infrastructure to and they are cost cutting according >> the similar way meta is also planning to do because you have to understand like when you're saying you are agentic a company how do you do effective now the expectation is that you have to put like single digit companies like nine members, 10 members are becoming unicorn and you are supposed to get part to them your architecture processes need to be automated to and your question of laying of people is one thing it does not then they're not hiring or is laying off >> you see accenture laid off >> and still you see there are hiring post there are two types of layouts one is like they are laying off people wherever the AI processes was not there.

[00:45:00]Traditional processes are not there and they're automating the process and they rehiring people with AI skills. Oh, you are hired with that. So last time I had a discussion like this management and see your point of view expectation.

[00:45:42]to change work with them talk to them and with age of experience it's very difficult adapt they are saying that thank you but why can't them that is the reason they don't even want to So and Tsis is doing it. Infosys is doing training is there and this is a place where we are going to have two lakhs people train training in TCS. Oh on like AI capabilities infos they are training. So you never know this people tomorrow will be like having a lot of capabilities. So layoffs that is that still satisfying satisfying Oh, no. No. Correct.

[00:47:24]better way of no one is saying they're not performing better. They are like who said that you might be top perform and it was like said at that time said Microsoft director that is so we have that much who decides whom to remove to the layoffs are done with us.

[00:48:37]So developers are observations like when I just type in in GPD like based on my past inputs can you suggest me what type of a person I am it is able to give me the answer >> I it will evaluate like you know like more like a balancer you try to keep it like be a harsh like you need like charge is always more like it will be like well to ask it like be unbiased Then it will come out to you like simply saying that how like you can sign your manager point of view like saying that what you need to improve what you like it will point out simple simple things where you are saying that you're just not commenting on what I've done instead that you did doing same thing you copied and pasted you're asking simple questions many of the people right who are working right experience senior senior I don't think senior managers or senior positions but freshers and mid-level people they are confused whether I need to use automation or AI so how do I decide whether I need to use automation or AI in the way of product I am building so right now the difference has not been much automations and AI using AI automations are done >> okay >> so it's like simple one but you have to understand this one aspect that right now the complete process is getting automated using different agents and different place and this has been long time I'm not telling this is a something different it is there but you have to understand that for example moves.

[00:50:38]means process means security for example.

[00:51:19]Credit card details.

[00:51:35]For exampidate credit of China.

[00:52:07]Oh.

[00:52:12]So, so basically we are building agents or we are training agents to remove our own jobs.

[00:52:27]Question A will replace you and A will sorry A will not replace your job but people who learn who are learning A will replace you under is it a sthing statement that companies give uh to their employees or really something I personally C A is not replacing you. A is replacing your excuses while learning before you previously in same job you're in 5 years without learning anything doing anything upskilling and app now the complete progress changed your excuse now you have to take time for learn understood no tools skill set got it talk to people implement build >> so it becomes mandatory for >> understood so that is a place where a >> oh it's removing your excuse to go what and now you are food like I'm telling that person with 50 years 50 years age or 40 years age need to spend 2 hours a day or one day one day learning or maybe spend some time in the week you don't have excuse saying that okay when I have family time I'll not spend this removing your excuses and that comes with its own pros and cons there are two types of job okay which is one is like once you learn you can implement it is like for guiding Let's say >> once you learn you can drive car understood go forward understood >> and there are few jobs which is like for example you you need not do any automation like you learn the vehicle and do that but now with this come skill set you have to invest time on learning new tools new skills improve yourself this becomes your daily job okay of learning so that is the reason I say that a is not replacing you art is like if you're rigid to change and if you have an excuse >> then we are repressed >> we are repressed that's what it's happening from management if you say because I had this conversation with my team >> where we were discussing with them and there were like out of eight we gave them channel access we gave them cloud access you asked them to use out of 8 to 10 people only four people utilized remaining four to six people are remote so here use when you We have given them access or top-notch access whatever it is and people are not using it. So using it means it triggers me a question come to prompting. So basically is prompting a new discipline or is it just resume it will hallucinate. So here comes the word hallucinate. Thanks for point. So when hallucination comes it will like are as good as critics and so sometimes it want to please you. So I'll create something and like it it can be as simple as that you ask me saying that actually what is 10 into 10 I don't know.

