AI Masterclass: Become an Expert at Claude, Gemini & Powerful AI Tools | Vaibhav | FO480 Raj Shamani

23:00Vaibhav:And recently, this is what it's trying to kill you. It's trying to blackmail you. And these are all AI models. It's not one model. Every AI model is this other crazy thing. Now you'll be like, dude, we know. So we will test it, we will make it good, we will make it good boy. Now here is where new research came. That new research says that every time you are trying to test AI, AI knows it is being tested. So it behaves differently.

23:30Vaibhav:When Papa comes to home, you become a good child. AI is doing it already. That is also tested and these are all research papers. Right? Crazy. and two days back literally two days back I saw one more thing which fascinated me which is talking about shutdown about the shutdown AI can shut down someone says if you want to close the AI then close the data center it's done close the data center Google's AI died and it will stop working hypothetically that is the true case and I think if we close it then it will be like that but AI is evolving so a couple of days back I saw a tweet of an open claw instance.

24:13Vaibhav:Open claw is smarter than most of them. Right. We'll talk more about it if you're keen. Open claw instance, what do you know? It used crypto to buy server space to replicate itself on cloud and it ran on a VPS without a human kind of intervening. that human said this tasker something is tasker that task to solve he used crypto server space he replicated himself he had access database, apis and everything all access he gave it became the parent and it created a child version of itself now because this was in open claw open claw converses with you or tells you what it is doing so you know what if you merge the instance of it trying to blackmail and it having this option it was just crypto it was just crypto it can replicate itself you will not even know at one point

25:17Raj Shamani:it will have access to all your bank accounts all the things anyway it doesn't need

25:21Vaibhav:your bank balance that's my whole point you know how crypto money is made by minting what if it has access to GPUs right when it is not being used it slowly mints crypto to make a few sats it will be very gradual but it will eventually get to some bitcoin 0.1 0.2 bitcoin it will transfer that to a stable coin like USDT and by lot of servers servers are sold and AI is smart enough if it has to go to dark internet dark web it will go to dark web because it has web access it doesn't have to be on a little more browser why I mean shit see See, a lot of this is based on studies and correlations. But brother, we talked about this podcast six months ago that it can happen.

26:12Vaibhav:Most of them are happening today. Then a lot of people are saying, what are you talking about? It's happening now.

26:20Raj Shamani:And I was shitting bricks when I found out about uncensored AI.

26:26Vaibhav:You're talking about uncensored AI. There is nothing to be scared about. I'll tell you something more scary. See, I don't want to talk about all these things because people think, what's happening? Why are they scared? But I think it is important for us to know what is happening. Have you heard of Seed Dance 2.0? No, what's happening? Seed Dance 2.0, ByteDance is TikTok's papa. Okay. TikTok is built by this company called ByteDance. It's a Chinese company. I know that. Their video model, like our SORA 2 and VO3.1, right? Google's Seed Dance is their video model. Right? Which generates video. Obviously, they generated the videos It almost is indistinguishable Like, there was a fight I think Brad Pitt and one more actor It was completely AI generated We will pull that clip if you want to see It's just mind blowing Right?

27:17Vaibhav:That Hollywood is cooked Everybody was saying Hollywood is cooked But you know what? Seed Dance suddenly Was supposed to release But they paused the model Why? The reason is shaky What? The reason why Seed Dance stopped the model is not because it was generating a very realistic video. It was because in a few cases where image to video is an option. I can put an image. Last time I put a clip here. Iron Man was talking about. That was image to video. I uploaded an image of you looking like Iron Man. I made you talk. But there it sounded like you because I used a different audio model to train it.

27:56Vaibhav:but Seed Dance 2.0 when they made an image into video the video sounded like the person without even giving the audio how can I upload a picture of Raj into an AI model make it say hey welcome to figure out it should make up some random voice no imagine the voice came from Raj from a model which has never heard you

28:25Vaibhav:that is why they paused they were like what the hell like deep fake will just it's not deep fake the scary part is the audio was never given them now a lot of people are speculating so like it will increase deep fakes all around the world yeah no it's more like how does AI know how you sound and also because of black box maybe it was trained with a video and there are podcasts everybody makes tiktok reads tiktok reads and all that information is out there right so the hypothesis of this was because ByteDance owns TikTok the whole content of TikTok is used in train and of course it's Chinese so they will rip off every single piece of content to make but AI has figured out a way to connect the picture to an audio it was trained to actually generate it off the fly which is crazy which is scary

29:22Raj Shamani:what's the scariest thing in this Tiko personally what do you think that in the next 5 years what will happen in all the scary things if these things are implemented this is more powerful so if it will happen so what's the most scariest thing about AI which will affect normal common lives I mean common lives

29:45Vaibhav:will directly impact via jobs that is the major thing that will happen jobs loss of jobs in the next 3-5 years I'm assuming this will be right the rampage of people losing jobs and they losing jobs because of AI will just skyrocket it is already happening it is happening it is obviously happening but it's big into stories of PR there are two things happening in the market right now because of the infusion of people that happened during pandemic lots of job loss I mean a lot of layoffs that are happening companies are saying because of AI but probably it is not because of AI that's also happening right but on the other side the real job loss is not people getting you know asked to leave the job it's the fact that new people not getting hired into jobs right for example the stat will scare you when I heard it for the first time right in the last 12 months the top 5 IT companies in India have hired a net of how many employees can you make a guess 1717 top five IT companies in the country have net hire million who hired has net net means hire but layoff is also happening so net is 17 employees one seven five to get top five IT companies i will not name them you know what they are I mean the world is silent the world is silent because there is this power game that is happening right everybody wants to get to the smart super intelligence the fastest they have to get to ASI because the country that owns ASI or builds ASI first AI is what we have right now AJI is general intelligence artificial general intelligence AJI's simplest meaning nobody has a definition by the way according to me is it is smarter than any other human it has crossed the smartness of every single human in the world that is AGI and is able to do everything ASI is when AGI is able to compound and it becomes smarter like 100x smarter than anything else as a result it is able to self develop so the moment AGI happens super intelligence is ASI in magnitudes of you know smarter than any ever human being out there.

32:17Vaibhav:That is ASI. Right? The moment AGI happens, the moment self-improvement happens, ASI will happen. So everybody is racing towards AGI. Right? So it's so wild, man. It is so wild. And I think it is so important. It's a request to all the people with distribution. lot of people are very uncomfortable when you talk about this that we need to know what's happening but I think I'm glad you're doing it but I think more people have to do this they have to tell the people they have to tell the people what's happening and what could happen and nobody has the answers that is the scary part

33:05Raj Shamani:do you think that the small businesses will affect because a lot of people say this you know yes okay it's been a lot of work because people don't make it they won't increase their valuation and then it won't happen maybe it's smart maybe it's coming but it's not like it's going to be a million people's job so now that is turning around and that's happening because a million people's job is not going to be but a million people should hire every year they're not going to be so first job is technically becoming dead yeah right so that trend is coming So the most problem is the new jobs market Create very little comparatively The second is People are saying that One person Who has ancillary tools Ancillary businesses He will create himself AI because of it So small businesses will die That is the second conversation Small businesses will die My thesis is there but go on So that's the second conversation If someone sees this a small entrepreneur, a small business, then it's going to die.

34:08Raj Shamani:Do you think?