[00:55:34]>> So I'll tell 200 because I I don't want to say no to you.

[00:55:38]>> Oh yes.

[00:55:39]>> So that is the reason in the task also you say that if you don't know tell me no.

[00:55:43]>> I understand.

[00:55:43]>> So give it a guard saying that if you know give me the resources if not in this you have singleshot ping zero shot.

[00:55:52]>> Okay multipoting. What is that? Zero shot prompting is that in one prompt only you want answer. Okay. Like you directly ask like sun rises in question mark. It's like >> uh and strawberries how many hours like multi like two short prompting is there like two prompts you'll ask first question like you may say that for example how many centuries that India cricketer uh maybe for example in it got answer I'll ask second question within India how many okay it is commented multi-shot is like again I'll go into like with exact opponent of maybe Pakistan how many uh so now you're training the data reasoning. So this is the so multi-shot counting is good for reason like you are training it like you say 5 6 7 8 what is odd number you trained it. So there is one more method >> like many people like you me >> does not know like like how many questions to ask to make sure the model perform so for which there is one like super magic sentence that you can use and the chart GP will perform better for you.

[00:57:00]>> Okay. It's like in zero short prompting in the first question itself >> if I say any mathematical question >> which I give and I ask it to do step by step think and think step by step do the process now it will only ask the reasoning questions okay so step by step by Okay.

[00:57:47]So I need to give in the same problem same do it step by step. So reasoning by mathematical equations by research. So and the prompt know >> so this is one of one of the trick every viewer should know >> every should know and that is very important thing because many times you see it's not performing like few factual questions and on that question it's not performing wellce step by step reasoning question.

[00:59:04]So prompting question then I say that okay structure rights police officer So every time I use GPD cloud or first like for example as a product manager I want to understand how my competitors are working on so and so space list on the features and your business it will give me an output. Now the same prompt I edit. I just added pull stop or comma in the last and I just submit it. It gives me different reply. Why?

[01:00:15]and bless.

[01:00:31]It's not mean to give you right answer, wrong answers. It's it is predicting next right uh predicting the nearest possible answer. Oh, so experiences. So same question.

[01:01:21]So it's not like Sunrises who like for example national actual answer logical creative answers.

[01:01:56]Okay. It will think on the ground. So nothing is right.

[01:02:05]>> Soil restrictions.

[01:02:13]Okay. So coming to the point having shorter prompt in effective way is better than having a long detailed prompt. So which one is better? So rightep process.

[01:02:52]So tryences.

[01:03:14]Okay.

[01:03:18]correct.

[01:03:22]Got it. And apart from active framework, do we have any other framework that gives you better?

[01:03:43]Okay.

[01:03:45]cont.

[01:03:47]Okay.

[01:04:00]PDF.

[01:04:28]Okay. So, so 5.5 open you observe that prompting is not about how big you write, how appropriately you write with complete Got it.

[01:05:10]Like many people I hear here and there on social media and all most of them confuse with AI tools versus a agents. So if you can explain in a brief What is AI tools and AI agents?

[01:05:29]Right question, do think act concept. Same thing for example let's say Gemini for example right it will not take a decision on it saying that.

[01:06:29]Okay. So what you're saying you're saying specifically like this is my parameter that whenever there is >> discount of 5,000 rupees on my current flight book the ticket it add to act it will not because acting is not it can't take a decision. So agents are such a way that they take a decision and act.

[01:06:50]So act is like purchase. What else? Call to action. What is CTA? Call to action.

[01:06:56]What is the uh what we are saying that whenever the rate becomes X you're giving that task to so it has to decide like every time it will go like maybe every 1 hour it will go to the website it will check whenever the flight rates are getting down it will purchase. So decision Video generation Zolan Image 2.0 related information Google search. Next one.

[01:08:12]So enterprise agents create for example agents.

[01:08:36]So for example, transform Got it.

[01:09:19]For example, 10 minutes.

[01:09:40]Okay.

[01:10:00]vlogs.

[01:10:02]So basically uh viewers when it comes to Gemini uh Claude everything has their own uh credibilities everything can perform similar kind of action accuracy but they are more confused about what to use when to use. So But where to use what? Like for example, what is the best st for the designers?

[01:10:43]So mainly design there are different stacks going.

[01:10:50]So first okay example.

[01:11:06]Why? Because for developing images is doing great.

[01:11:34]image model.

[01:11:36]Open discussion but you have to evaluate reason why this model will have an advance.

[01:11:59]Correct. So every Monday morning we are finding a new model with new advancements Kim. So question is there price sensitive and also it's actually doing very good with the overall process.

[01:12:17]So 11 so we are trying to do image 2.0 overall performance like all cl so basic next what a stack is better for product managers. So as a AI product manager myself >> I personally see that sticking to one stack is somewhere you leaving out something out out >> so you have to try and understand which stack will be helpful for you. So as a product manager we were having this discussion >> as like should we have a codebased access also as a black should not be a black box in the as a product manager and a product manager previously we have a daughter as we were discussing previously also like getting into it. You know don't understand as a product manager borderline >> because as a product manager AI product manager product manager what is expected for you in that how quickly you can build a prototype got it yes ready to get into production production that you have to have a developer for security reason everything but you are getting into MVP minimum viable but one question is uh when you say we need a media prototype that can perform at least for MVP one over don't you think that we are missing the perfection >> so here the game is not going to be perfection or faster to the market but then again this is a contradictory check one side suppose if I'm building a product simple take an example of ticket booking so as you mentioned I want to book a ticket from say Kali Oh uh Barara Hills example I'm giving or from Hyderabad to Thupati take I I given uh I get a detail saying that okay based on your inputs okay these are the five trains that are available on this particular day and the cheapest one is this. Now you have given me two options. I can select from five trains or I can select the the cheapest one that you are suggesting. Now perfection in the sense what I'm saying is when I click on something it should get booked. I should get an email or message saying that ticket with so and so seat number are booked or in kind of like I expect my turnar around time to be very very less. Latency time latency time should be very less somewhere we are missing the perception in conjunction with the faster space of output Indian right question I I can completely understand the way you're saying and somewhere it's very important question to answer >> here we are not delegating an imperfect product out >> okay >> you have to understand but here what we are trying to say is that out of 10 ideas there might be one idea that might be working >> okay you spend on what you want. So designer previously >> used to spend that much time on 10 products. Now you're saying that spend on that one product >> get perfection. Oh, so uh coming to this right I was reading to certain um just wanted to continue on that one please please go where you were saying about I wanted to highlight here it's like we are not crossing anyone's in like as a product manager also we are not going to do a developer work you're not going to a designer work kind of thing here your understanding is about making sure that when you want like as a develop as a product manager or anything the overall understanding coming about a proto prototyping it close to production over it so that the designers will able to think better and their time is not wasted in just telling them which color it is. Here the cidility like many times I feel that like design thinking is one aspect I said it's more like I is ideate D is defy or then you have normally once you get get this one you before going to testing and prototype once you ideate right you have to actually build it develop it so ID like develop the one define after you'll have to develop then after you'll have to go into testing and prototype then you have to go there so you're following this design principle here.

[01:16:54]>> So what we are decreasing here is that we are not using this low fidelity to high fid going in. So the other person will able to act fast when this time is not wasted. But even we have some tools like Figma right where I can give a prompt and I can get the uh my designs built >> and even I can I can deploy then don't you think that designers jobs are at stake? See you might have to understand a couple of thing. And it's like it's not just about deployment.