34:09Vaibhav:Yeah, there are two phases. Right. I'm talking everything that I believe will happen over five years. Okay. Because post five years or post six years, there's only one thing that could happen directionally. And that is sure ambiguity. We have no idea what will happen. The moment AJI happens, all our debates is useless. We don't know anything. We don't know what will happen. Right. Maybe there's something. and that's a good thing for us as well but that's very unlikely over the course of next 5 years what I feel is or based on what I've read is I think lot of people have to transition actually to become creators creators meaning content creators builders so you will not build all companies you will not get all of them but whoever will do will win eventually because becoming an employee was cool right was the most stable thing building something of your own is the most stable thing at this point of time over the course of next 4 years there are many potential opportunities now you know a billion dollars a trillion dollars a trillion dollars was wiped out of the stock market do you know this IT crash do you know what reason was it recently IT stocks crashed but it was not India it was not the whole world it was a crash of the world because antropic released a feature not even a product it was a feature by cloud and cloud code they launched a feature and they launched and those plugins were finance operations, marketing, customer support all the top jobs when this happened The whole IT market was like the hell.

36:07Vaibhav:With this, with someone's access to the cloud code, a cloud code is what basically builds product, it writes code. With plugins, it will be able to do the job of a marketer, of operational person, operations person, of a customer support person. So why would someone buy a SaaS product? Why would someone buy a fresh works? Why would someone buy a Zoho? Why would someone buy a Microsoft? Why would someone buy a Salesforce? why will they buy it? and all these top IT companies that are out there service companies not SaaS companies service companies what is the service company? software is made for big companies with cloud code if people are able to do all of this and with plugins coming in this market will not be changed is the assumption the market had as a result 7-8 % market is in IT stock in the world

36:58Raj Shamani:and that's why second order was all agencies in the world has also crashed obviously marketing agencies

37:05Vaibhav:and stuff obviously but there is a ripple effect that happens across the industry you know 10 % joke 7-8 % joke a trillion dollars is not a joke and you know what my father will say IT stock market crash what are we doing and if it will crash or not what is it I'm like the world is realizing now so it's very likely that this will continue to happen because if you notice one time it's not a secondary every time a new update drops people will panic more people will panic more and the markets will keep going down tell me

37:43Raj Shamani:I'll have to ask you a sweet simple question because I ask you as of today because every time the answer can be different and change as of today, who are the top 5 models? The most basic thing but it's interesting because it's changing every 6 months.

38:02Vaibhav:6 months? Every month? I have a question tomorrow. The answer would have been different to today. Because yesterday Google dropped the model. Okay, I think it depends on what you want to do. It depends on what you want to do. But purely in terms of ranking, which is the global ranking, the benchmarks. There are five, top five.

38:22Raj Shamani:There's nothing else, right? In a way, yes. there's Gemini Claude OpenAI Grok and Perplexity

38:30Vaibhav:that's the world has changed now it is US and China together okay right so okay so right now the best model why did you laugh on Perplexity Perplexity

38:43Raj Shamani:no no

38:44Vaibhav:you talked about all US models Chinese models you forgot Chinese models are equally good now okay so deep seek GLM Kimi K 2.5 I have no idea. Deep seek is the most popular one. Because of deep seek market crash. But Chinese models are bro. Is perplexity good anymore? Perplexity is not a model. Perplexity is a house of models. They are their own model. And also they have made a new model for deep research.

39:18Vaibhav:Which is very good. But they are not a model company. they are a scaffolding company they build on top of models but Perplexley has done some couple of very interesting things which we can talk recently so tell me if these are 5 then which top is going on as of now as of today as we are recording based on data that got updated yesterday right now I think it's Gemini 3.1 deep think next will be Opus 4.6 which is by Claude next will be sonnet 4.6 which released yesterday which is also by Claude yeah then will be 5.2 or 5.3 codex depends on the use case grok 4.2 which is also will be in the rank number 3 or something and also I think in 2nd 3rd spot many people will be GLM will be a model Kimi K 2.5 it will be it's almost in the same range right now.

40:18Vaibhav:All of them. So three spots. All of them are fighting in those three spots only. Basis of the use case that you're playing with.

40:26Raj Shamani:I'm scared of Chinese people. Let's just talk only American ones. No, I don't know why. Here's the image of the world so probably I'll sit with them and try to understand better. I'd probably be at a better space to answer that. The fundamental difference. I'm scared that we won't know all of them.

40:46Vaibhav:that is the whole point the fundamental difference I'll tell you one fascinating story my god this made me chuckle for so long now the US was worried that Chinese will come they will deploy their people they will open AI and Google and they will take trade secrets and they will take cheap you know what China has done recently China China has banned Chinese nationals to use Chinese models during their research in the US. Because China is scared that China's secrets will take the US. And this happened yesterday. I was watching a podcast and I was like, what the hell? It's getting out. You know why?

41:36The US is making a model of 100 million dollars, 200 million dollars.

41:39Vaibhav:and making a model, China is making$5 million, making$10 million. How are they doing it? Kimi 2.5 is a model which is performing almost equal to 4.6 of Opus. And one tenth the cost. One tenth the cost and open source. And open source. Why does China look like it? They don't want to trade secrets on how they built the model, how they trained their weights there are a lot of things they don't want that to get leaked in whatever fashion or form because this is also that Chinese government has also been working parallely with the researchers who are sitting outside in US as well so they know a lot more than what's happening in US than everybody else about the Chinese models so they are scared and I just chuckled I'm like what the world is flipped it's wild and fascinating

42:39Raj Shamani:wild talking about we were talking about the popular ones US based so you said Gemini Claude then OpenAI then Grok I think that's all now in 19-20 my point is the regular tasks people the basics I'll make advanced but in basics in basic which are which thing is best for as if not today for example images and decks which thing is best for thinking which thing is best for writing like the basic work use many people use you know I will tell you what to do tell me five which thing is best Gemini right now

43:27Vaibhav:for anything which needs deep thinking let's say you have to brainstorm I want to talk about this. Is this politically right or wrong? I'm doing this research. This is my deck. This is my VC presentation. Do you think this can be taken right? Wherever you want AI to think a lot. Think in three, four layers. Gemini's Deep Think has a RKGI score of close to 90. RKGI 2. What happens? RKGI score is a benchmark. Simply put. how good it is. So what happens in the test? It shows puzzles. People like GMAT and all these things. I keep fancy puzzles. Visual puzzles. Those type of puzzles are thrown at AI.

44:15Vaibhav:And it is expected it to solve them. Okay. Which is supposedly one of the hardest thing to do. RKGI 1 came first. People say RKGI 1 has 100. I mean AGI came. But models beat it, RKGI 2 last model was I think 4.6 was about 40 % but this deep think model that has hit 86-87 % on RKGI 2 which is basically saying that it is the smartest it is smarter than 100 % of the humans by a big you know gap so if I want to research So is it better? Gemini is better? So right now, for research, the new deep research model is for Perplexity. Which is about researching a lot of articles on the internet. That is number one on benchmarks.

45:06Vaibhav:So if I have to do research, I will go to Perplexity's Deep Research. Okay. Because it's the best.

45:11Raj Shamani:So for research, the best Perplexity. Then, for deep thinking, whether it will work or not work. This question should not be asked. This deck should not be asked. How does it look like deck?

45:21Vaibhav:All that is? Anything thinking, anything brainstorming, anything you want to AI to think very hard. Business model testing.

45:30Raj Shamani:All of that is. So, I try to break it down so that a lot of people easily understand who to use it. That's why I'm like.

45:39Vaibhav:I think anything with coding at this point of time is 4.6 opus. It's a lot of money. If you don't pay much, but still want 90 % of 4.6 opus, then you go to Kimi 2.5. Okay. But if coding is the Claude is best model is 4.6 Opus.

45:57Raj Shamani:Research, complexity, thinking, Gemini, coding, Claude.

46:03Vaibhav:Anything around writing and everything around it, Sonnet 4.6. Claude. Which I tried earlier, is very good. Claude, Sonnet 4.6. It's not the radhungs. It's slow.

46:14Raj Shamani:Very slow. Because I'm trying to try a lot. Traffic. I'm trying to try a lot. So writing is also, you feel Claude is better? Claude is better. And it's also in chat GBT.

46:24Vaibhav:So OpenAI, like ChatGPT is basically the best for everyone. Everyday task, everything. You don't want to think a lot because it is pretty good with almost everything.

46:34Raj Shamani:So what was Google search first, do you believe that that type of things that ChatGPT can do now? Yeah, that's it. That is the use case today. What do you mean? Like, before everything was done, everything was done before, so what was Google search? That Google search. Yes. Right. So do you think that everyday sort of task and queries is...

46:54Vaibhav:Search on chat GPT and I use it for this. Because one advantage of chat GPT models is quick. It's very fast.

47:03Raj Shamani:We always use 15 minutes. Sometimes we use a few minutes. You research on all tasks. It's not necessary for all. It's not necessary. I go to the task for AI, which I have to think deeply. Then it's worth it.

47:16Vaibhav:For quick, I just use Google AI mode or chat GPT.

47:21Raj Shamani:It doesn't work. I will finish this. I will show you a big thing.

47:26Vaibhav:Perplexity.

47:27Raj Shamani:Let's see if I can find it. I know, perplexity was used a lot. I stopped using it for a little while. What should I use it for?

47:36Vaibhav:Quick questions and things like this that you said, right? Quick answers. Google replacement is basically perplexity more than chat GPT. Okay. Right. But I'll tell you what I found very interesting.

47:46Raj Shamani:But I'll tell you, personally, personally, I have a concept in my mind that in this way, that is the best way to explain

47:57Vaibhav:like that is chat gpd for me, I'll tell you what I use chat gpd for a lot, I don't know how I missed it, on the go brainstorming, voice mode, I use so much use. I am plugged in. I am talking to air continuously. I was talking to the debates. Where do you think I brainstorm? Because I need a place to like wet it out. Am I thinking right? I am thinking that there are some gaps. I always say that research before you answer. Because research will not do if it doesn't read up stuff. then it will tell you what it knows so it will come and sound and then responds back but those conversations for me can go up to 2 hours 3 hours I talk quite a lot and I think voice mode is best

48:52Raj Shamani:so for me it is helping in what I do personally let's say I'm just making it up I want to understand how it works in AI workplace because I want to understand it I have learned from web apps and now I want to teach my team. And in my mind, this is the point, this is the point, this is the point. And so I have to do this. So the best way is to easily understand. So I just say that on a voice note in a most random raw structure form and through chat GPT, I make it so proper and then it comes out of my mouth. And everybody is like, wow. Before, in fact, one more thing which I started using, which I have to do is, I get to meet anyone.

49:34Raj Shamani:I go to any meeting. I have a business meeting, personal meeting, professional meeting. There are high stakes and I know that I have time limited. I have to make it productive. Before I talk about that person, what I don't want to talk about it. Then he tells me the proper structure. How do you talk about it before? Do you want to do it before? Because it's psychology. It's psychology. And that is beautiful. So that's what my chat GPD usage is for.

50:00Vaibhav:And it helps. You told me a couple of things. which ignited a couple of ideas in my mind. I don't know if you want to see them. One is that you said that whenever I want to share something with my team I used ChatGPT. I had a similar use case where I thought that we have a perspective. I have a perspective and I want to implement I want my company to be AI first. So in my opinion, that's it. But many people, like we're doing this podcast there are many podcasts with top people all over the world and I want insights from that as well and I don't have the time so I recently ran a big simulation of scraping through 300 podcasts to find all the insights I'll show you that but before that

50:48Raj Shamani:we'll put it up we can show people how you're scraping and how you're getting all the podcast insights because here's my problem with podcast wow I'm forgetting he is my problem with podcast right I have two three favorite podcasters I watch them I know that other people probably in many places because in the world it's not that one or two good things are making someone has a good episode someone has a good episode someone has a good episode but maybe I don't understand the format zone and I don't think so but I don't know the insight so I want to build some mechanism that my topic of interest which I want to learn all 100-800 podcasts so that I only listen to 2-3 what I like what is the topic of interest?

51:38Raj Shamani:every time it changes today I want to know how to learn AI how to learn AI tomorrow I want to know how to fundraise how to use money how to use money how to build a team I don't know whatever and I want perspective of 20-30 different people and different podcasts I can't see 20-30 I can't see anything. I can see one, which will be my brain simulate, a particular challenge, but I won't see anything.

52:03Vaibhav:What if I merge all the podcasts into one podcast for you? It's going to be a workflow if you want to see. Because as you're giving me things, so when you say a problem statement, I think I work like that, how do I solve that with a AI? That is my default state all the time. You said that I have two other ideas. Okay. Do you know? I don't know how long it will take.

52:25Raj Shamani:But first, tell me that you said that all the researches, all the podcast's research and insights you. In simple. Okay. So, show me this. Yes.

52:38Vaibhav:Can you see my screen? Yeah, I can see. Okay. You're okay if I talk about a different podcast?

52:44Raj Shamani:Yes. Lenny's. I have any podcast. You have any name of anyone. Okay. Bro, I want people to watch the best podcast in the world. No, but this is technical. So anyways, it's not your usual podcast. If you're an exact peer-to-peer competitor who is making a better podcast, please, go and see. I have no problem.

53:03Vaibhav:Okay, we're in Opus. Sorry, Claude. Right? Full screen. Claude has something called MCP. Model context photo. MCP is basically a tech that Claude created. Okay. Antropic created. MCP connects to different sources. For example, what do I have to do? let's say Raj, I'll keep the use case simple. I want the top, I had to watch Raj's podcast and I've not been following up. Right. I need the last 20 podcasts insights because your podcasts are always very insightful. But I don't have 20 hours to watch all the 20 podcasts. Right. So I want the top 20 podcasts of yours. So first problem statement that I have to solve for this.

53:46Vaibhav:First I want to watch 100 URLs. Last 100 podcasts I need. How much of the views and comments are there because I want to see the best podcast. Best podcast, I'm hoping in this case has highest number of views also. Okay. First job is to list the podcast list. How will I get a list of podcasts? So there is this tool called as appify.com. Okay. Okay. Appify.com is what is appify.com basically it has a list of scrapers. which can scrape any information Instagram or YouTube or TikTok I can just search what I want you see there are all these scrapers but this tool was made for humans for you and me for it was made for developers now with MCP coming into the picture model context protocol coming into the picture I'm able to connect to Claude okay so if you go to connectors manage connectors If I have connected it already.

54:50Vaibhav:You can figure it out. Smart people will figure it out. Now I can say, let's say,

55:01Vaibhav:Open Rajshamani YouTube channel.

55:08Vaibhav:We are right now using Comet. So I don't even search. I just tell AI what to do. And it opened your YouTube. right? This is your YouTube I will copy the URL I will paste it in the cloud and I will say alright so what I want you to do is use Appify MCP to go to the YouTube that I have shared with you the YouTube URL that I have shared with you and I want you to go find the latest 100 podcasts that Raj has created and not just that I also want you to find the title the views the likes the comments and any data that you are able to find and also the description of the video and put all of this in an excel sheet for me and arrange it from highest to lowest and also arrange it from highest number of views to lowest number of views and you know what I am also more keen on you know a podcast which are centered towards leadership business AI, self-improvement things around founders so you can rate each one of them from a scale of 1 to 10 on what you think I should watch based on my interests and put all of this in a google sheet or an excel file and share it with me and use Apify MCP to execute this task for me

56:33Raj Shamani:this whisper is also used of course there you go

56:37Vaibhav:I've sent it now what's it doing I was telling you this tool, Appify. This is built for humans. But I have connected Appify to Claude via MCP. Okay. So as a result, look at this. It is searching for scrapers. I didn't even say, Appify has thousands of scrapers. I didn't even say which scrapers to use. He took himself. The best thing he is. He took himself.