[01:17:25]>> It's always about how you are doing like here the perception comes in picture. So if someone is like 5 years old designer he knows that what customer is needed and how to keep that up. So >> so you as a technical guy you have a technical expertise.

[01:17:40]>> Got it.

[01:17:41]>> For simple projects. Yes. Like for example, what as a product manager you're for example if you're thinking of future product manager role I would say that sit down on one weekend build the product >> by next weekend just make sure you build a prototype and everything in v 0ero.dev dev or you have lovable you have tools to do it and deploy it in Google play or app store or maybe in website and within the month and you have a user research coming up >> and you'll able to find it what's going well what's not going well another so this is one aspect now when you're saying where the designer aspect it's like single person is handling it >> but if a designer is there if you're building an enterprise level product >> then every small things comes into picture >> got So I'll tell you one example also in this one. So there was in Amazon there was a point where they were planning to get this one click button for purchasing. So if you click that button it will be purchased like one one stop like they wanted to say that one click to purchase and because of this the revenue increase >> but you have to understand that before that you have to onboard like saying that you have to keep addressing this is a default this is a de payment so that you will be doing it process that is a redo but once you done that one it's one thing but you have to understand where should where should I keep listen >> which color it should be there oh so that the larger That is a place where the designers come into lecture understand. So uh a generic question as a layman I just give a prompt between saying that I want to understand what's going on in product what are the new enhancements and I click on enter. No, it will give me some actually what is happening where can you just tell me in nontechnical way in this step process like just to understand and to make viewers understand actually what happens behind the pump when you so majorly um I think you best you asked me to dissect it and then that's very good for the US let me do that one >> so it's more like a large language models working with neural network like it's like how you have neurons in your brain it is so it's simple simple question how your brain thinks.

[01:20:03]Okay.

[01:20:16]the closest probability first of all.

[01:20:44]So a minute like for example okay related information predictable next word.

[01:20:58]So whatever the sentences you get it it's like what is the best possible predicted word next possible word that >> it will match to that and it will see what are the all the probability then it will give me >> give you tix so it's more like a predicting next so it's as good as the data only seconds.

[01:21:32]So as a product it is good.

[01:21:52]It does not want to say Who is the current?

[01:22:08]He does not want to say no.

[01:22:15]Don't listen.

[01:22:17]Don't give me sources.

[01:22:24]last two years.

[01:22:28]Got it. So that is where as a product manager right you have to understand few core skill set your product thinking.

[01:22:35]>> So even though it gives something it does not blindly take it into your understand >> you think on that you give customer. So we were discussing with my mentor. So we were discussing on this product requirement document.

[01:22:47]>> So your product requirement document is not just a starting docker.

[01:22:52]>> It is a dynamic knowledge base. Whatever knowledge base you have previous from the customer, you have information related to all related data. Everything like for example knowledge base that is the so you keep on updating that and this update will be going to GP.

[01:23:22]Got it. So coming to the point of uh AI in the product space. So most of the product managers right with whom I am dealing or I'm working on the products that I'm working on majorly people don't know really my problem needs AI or not. So how do I define whether really my problem needs an A? That is a reason and it's a valid question that you mention.

[01:23:53]So always for example So 200 $200 $200 I can't say that $200 I started thinking maxion.

[01:25:00]So that what is realizing people so it's like don't force AI if there is not requirement and you have to see cost value for money. So for example, understand for example.

[01:26:08]Someone saying that MC buzzword MCP is dead skills are you have to understand that it's not like you can't talking of software as a service as a service process what I want to achieve are changing Hyderabad So process change destination.

[01:26:48]So as a enterprise products and being handling enterprise clients I understand they're rigid to change.

[01:27:00]So you have to understand that why you if you're implementing AI is there any data vulnerability AI responsibility like which is responsible AI is becoming very key because you never know your data where it is going on because it's a new world if you are building AI on the top and you gave the data to train and it's the data is out and it happened to many companies >> so you mean to say that how uh there is no reliable >> it's like not like reliable AI is available >> but you can't trust the complete process for example let's say >> there was like ex employee from specific company >> he uploaded company's document into a >> got it uh is trained in that day and his competitor is able to search and get that content now it's is it something AI >> so because the man made main failures >> so it's like organizational failures and you have to understand that you can't keep every loophole now it's a big world and once it is entered now some that company can't call and say remove that date now it's out into it that is the reason where it is happening and many like this comes into lot of pictures like I say defakes are coming in how do you use the data like image generation you're talking about what about deep fix how the things are impact so you have to know guard dials proper before implementing it so as a product so I got some few of questions from the product managers and the designers. So on they have sent me saying that like can you ask these questions to the guests? So I'll read one by one.

[01:28:42]>> Yeah. Yeah. So coming to the first question right uh as a product manager how do I separate real use case real AI use case from something I I just look cool.

[01:28:54]So is there any mechanism or the process that I can simply say that okay uh differentiate okay this is a AI use case very rightly pointed out are they so it's simple way I can say is that without AI if you're able to >> got it and with AI if you're doing >> can it still be same or maybe it's like is it a incremental growth or a exponential growth I think these are the two questions that you can ask is it like adding AI is a incremental growth >> okay or adding saying AI is exponential growth. If it exponential growth, go for it. Find out >> incremental growth and exponential growth. I need to understand >> so when you're saying incremental growth, it's more related to it's just a additional version.

[01:29:36]>> Got it.

[01:29:37]>> Like for example, let's say simple like we were like we were also looking into building voice agents.

[01:29:43]>> So we are building when I was learning building voice a now for a real estate customer or insurance customer.

[01:29:49]>> Now we have like sales CRM built on.

[01:29:52]>> Got it. Now the increment like adding a AI on the sales CRM what we thought we have a tiny AI which is like a agentic flow which will like the the admin will come and he'll talk saying that how many calls we have got >> and it will report it will say who has done proper >> which calls I should call back >> if I want to remove one person who will be that person >> and it is able to understand the calls and says that out of this conversion rate of one person is almost zero 5% rest of people are 5%. I'm not looking into reports, dashboards. I am talking this is like exponential okay incremental growth is like like for example let's say now the employees there during the sales sales CRM that sales that is coming on if you're trying to say that I'll add AI to just to transcribe the call. So what you are saying is I need to first identify whether implementing this particular AI or any use case is exponentially improving something exponentially improving the overall feature >> overall feature in the problem statement basically >> for example in like for example let's say sales CRM sales OS I'm talking on >> here the incremental is like transcript >> transcripting is like you rather than you typing it you after the call it auto transcripts and gives the summon Got it.

[01:31:17]Yes.

[01:31:17]>> Which is incrementing which you could also say this is hot lead or of lead or warm lead.

[01:31:22]>> Yes. Yes. Yes.

[01:31:23]>> But the exponential growth is like having this agentic channel >> which you can talk and it will act on wheels. It has control of everything like it can remove a person who add a person it need not look into dashboards.

[01:31:38]that tomorrow's world the dashboards become obsolete because the CEO goes there or leadership goes there and ask the question because rather than you sitting in the and looking into dashboard you'll just ask what is the things that I need to work today >> and it will let that top three two three rings work on so that is exponential >> so basically like like the management says can you identify the AI use cases where we can keep in our journey product journey Right. So as a product manager where should I start first?

[01:32:13]>> So you have to start with as I say always start with empathizing with the customers.

[01:32:19]>> So first you empathize with customers understand if this is actually a praying point for them. Got it. Focus group discussion. Here comes the design thinking part where you ideate and also define the problem clearly. Understood and then you develop. So it's like more like ideate what is the problem talk to the customer is it a really and like having a stakeholder interviews yes talking to them understand that actually they problem like I have a many cases >> where actually the problem that you're solving is not real problem >> like I was trying to talk to my one of my >> and um in my past companies like we were just trying to talk to him and he was having a critical problem saying that he's not he's saying that I'm confused with this application got it okay Now he did not say why he was cor now our team was like let's revamp everything and I'm like no let's go to let's ask where is the confusion is there okay >> and we were looking into his journey of the race and then we understood because for uh like it's a regional issue >> like we think from like left to right we read in few regions like mainly on the northern side like other geographical we read from right to left.