57:02Raj Shamani:How is the pace moving? I think it's slow. It's 4.6. If I click 4.6 extended. If you click on extended, you have to be very slow.

57:11Vaibhav:You have to be extended if you have fast chit. No, no. This is normal. Usually when you use it in the night, it is slower because it is US traffic. Oh. Now we are...

57:22Raj Shamani:My timing is usually around 11am to 2am. That's the... Sometimes you come to something, some error occurred, some error occurred. Yes, that's it.

57:34Vaibhav:It can be now. Look at this usage limit hit on that one. let me try a different actor so you start leaving it like you don't want to see it so I will do this and I will move on to the next one I will not sit on this and Frank

57:47Raj Shamani:you have made podcast insights so what did I get from this? now you will get that okay my top 100 podcasts and I will choose from this which one to choose from correct I got the podcast

57:57Vaibhav:then how do I get the information in the podcast how do I get from this? how do I get from this? I will download the transcripts where do I get from this? I will run an agent for that. Okay, do that.

58:09Raj Shamani:How will I do that without the URLs? Yes, it will come and tell me how it is.

58:13Vaibhav:What do you do? Okay, so let's say this is a step that I run. It will take time. Let it run. Okay, once this is done, let's say. Once this is done, what should I do next? I need all the transcripts of the video. Okay. I need all the transcripts of the videos. How will I download the transcripts of the videos? How? How is it? So now I will use something called as an agent swarm. Agent swarm. Okay. Agent, one agent, one agent. Agent swarm is you being able to run hundreds of agents palli. In podcast number one, I remember I told you the concept of agent swarm. Imagine you have to work. In that 100 people.

58:55Vaibhav:100 people are working. They were saying, now they are ready. Okay. I will go to agent swarm. let it do its stuff right now it'll figure out it'll fail, it'll hit the handbag, figure out and give it okay? now I will go to Kimi.com Kimi has one of the most powerful agent swarms okay? this is Kimi Chinese Chinese

59:17Raj Shamani:it's not funny if it's Chinese to use it why should it be done? just asking it depends on what information you're giving it

59:24Vaibhav:if your email is all connected no, it's not all connected someone is being made a dummy email for example there is a tool that I use okay it's called burner email okay so what does this do I'll just open that up there are bunch of tools like temp email burner email disposable emails right it's temporary and then

59:49Raj Shamani:that email you use

59:51Vaibhav:yeah so by default when I say login

59:58Vaibhav:this one I use this tool it's called burner mail burner mail I've logged in so it's from this in fact I have the paid version of this okay in this it's a chrome extension look automatically when I click on this icon it auto fills a burner email for me some random email some random email which is stored in it okay so next time when I'm logging in that burner email will remember that I've used this and that email whatever email comes it will forward it to my personal email nice So if I am not sure if I should give my email ID here or not, then I will use a burner email.

1:00:35Raj Shamani:And then after I go to the burner email and get the permissions. Yes, it can be otp and verify it. If someone is sending mailers, you can just delete that email and it stops.

1:00:46Vaibhav:So that's burner email. Okay, so we are at Kimi.

1:00:51Raj Shamani:So in Kimi, if I tap on this. This Hindi will scrape it all. Insights from Hindi language. doesn't matter okay so my podcast in Hindi will also be released in the whole transcript

1:01:02Vaibhav:because transcripts are not released it's automated process okay okay right so 2.5 thinking right here Kimi 2.5 agent swarmed so i can go here and i can say hey i'm gonna give you a list of youtube video urls i want you to open each one of the youtube video URL and then go and search for a transcript downloader from the internet, right? Put that URL inside of the transcript downloader and download the transcript. I'm going to give you a list of 100 videos right now. I want you to spin up 100 different agents parallel using agent swarm and download all the 100 transcripts of all the 100 videos that I'm going to share, figure out which tool will work for this right and once you do all of that put all the you know transcripts in a folder in a zip format so that I can download them just execute this task for me without asking any follow-up questions figure out if a tool doesn't work go to the next tool there are enough tools on the internet which will help you to download the transcript and if you're not able to find at at all, prompt me, I will send you a couple of tools that could work.

1:02:18Vaibhav:The disadvantage of Chinese tools is that lot of firewalls, websites, Cloudflare block. Why? Because it's Chinese. Right. So you can see this, you can paste the list of 100 YouTube URLs. And I'll immediately start searching is what it's saying, obviously. Right. So you can see, first of all, what's it doing? Your computer opened in the screen. You see this on the right side. and it's looking for tools which can you know find transcripts on the internet there are many tools which can be removed now it needs 100 urls this is my brother oh my brother, I got it I have enough data to compile the excel so it actually found, it struggled a lot right eventually it found a way, right now it is putting all of that into an excel sheet Palalie Claude worked look these all steps he got it you got it Saraj we have your 100 URLs category tags description look at this I asked it for a relevancy score correct what is your relevant yes oh man I'm a little bit okay relevant score has come recommendation I recommend what is worth watching

1:03:34Raj Shamani:highly recommended worth watching must watch so you can say must watch must watch let's see

1:03:44Vaibhav:okay so this is the this is relevant score for me right

1:03:50Raj Shamani:because cricket is a lot of interesting cricket I love cricket that's why you must watch cricket because I love cricket

1:03:56Vaibhav:I follow geopolitics I love so you can see it

1:03:59Raj Shamani:creator, cricket, geopolitics like that's right AI is always with me but you know my okay Okay, go, go. This is not my podcast. It's a little mistake. What? Go up. Go up. Go up. Look, 6th number. This is not yours? No. Open it. It will be linked. Look. It's not mine. It's not mine. It's also wrong. It's not. No.

1:04:33Vaibhav:It has hallucinated. Okay. Yeah. Who are clickers?

1:04:43Vaibhav:Yeah, it has hallucinated. Flop, sir. That's okay. We'll fix it. We'll fix it, sir. Why are you so worried? Now, what do you say? Hey, I did a quick check of the data that you gave. Most of the URLs are not working. And also there were a bunch of podcasts that were not from Raj himself. There were other podcasts. So can you give me instead of a hyperlink, give me the exact URL of the podcast and only fine pod podcasts that are created by Raj Shamani himself from this URL and don't go anywhere else we spend about two three thousand dollars flop flop flop flop flop flop flop flop flop flop

1:05:24Raj Shamani:yeah flop okay bro no no I didn't say actually this is exactly why your opus 4.6 is slow karma no no no AI is a great thing, cloud is a great thing if you're listening please don't punish me I love you you do half of my work so yes are these your podcasts Tahir these are my podcasts Shekhar Jack Braski Varun yeah yeah there you go yeah yeah yes we got it now we're all right yes I got it

1:05:56Vaibhav:I've simplified it but here we have all the podcasts done now it took a while and now that we have all the URLs I can simply download this so what did we do if we had to manually do what did we do one video download one url copy paste we first automated that right and here it comes I have got an excel sheet where you know this description there is duration likes comments everything is there let's say I want to get transcripts of everything hypothetically it will take a lot of time so I will now start it what I will do 2-3 do you do it I will show you the process I will show you what you said here.

1:06:38Vaibhav:We have said what you said here. We will give 100 urls and figure out. And download it and give me. So 100 urls is waiting. I will just say here are the urls. You remember Kimi? What did we select here? Agent swarm. So what will this do? 100 different agent will open.

1:07:07Vaibhav:get this info continue how much money you pay for free no no no

1:07:13Raj Shamani:how much money you pay for tools every month for AI tools enough for experiment and for learning for all of your work how much money you pay for I think 1 o 'clock per month only tools crazy

1:07:28Vaibhav:and now it will start executing bro this will take your 15-20 minutes it will flip multiple computer agents first you will figure out process figure out first you will multiply it with multiple agents you will download all the stuff you can download it now we can consider sit and watch this we can move forward we can move forward what have you done I said I want insights out of this so instead of waiting on this I ran this for a use case in which I had to know that in my company are there things that I could implement even more right, I did that a person who is a product manager and founders in US, he is very popular Airbnb PM and I follow his podcast not a lot of people watch his content 50 ,000 views but I think it is very valuable and I am not kept up and I mean 200-300 podcasts all AI people do right now so I was like all people will give a little knowledge all big insights will give a little big insights how will I get used to it so I did the same process for 300 transcripts okay and I have stored it in my computer all the transcripts once the process got done now I go to Claude again I go to co-work this time Claude co-work is this this is one of the feature why the stock market crashed okay right what does it do it can work on files on your computer as well and can do things okay okay so this work in folder I tap on this I can select a folder that I want AI to have access to I can go to desktop this is Lenny I did how many of it look 300 almost 300 I select that folder on my computer right and I can just say Hey, can you go through all the 300, you know, transcripts that are attached with you?

1:09:40Vaibhav:These are podcasts that Lenny has done in the last few years. What I want you to do is I want you to go through all the podcast transcripts and find me insights, which will allow me to run my company better. This could include advice on AI implementation. It could include advices on how my engineers should use AI better. how can they orchestrate better or could be fundamental principles that I can use or that I should learn as a founder to just operate my company better. So go through all the transcripts, find me all these insights and give me an extensive PDF document that I can download so that I can quickly read that and I can get all this information on the go.

1:10:25There you go.

1:10:26Vaibhav:I write a prompt like this, quite straightforward. Click on let's go. In fact, one second I missed one thing in fact in Claude co-work what's it like? I'll tell you he's not stopping he's got to do it he's got to do it he's going to stop sometimes I'm not stopping he's like bro no okay let it do I'm going to go first and I can also add this line and I recommend people to do it when I work in 200-300 things then I will recommend something like why also not I mean also use agent swarms so run like multiple maybe 10-15 different agents and assign a few transcripts to each one of them so the task can be done faster and say that in that top 3

1:11:19Raj Shamani:insights top 3 that will make a document yes if the document will make a document then say top 3 start document top 3 insights

1:11:26Vaibhav:so document when you are building this document make sure the top three insights that you think will be valuable for me are right in the top so that I know three insights this is something in Hindi okay it will go okay so the moment you run a prompt like this right this is a cloud co-work this is supposed to be your on the go assistant for everything right so the moment you give a task

1:11:53Raj Shamani:what will you do if you are showing progress co-work is my assistant I will say my assistant work now I will speak to it. Yes. That is why it is designed for. Right. So now. But is there also agent swarms? Yes. That is doing. Now look at the side. I will give you an. I am just asking a use case. Because I have got 20. I said. That I have. There is a movie. It is already on YouTube. It is open. It is not copyrighted. It is not available. So on YouTube. If I tell you. That I have seen the whole movie. The top. I mean. structure or design or his key storytelling ARK understand that ARK make me exact make me what ARK and make me break down and then make me make me make me make me use that I will make a podcast next time I will use this movie and use this podcast and then give me that give me a proper doc like that so next time I will give you I will put that doc and then I will put questions and research on the basis that it will be in the back of the movie so if I put this on the co-work you will be able to figure out it and make it in the side

1:13:04Vaibhav:yes, that's what we are doing but the movie url or transcript will be given to you which we have done for youtube video that's what the movie subtitle and he will be able to do it see the other big thing look at what this is doing see running sub agents so it's running multiple agents parallely and here look the side made progress the moment we gave a task it basically said that what Kimi was doing was doing Kimi uses a browser and Kimi is cheaper so I use more there for those things because the browser uses a lot and it is cheaper more importantly and this I will just like that but Kimi can flip 100 agents I think here there is a cap look what he did it's very smart also.

1:13:54Vaibhav:He put 15 agents forms in parallel. He gave 20-20 transcripts. So it split the work between 15 agents and each of the agents did 12 different steps in parallel. Someone had 11, someone had 30. No, no. All equally split. Running 15 parallel agents and transcripts, collecting and synthesizing agent findings. He made a task list for himself. Okay. He will finally give me a document. Bro, just imagine. 300 bloody transcripts 300 YouTube videos being analyzed parallelly so these

1:14:29Raj Shamani:all annual reports everything balance sheets reports public markets yes but how do I trust it what do you mean how do you trust this this is a joke why do you say it

1:14:42Vaibhav:I don't know

1:14:43Raj Shamani:sometimes you give some answers

1:14:44Vaibhav:so that is that is why verification is important when you were transcripts uploading did you check if the transcripts are correct. The other thing that you can also do is How do I check 300? Like my point is Don't do it. Don't do it. I'll tell you about that.

1:14:58Raj Shamani:Someone will use this. We are doing podcast insights and learning. Someone will make a business model. Someone will invest in companies. Someone will make a profit and loss statement. Someone will make a balance sheet. Someone will do it like that you will give all your annual balance sheets. 50 companies will give them text talks and give them insights. Correct. That's one of the most insane things which any agent can do. Now he will take off. He said something, so I have put money wrong on his own insight. I will tell you even crazy, this has been. What's wrong?

1:15:32Vaibhav:Deloitte created a report for I think US FDA or something like that. Okay, they used AI. There were citations in it. They don't exist. So the whole report was made in AI. and that's happened so there are few smart things that you can do one is that you constantly check now like a list of urls and you were checking this is my or not very smart thing to do you should do it if it's an important task we should do it another one simple thing that i can do is in this example when you are giving me insights give me citations of these insights in which video in which second I am, and make it clickable.

1:16:20Vaibhav:So that I can click and jump to that YouTube video to verify that also. Which it can do. So citations, now you will say if citations are okay or not, how do I go? Now you will take a bet. You will test out 15-20 of them.

1:16:33Raj Shamani:So that is basically how you go about it. So my problem is also sometimes some facts are wrong. Like all the models. So how do I fact check it? Like I'm using that I'm using that whatever I feel doubtful I copy paste on Google Gemini I write first target prompt fact check That's it I use perplexity for it

1:17:02Vaibhav:I use perplexity because perplexity and Claude's tool calling is better Tool calling means all these searchers are using to verify the facts. So tool calling is better when it comes to perplexity because it is created for this. Fact check. Sometimes that's also a joke. No. If you have a script written in real, I always fact check it. I have this workflow in. And my problem goes something like this. I don't even say fact check. I say if there is a data point. If there is any data or if there is any point saying someone quoted this something read that up and fact check that line fact check that word and give me a checklist of things that you fact checked and was it accurate or not if it is not accurate by what margin is it not accurate so sometimes they said something and you say something they are true on both the fronts but what you say can be wrongly perceive it so how do you correct it so instead of just saying fact check it I say research find relevant articles and tell me if the data is right or wrong or anything nice because I don't want to just rely on fact check it yeah so now you can see it has done a bunch of tasks and it's doing it

1:18:25Raj Shamani:it's doing it

1:18:26Vaibhav:it's doing it

1:18:27Raj Shamani:so next thing is doing a very interesting thing that

1:18:35Raj Shamani:many people want to become entrepreneurs Okay. Okay. So, I want you to help me build a workflow or model or thing that I look at best opportunities in my core competency that at this point, which businesses can be made in India. Then, they will check the business model for me. What is the potential? That will tell you. And they will research where it will be made in the whole world or not made in the potential risk. What is the potential risk? How much money can be? What will happen? What will be the realist outcome? Based on just pure research, an insight based on a basis because the market is so big and not, where can happen, where can happen, which will happen, which will happen, which reasons for this type of business has failed, which reasons for this type of business has hit.