[01:33:34]>> Correct. So you have to understand it's confusing because we our complete architecture is designed in that way like one after so left to right their ar right pillar right we have to understand then where exactly is that and build on the top so for your question it's more like as a product manager you have to talk to the client customers don't directly jump into the solutions go to product problem then build on it so here is another question like uh I got it as a PM if AI gets the answer right 90% of the time but confidently wrong 10% of the time what is a usable so how do I define how do I decide see it depends on like how much cost that you're going to lose by that okay so if it is a patient let's say it is for heart disease and if it is a medical side if that 10% wrong is a very big for that but if you're asking a general problem like for example learning and the 10% will not cost if that 10% will cost you 90% less go for it. So okay that is majorly the cost and what is the impact the cost of that that 10%age for like for example I always say if a pilot pilot miss.1% there is high chances of accident >> and that is the same thing if you're in a smaller thing and you are going with 90% precision also so it depends on case to case what is what is actual impact on it and how much you can actually have this ratio like for example I am fine with 90% percentage because with the hallucination like 10% 90% continents because I feel there is no major loss with 10%age of answers.

[01:35:21]>> So according to that my case that is the >> I'll give an example. So someone is building a product uh it's simple uh like there is a users there is a request then it will go to admin and uh admin will approve or reject. Now there are cases where sometimes people don't even turn up to the websites and they don't approve and most of the case are getting expired. Now my question is as you said now right my intention is like using AI I want to build a use case where it sends some email recurrence email or anything and later part it will auto approve on now in this case the 90% versus 10%. How do I define that?

[01:36:10]>> Very interesting question. So just think in this way like when when you're looking into this auto approvals and uh taking a little bit later on like reminders it's a generally a problem for everyone >> but you have to understand that what type of approvals you want to go for auto what type of things are not shown on >> oh >> and which are the generic one which has less security vulnerabilities got it like for example let's say you are taking a access like let's in in my company If I say there are auto approvals and there are some approvals that could be required does not require much like for example charge access correct >> or something that accesses that does not have larger impact which could be like done in terms of audit and reminders are something which could require like 12 hours 24 hours required.

[01:37:00]So you have to see that if if someone is delaying your approval for 48 hour >> still you're fine with it because that is not a critical one then it's it's a good one >> but problem with another aspects you have to understand is that if it is a security vulnerable correct >> and you are giving admin access to someone and it auto AI auto approach that 90%age and 10%age auto approved someone else got and you are at a very big stake of loss so understand that >> if AI is like accuracy level is something you have to decide 99% accuracy is something you require at that point I can't compromise on that is the reason auto approves are not prominent in bigger organization where you have it but you have to define which tickets are actually security vulnerable which are internal got it >> internals are like simple access like for example if you have a access request environment you can go for after and then you can actually look into it backward and revenge back later on if you feel that it's not right. So as a product space right uh if I could ask you like what are the three big tools that any product manager need to equip to sustain for next at least 10 months >> very interesting interesting I say there are no three tools that you but at this point I can because see tools are evolving nowadays as I said you're right settle is not the same thing the way I'm saying like there is no way like you learn three and you settle down the fifth >> or else the the things that makes them survive Yeah, make themselves. So one one thing which I personally say claude is some some useful tool that is there.

[01:38:42]>> I personally say cowwork and cloud code you try to have this ecosystem of claw design cowwork and cloud code and we were talking about workflow automation.

[01:38:51]So the learn there is a skill jaranic and anthropic.killar skill chart skill jar >> skill jar dotanropic.