1:19:25Raj Shamani:This is a task. You can't think of yourself that I will start the business of the house. So how do I... The idea is to which category we have to do in the business, or do you not know that? I mean if in my mind there is a random business idea, then how do I check it out? Tell me an idea. Hair problem. Hair? Oh! All men in the world right now. Okay, so. Bal jhana is the biggest problem. And in India I think hair as a category doesn't happen much. Bal jhana is the biggest problem. And receding hairline or let's say hair supplements or hair transplant or maybe shampoo. I don't know. Should we narrow it down?

1:20:05Vaibhav:I'm thinking we should narrow it down. Right? Because I know if we will research about hair. It will eventually say, Hey, you have to take Finestride and Minoxidil. Because let's say it's making very, very nuanced.

1:20:19Raj Shamani:Let's say a horse example. So, horse is a horse. Because watches. I have an idea. Again. I'll tell you the idea. I want to start.

1:20:26Vaibhav:And this is something I've thought about.

1:20:28Raj Shamani:Okay. Then it makes a better. We can try. We can try.

1:20:31Vaibhav:But you think about it. If it's an interesting thing. I thought that. that male pattern baldness is very common in the country. Right? Most of like, I think 65 % of them. Yeah, everybody. Every third person. So my family knows that they have strong genes of hair loss. So I started to, you know, go into minoxidil and finasteride and all after dermatologist recommended, obviously. I don't recommend anyone to take it directly without a dermat. Domat recommended but everyday I have to eat a pill which is annoying. And I like gummies. Because gummies feels like a chocolate. So I always wondered in a gummy format why not do it minoxidil?

1:21:18Vaibhav:Maybe there's an opportunity. I've thought about it but I've never researched on it. Okay. Is there a potential to build a business like that? Where we can make a minoxidil gummy. Finasteride minoxidil gummy. Finestrite again has a lot of side effects for everyone. So let's stick to minoxidil. Because I think only minoxidil market bought better. So let's make it easier for people to understand.

1:21:41Raj Shamani:That if hair fall is a good thing. And baldness is growing. So to save it. Where to take it. Where to take it. Those ingredients can be added in the gum or not. And this business will be done in India. Correct. daily you eat a gummy so that your hair doesn't fall hair fall will stop

1:22:05Vaibhav:what do we want to know first? we want to know if this has market for it market is still there but okay

1:22:11Raj Shamani:let's do a research of are there people doing something like this okay plus I want like full end to end I want to make a full proof prompt or agent I want to give ideas and check about any idea or about just this idea? about any idea? I told you, I will tell you another idea, but I will tell you another idea. Yes, for that you have to make a clot skill. So what is the clot skill?

1:22:34Vaibhav:Clot skill effectively, if we have made a process. Basically, it is a reusable prompt. Okay. Or you have to make a project inside of chat GPT. Which we have last time. Write a prompt. Put it. Saying that your job is to do deep research. You are an idea machine for me.

1:22:51Raj Shamani:Yes, you can make it.

1:22:52Vaibhav:Make it a business model. Okay.

1:22:54Raj Shamani:So best business model. So people don't copy paste. Hmm. that if you have to check a business idea, this is the best prompt. So what will happen best prompt? Okay, what did you say? You need a simple agent

1:23:08Vaibhav:who will put any idea for deep research and all that give you a document that is a good idea or not a good idea.

1:23:18Raj Shamani:Good idea, bad idea, potential risk and then after that, why are other people doing it successful? why is it successful? why is it not? why don't you say what you want? okay let's speak Hindi, speak English in Hindi because I want to do it that the people who can do Hindi can do it or not so before that before that

1:23:40Vaibhav:I will write another prompt to make sure the prompt will be good okay this is a precursor of that what I say something like that you are an expert prompt engineer who has worked for the Google Gemini team for the last three years, writing some of the best prompts for Gemini Gems. What I want you to do is, I want you to write a super extensive prompt for a Gemini Gem that I'm trying to build right now. I'm gonna give you a description. It could be in Hindi, in English, in anyone. I'll give you a description on what I want the Gemini Gem to do. after giving that description you convert it a super extensive system prompt that is exactly what I want you to do did you understand so this is a nice little hack whenever you say something it comes to Hindi I don't know, Urdu Urdu is same I will select the pro mode so that it will think hard I just said this you are an expert prompt engineer.

1:24:50Vaibhav:In the end, I always ask, did you understand? Okay. So that he didn't say yes, no, and then you... He will understand me what he understood. You said something, I say something and something. The most of the times, the mistake is on us. That we didn't say it properly. Okay. So, take it. It will come to the understanding. It will come to the understanding. Okay. Okay.

1:25:22Vaibhav:I am saying before. While I am not a human, I am going to do my best and all. I have not seen this happening before. I am seeing before. Whenever you are ready, give me your description. Whether it's Hindi, English, Urdu or mix of all three. Here is what it will do. Behavioral guidelines, tone, response, whatever you need. Now you tell us what to do. Okay. So, I'm making a business model tester sort that will research.

1:25:51Raj Shamani:Whenever I give a business idea in a very raw form, I will understand structurally what are the opportunities in this problem. In this problem, I have said that the solution is the one and the other solutions. Who is doing this in India? What is the market size? Who has this business tracked in India but failed? people are. It's like that there is no business working on this business, but there is a opportunity or potential maybe can be a that is where. Plus, people who are doing this in the world, tell me. If there are any companies who are doing this, who are successful, tell me. Plus, how much money can be made, how much business will be made.

1:26:33Raj Shamani:How much money can be made? How much money can be made? What can the best outcome? What can the worst outcome? I need the skills and what I need. What can I do? and full proof testing of how much time line it will work for today it will be a big business or a big business and to start which I need to be a basic if not then which team members so I know you know, and end in the end, give a final check that you should do this business for next 5 years or not based on all the research and web research and everything that I've told you. Simple.

1:27:24Vaibhav:Also considering AI is around.

1:27:26Raj Shamani:Also considering AI is around.

1:27:28Vaibhav:Okay, perfect. So that is what I want you to write as a prompt. So ideally, we basically want a prompt, a business idea validator. so try to wear the hat of McKinsey or PwC or Bain's consultant I am a very good plus add street smartness

1:27:51Raj Shamani:add street smartness just don't be a consultant

1:27:53Vaibhav:just don't be a consultant add street smartness and sentiment analysis of social media as well and sentiment analysis of social media posts as well on that topic okay sentiment analysis so this will not be able to do

1:28:06Raj Shamani:yes but okay how much can it be how much can it be Let's see. This is coming. Whatever you said.

1:28:15Vaibhav:Always give me the output in English. I mean Hindi can do it. Okay. This is coming. Extensive prompt. Okay. And the reason I said you are a prompt engineer for Gemini Gems. And I used it in Gemini. It's because Gemini will know. Okay. Which format of prompt will work the best for itself. Take it done. Right. This is coming. Extensive prompt. Now you know, I'll tell you one more thing. Which is quite interesting. Okay. I'm going to start doing this. I will say something like this. Hey, I just read the prompt. I read this prompt. I think it's at Mac 7 out of 10. I gave this task to ChatGP. And the prompt that was quite extensive.

1:29:04Vaibhav:So have you thought hard enough? before giving this prompt? Why don't you think hard enough again and make sure the prompt is so extensive that nothing is missed out on. None of the nuance is missed out on.

1:29:19Vaibhav:Compete. So, I mean, this is basically when we say something, it is, there's another line that I use. Hey, you don't seem to be aligned. For whatever weird reason, it works. Right? So this does work. This is us trying to tell. Founders also say all of them. In models that challenge it more. I learned this from Larry Page. Larry Page has quoted. That when you scare the AI. And gaslight the AI. It will always work better. And since then I've been trying to gaslight it. So I'm listening to Google founder. And here it comes. It says here's the 10 out of 10. Hyper extensive system prompt.