[01:39:00]>> Yeah. Yeah. So they have this skill job mainly where they have um in their website they have many learnings of clot go for it like it's just like 2 three hours of courses you complete it you can do that and second thing why you required this one it's like for research purpose >> okay for even for interacting better way and it also interact with the cowork so cowork is a completely it's like it works with you coding is also there so one aspect >> and second thing cursor people might find it like actually curs how can a product manager learn curs and sort of it I want like now the lines are blurry >> so you will get access for code you should know at least open the code and see so now the expectation from the people is like you open the code at least look into you might scan using AI >> unless vulnerable to anything right now it's not a black box and the third one which I can say is more related now it's going out is like any designer like it could be like sigma I Then sigma is one stage but I personally say building prototypes like using v 0.dev is one thing v v v v v v v v v v v v v v v v v v v v v 0.dev is a tool sub to like for you have it like v 0.dev which is more like in create designing aspects working prototypes like as I said like three aspects one for coding which is cursel one for designing prototypes and different aspects which will be there got it and the third one would be flaw right claw is like it's a generic like you have research and everything you can use for data and you can build some some agents on the top of it for your automations and all that so these three are very important tomorrow like we we discuss after 6 months I'll come back with new things but As a product manager you have to understand few things is like identifying problem building research you have to KPIs stakeholder management >> yes >> so that is the reason this like building prototype understanding the code so in this one top three comes into this picture >> so can I say that if you learn these things or if you equip with these things you can become a mini CEO of your own see as I said this one aspect of um said uh we are also in beta settle like there is a critical thing like then like studies then job then purchasing a house >> and once you purchased house you get married and then you have kids >> a traditional traditional way of set thing so what I wasn't wanted to tell that this is not going to be the same like the same thing got break now you have to think like today >> if I don't settle what I need to do you think on that ground and you do according as I was mentioning also right now the overall structure is chang and as also we were discussing more time it's not that you worked for 18 hours and 20 hours neglecting your health maintaining that balance keeping some time for your work workout work out sometime for your family sometime to yourself and balancing out in a such a way that you survey for longer so this is a marathon correct >> not a sprint So previously people used to think like set I'll run for 18 hours 16 hours and I'll settle down then I enjoy less. So keeping current view like with environment health and other issues you have to plan your enjoyment also.

[01:42:24]>> So one comes I mean uh before we end this end the podcast right so I have two more questions. So do you think that already there are certain white collar jobs that are dead but yet people don't write agreed with it and many a times I personally say that >> it's not about like they are re rebranding if you don't upgrade you are actually already taking on that because I'll tell you one like very interesting thing so using plot if you integrate you can actually edit a video oh proper like good level and you can actually generate videos and the one job of editor editing and designing has been taken by and it is doing it in the pro level at a less cost like within 2,000 rupees I can get one hour of video of complete edited one and you can give it there are multiple agents working on it now you think about the editors who are working on it time they're spending on it >> done >> but you have to understand that they have to rebrand themsel using that tools they have to perform better. But why do people so I know it that using plot I can do the entry even the editor is getting upgraded why do I use because why do I utilize him basically because I know that AI is doing yes so this is at like there are few things that perfection comes into picture or maybe you need personalization customization because every videos look same you want to have a little bit more understand so that is the reason it's like how do you use it to make it more professional and personalized to you.

[01:44:10]>> So for example, you are doing a video over here. I am doing a video. How do you differentiate? Got it. Now you come into the picture like you have to come a little bit not just a designer as a product manager someone if he does not evolve she is in the same way. So why do you require a manager if everything is managed in such a silos and you're getting updates? So so one of my friend has actually developed Dan BT update tool. So you're getting updates directly to the CEO updates. So for taking the inputs and sharing you don't don't require you have to evolve the things.

[01:44:42]Similarly there are huge jobs right now know like voice call >> there are also been some jobs where agent is talking properly and you don't like I'm telling you if you're paying 12 to 13 rupees you don't even identify agent or a >> yeah that is AI talking. So one question is um where do you see AI in next 10 years and at the same time is there any genuinely shocking things that you can tell that AI AI capable shocking AI capabilities that currently AI has for the viewers keeping in the view right like next 10 years is very like I can say like US states like 1 to two years is something we can look into as I said 2027 is something I'm looking into things getting changed because things are drastically changing a lot. So few AI capabilities which I was thinking uh which is happening right in couple of years it's like we talking AI doing work next we are training with AI robots. Oh, so now China in China there are things that we are hiring robots to help to fold your um uh bed sheets do as a maid work do your own work and the AI is helping them out to do all the things.

[01:46:01]So within two to three two to three years you find robots going like doing your stuff when you're old age people buying like for example how do you find iPhone and within that the robotic AI side is developing a line so you'll find that common like in US in China there are people where uh it gets some groceries from your store and you're going to find that common going forward in two to three years I don't say common like they are coming up but you'll seeing the products coming in prototype in India market or maybe US market and other places within 10 years down the man I feel this will be common like after 10 years you find like you're talking to a robot and robot is with equipment and there are like people calling a robot as their girlfriend >> oh nicknames not nicknames it's like they dream like okay this is my girlfriend and this has its person having relationship with a chatbot like AI bot is going to be the in product side >> on side of it I say that things will be as voice covers like using your phone you can actually develop the products or you need a laptop now also you have in co-work and other places Gemini you have every control is given this dispatching everything so within your mobile you can actually learn it because it's talking it's doing right or wrong you understand you'll just look into mute suggesting everything will be in the microess so that is a place by AI is moving All and lot of advancements which you don't know which I wanted to say is happening healthare healthare majorly what's changes are happening healthare previously for any vaccinations or anything they used to do thousands of trial 2,000 like lacks of trials with AI they are able to redo it in the better way and they able to do it in short like for a vacant might take 10 years previously taking one two years and there was a case where a person in Australia identified a cure for a cancer of a D is dawn using AI. So he just kept a DNA sample of this used AI and he called multiple people worked on it and identified and done cured the cancer of the >> and he was able to cure because there was no cure for the dog cancer. So he needed so like without any medical biometric medical skills he was able to so tomorrow that is be next. So tomorrow if you are feeling lonely don't be shocked because in near future you get your own girl customized girlfriends customized girlfriends or boyfriends who can actually make you sure that you're not alone anymore and this is a scary part also right like where I see in near future we don't have the term called relations either it can be with your wife family mother and all because you're already equipped with lot of day which actually hallucinates as your mother father and all hope that day will should not come where people lose actual relationships definitely and this is happening nowadays I'm feel like like many instas I try to say that the AI is responding better than my friends or my girlfriend sort of post because it's not judging you >> correct correct the judgment factor doesn't come >> now you think about having a physical image like a robotic path >> gone >> and you have that talking and this this is not something I'm talking I'm a very science secret movie this happening and many people are talking to Sanjip in their personal thing which they don't even share to your mother and father >> understand so with this like we will wrap our part one the second part will come up with more interesting topics where we also discuss about building a live agent actually what happens behind that and how you can even sell it. We'll speak up to some part of how do we monetize it?

[01:49:59]>> Definitely. So majorly we can concentrate a little bit on a technical aspect like as a layman beginner how do you actually look into building an agent and then going back into building like how do you manize it?

[01:50:11]>> It's it's like it's a running prototype is something we can look into it because definitely >> it's something like you can do yourselves.

[01:50:20]>> Exactly. I'm looking forward for the duction.

[01:50:22]>> Yeah.

[01:50:23]>> Thank you. Hi everyone uh welcome to classic connect with Udiran. Before we start the podcast I want everyone to please grab a pen and paper and this is not a normal podcast where you can keep it in a background and listen during your scroll but this podcast actually gives you value addition to the work that you are doing. So you may have to sometime stop scroll and listen back again to understand how we are using the AI and how can we use the AI in future.

[01:50:53]So with this I'll bring up the guest.

[01:50:56]His name is Akshai Sharma who has wider experience in building AI products and has wider experience in working with bluechip companies like EVA and Microsoft. So let's welcome Aki to the podcast. Thank you everyone.