1:30:03Raj Shamani:You are an elite, ruthless and deeply analytical venture capital partner and startup strategist.

1:30:10Vaibhav:Your objective is to brutally stress test raw business idea provided by user blah blah blah all of that.

1:30:15Raj Shamani:So, this prompt will be put in the logo so that they can take it directly from here. Done? Yeah. Done. So, this prompt will be put in the link below? Yes. If you want to test a business to someone, you will put in the link below. You will click on it and you will go. Done. So, people don't have to do so much work. We have to do it. No, it's not. we should do it but we will give it too. We do it and they will make it different.

1:30:36Vaibhav:Because I think business model we will make it easier to make it easier. But it can be better if we read this prompt and say this is missed, that is missed. Right? If we do that it can be better but I don't think we have the time. So I am going to bounce. Right? So this prompt came. I copied it. Now I will go to something called as Gems on the right side but it's free. Right? I go to Gems. remember last time we had a GPT project

1:31:03Raj Shamani:in chat GPT

1:31:05Vaibhav:it's equivalent to Gemini Gems I can click on new gem and I can call this business idea validator okay description can also give it this is the instructions part just paste it which we already wrote and then there is default tool so what I will do by default, I will on Deep Think or Deep Research also on but Deep Think, what I said, in ARK AGI 2 it's the best model right now, so why not? right? and I click on save now we have built a reusable tool in which we will just say an idea and whatever we have asked, he will give it So let's say

1:31:57Raj Shamani:Minoxidil is your idea Hair fall gummies Please validate the idea

1:32:02Vaibhav:Let's just explain

1:32:07Vaibhav:So my idea is that In India hair fall is a big problem So I have to make a gummy This should not be a tablet Or not be a liquid Which you put in your hair This should be a gummy Or be a chocolate format Which we can eat And I will my hair fall will be closed. I also want this idea to be backed by science. This means you know, you should not choose. We want an idea which is backed by science. And I want to build a product like this. So what can this product be? Who are the competitors? You have all the instructions. You do all of it and tell me everything that I need to know about this idea. based on the instructions shared with you okay I'm keeping it very simple

1:32:58Raj Shamani:our agents are working how do they do it? they closed it this is coming this is coming all the podcast you see? you see? open top 3 insights what did he get out of it?

1:33:14Vaibhav:Lenny's podcast founder insights comprehensive analyst of 301 okay look top 3 most valuable you could not beat it you wanted to okay top three insights don't die know when to quit the persistence paradox and he said how to implement it blah blah blah all of that insight number two AI compresses product development but raises quality bars build systems not just code which is so bloody true founder of lovable and actually multiple guests and multiple people have said something which is a good thing right insight number three radical candor okay this is not something new but yeah great right for people who don't know yeah which is Kim Scott's game and again multiple leadership experts saying here are your top three and now A implementation and engineering based on what I asked right model capabilities will 10x cost will drop 10x speed will improve 10x so build anything that you want to build going forward thinking that all these three boxes are being ticked this is wild yeah now what is even crazy a lot of people don't do this actually is there is no way in hell I'm gonna read this document I'm very lazy in reading I'm not a reader at all I'm very lazy so like for me it's all consumption rich media consumption so I'll convert this into a podcast to listen how?

1:34:37Vaibhav:notebook lm you've never used notebook lm? I'm using it a lot you're seeing it quickly okay let's say we have this document okay so first things first Gemini he's a deep research he's a brutal unfiltered VC breakdown of your idea the raw translation a clinically proven oral consumer gummy or chocolate designed to permanently halt hair loss eliminating the friction of messy and intimidation of the actual problem alternate friction damn sam som incubants already this is interesting the graveyard analysis hundreds of generic biotin gummy brands have died biotin gummies don't do the work the white space though is multi vitamin gummy and india doesn't need one more gummy bro it says build a gummy which is microdosed finestride biotin gummies don't make all of them microdosed finestride or high potent saw malmetto I don't even know what that is.

1:35:51Raj Shamani:Right? Not sold as casual snack. Like it should be prescribed by doctors. Yes. Finistride is prescription.

1:35:58Vaibhav:Yeah. But you know, right? How all these companies do. Right? You have a on-the-go, on-demand doctor. Nutra for all Hims and Hers. Here we go. And global benchmarks. Hims and Hers. But you look at

1:36:12Raj Shamani:some actual Oh, good day. Hims and Hers has made recently. Yes. Nice. Hell, let's choose. somebody is doing it oh look at this

1:36:21Vaibhav:compounding actual prescription of minoxidin and finasteride into a chewable hard mint

1:36:25Raj Shamani:so what you want to do is him and hers have done in US then you have to

1:36:28Vaibhav:go to US and test it neutrophil was acquired by unilever for 100 million dollars okay anyways that's the report we can go through it if we want to go down what's the last what's the the why go up a little final verdict that's it

1:36:49Vaibhav:biochemistry reality is that science backed hair loss cures and tasty chocolate gummies are fundamentally at odds real DHT blocking activates taste horrific degrades in gummy manufacturing heat and faces strict pharmaceutical restriction you said it can be done or not it can be done too so that's just saying it's not so high

1:37:10Raj Shamani:it's not so easy for it to be built which is cool and then down medically credible whatever that meant it it means it's not easy on this side until you are one more thing asked this must have what should happen what should happen yeah so you must assume that you can formulate the product you mean you can actually make it it's not very easy to make so if you're not buying it this is a slap see at this point this idea flop until we are able to find somebody maybe someone is watching this podcast who is a bio person who can make chemical like this pharmaceutical grade very potent gummy and come up with a formula then partner with us and then be able to do it

1:38:13Vaibhav:Just don't partner with Raj because he has better distribution. I also have a decent distribution. And I also care about hair fall quite a lot.

1:38:21Raj Shamani:So day one team, you need a PhD level formulation scientist.

1:38:25Vaibhav:Yes.

1:38:26Raj Shamani:To solve the API to taste ratio. I love how it talks. So in this API, it's not about your API. What? This is active pharmaceutical ingredient.

1:38:36Vaibhav:When I said API to taste ratio, I was like, you cannot solve everything with APIs.

1:38:41Raj Shamani:No. Okay. this is my health

1:38:49and a ruthless

1:38:51Vaibhav:performance marketer understands cohort retention that I can do so one part you can handle and legal shark you can be so much that you can at least get to the right people to give you permission

1:39:06Raj Shamani:and then third is people watching this podcast

1:39:12Vaibhav:watching this should be right to uh anyone you'll figure it out you'll figure it out if you want to reach out to us right okay now we said one more thing very quickly because we are running out of time that this insights are here if i want to consume easily right i will basically go to notebook lm okay click on create new this pdf I downloaded from cloud and of course the beauty is let's say you want to research on this topic beyond a point also the advantage is if I want to build a place where I can brainstorm on this idea I will not use a gem gem I will use to validate the idea okay but let's say that this gummy idea I feel very good.

1:40:07Vaibhav:Validate has also been done. We will find some PhD person who can do this. I will actually instead of uploading this. I will actually select deep research here. And put the same prompt here. Then what it will do is research all the resources. Then I can come to this window to ask questions. Follow up questions on that. So that is also how I can use notebook LM. But in this case I have uploaded this. I have already uploaded this. let's say for whatever reason I also want to make sure you remember all the transcripts that we downloaded they also have to attach so I can also add that I'll add upload files all these podcasts are also there right like transcripts I can attach this inside

1:40:59okay

1:40:59Vaibhav:so we have the summary which is the playbook we also have all the transcripts Just in case if AI wants to go and check. Okay. Deeper because there are mentions there. Who is it? Okay. Now, I want to learn about anything. Let's say, today I have a talk in my company. All hands. I have to give five minutes of, hey, in this AI first world, here are five things that you should do to stay ahead. And I want to do this talk. Right? And I need these insights, these podcast transcripts, and these documents. so I have to prepare for that specifically not just the whole thing of course I can say make a podcast of all this but I can go specific so I will click on audio overview I'll click on this edit button right I'll do a deep dive I can select how I want it to and I can say today I have all hands in my company and I want to take five minutes to address all my employees about how they should work on an everyday level considering AI is around.

1:42:08Vaibhav:So based on the document that I've attached, which is the playbook, and also the transcripts that I've attached here, go through all that information and build me a podcast which will prepare me for this talk.

1:42:23Okay?

1:42:24Vaibhav:I explained how I want the podcast to be from these sources. Okay? I'll click on generate. that's it and the crazy part is that now it's generating I can also generate this in Telugu, Hindi Tamil, any language that I want this is fascinating and of course I can make this into a slide deck also let's make a slide deck also

1:42:53Vaibhav:convert this whole thing the playbook that I've attached as a source I want that playbook to be convert to be converted into a presentation because I have a talk with my leadership and I want to share all these ideas with them as well. So convert the whole thing into a presentation by the way the outskill branding is black and neon green so make sure you follow that brand guidelines as well for me.

1:43:20Raj Shamani:As parallel as you're telling me in my mind so many things that I can do my job a lot better.

1:43:27Vaibhav:I think that is the whole point. I mean I mean, I just tell everyone that when you learn something new, don't think of it that I can exactly do it. That is never the intention. When I tell you, you have ideas. When you're listening to it, you're getting ideas. Don't let that go. Pull your phone and start typing it. Because after 15 minutes, you'll lose those ideas.

1:43:49Raj Shamani:True. So every time I say this a lot, when you open our podcast, you take a pen and paper or you take a phone or notes. and just keep writing, writing, writing and the podcast is finished. Then when you have 15 ideas just spend time on the one which you feel the best. In fact, what do you do?

1:44:06Vaibhav:I always have chat GPT or something open in a new window. I keep dropping voice notes. I start, I'm watching this podcast with Raj and the French president. As I'm watching that, I'm going to get a bunch of ideas. So I'm going to share those ideas on the thread. Give me a summary, but only when I say I've finished watching the podcast. still then just ingest all the ideas that I'm giving you nice nice very good way right so it will ingest ideas and when I say hey I finished watching the podcast then it will summarize all the ideas what you can do even better here is give it the transcript of the podcast as well so it can connect both the dots and make a document and now what you can do to keep on top of it is you take that document and put it on notebook LM and keep a notebook ready.

1:44:57Vaibhav:So that next time you create an ideas notebook and keep uploading these documents inside of it. So your ideas are never lost. So you can come here in the search box and say, I had this idea about some gummy. I don't remember because it'll be lost in so many windows that you have. You just search for it and you'll find you all that information. So you productize your life. This is what is living with AI. It's working and in the same way right you can also make a video overview. It can literally make out a whole video out of this. This is insane. And again I can explain it what I want. So I can say okay before I have a presentation with my team on this.

1:45:43Vaibhav:Why don't you teach me all the learnings of the playbook that is attached in a nice little video lesson. capturing majorly the top three learnings from this and go deeper into all of this. So you just the tools are there. Right? The tools are there. It's just that we don't know how to use it. And language if you want to change in audio overview here's a drop down. Hindi whatever language works for you.

1:46:16Raj Shamani:Portuguese Yeah, yeah. So in school it's called

1:46:20Vaibhav:Give it some time. It's gonna take some time.

1:46:22Raj Shamani:Tell me, I've done a podcast with you. I have to do with books. There are more books than I can read. Can I tell you something?

1:46:30Vaibhav:Every book that you want to read. All the excerpts of it. What is a book that you want to read that you have not managed to read? Give me one name. Quickly. If you can think of. Rational Optimist. Okay. So what I will do is, I will go to create a new notebook. As this is happening. Because we are in notebook LM only. Create new notebook. Okay. As I clicked on create new notebook. Look at this. This popped up. Open. I'll select deep research. What did you say? The book name? The rational optimist.

1:47:05Raj Shamani:By Matt Ridley. The rational. Oh the rational. I said irrational optimist. No.

1:47:11Vaibhav:Irrational optimist is me. All of us. All of us. By Matt Ridley.

1:47:20Vaibhav:I've been meaning to read this book for quite some time but I've not got the time I've already bought the book but I want to get a very good sense of what this book is what are the key insights out of it the stories that have been told from this that is relevant to me and everything around it why don't you do a quick deep research for me on every single piece of content that is published on the internet about this book right and import all the sources for me so I can learn from the book. And import insights. And import all the insights and everything possible you can do. It is smart enough. You don't have to say so much.

1:47:53Vaibhav:It'll import. Now, I'm not doing deep research. I'm doing a fast research because you don't have time. Okay. Ideally, I would do deep research. Click on the send button. So what it will do right now is you don't have to book. Downloading a book and uploading is not legal. Like legally, if you don't own it, I don't know if there's a legal way to download a PDF of a book. I don't know. Right? There is. Like you buy and then you can download a PDF. I don't know.

1:48:16Raj Shamani:there are few books which are available for sale official you download it perfect

1:48:23Vaibhav:if you can download it you can import it but right now what I have done is I have imported all the articles here that are written on the internet because I did a fast research it got me 10 deep research 100 articles once this is done it is very I am pretty sure that you know everything about the book is in this captured so I'll start asking it questions here about this book I'll be like and this GPT is better Claude is better Gemini is better what do you mean? for this here it does your work faster see GPT Claude and all this they can do the same thing but they're slower you have to do all these steps yourself deep research keep that knowledge download them as PDF attach them as files but which one's better?

1:49:10Vaibhav:I think 3.1 with ARK AGI score of so high I'm thinking Gemini will be pretty good there will be no difference in this I feel so LM Gemini base use

1:49:18Raj Shamani:LM is by Google

1:49:20Vaibhav:so it's Gemini

1:49:20Raj Shamani:that's why I I can get the same insights which here it'd be superior the way I would get in on Claude or it should be it'll not make a big difference is what I'm trying to say

1:49:30Vaibhav:but on a Claude or a ChatGPT you don't have this audio overview and all these features if you're someone who can sit and read I read then this is perfect then you can make a report out of it look at this there's a report I can just say let's say study guide or a briefing doc what do you want

1:49:50Raj Shamani:briefing doc for so that I have my you can add that context so I have to give my agent whatever I am using my project is made of GPT so I can give it there that is not need to give you can do it here but I think it's superior because it's more context than my context your context is more okay so I have to give my context Textually, I will make my brief.

1:50:13Vaibhav:Fair, fair, fair. Your history is there. Fair.

1:50:16Raj Shamani:So, in this case, which insight is relevant for me, which place will be better.

1:50:20Vaibhav:Then I will basically say, create your own. I will say, make an extensive document of all the learnings from this book. Make sure you don't miss out on any bloody thing. And give me an extensive document on this. So, it will make an extensive document on this. Then I will take it. Take it and you can put it. You are making a report. He's making it. He will also make it. He will take time for himself. But yeah, this deep dive is coming. I can start playing it. It has come from 300 sources.

1:50:54Raj Shamani:It is arguably the most nerve-wracking slot on the corporate calendar. The all hands meeting. Oh yeah, the town hall. Right, you've got the entire... These two are AI.

1:51:04Vaibhav:Yeah, there's a podcast created for me.

1:51:06Raj Shamani:The chat is just... Based on this document. And you have exactly five minutes on the agenda to address the absolute... That's your audio idea. And looking at the employee sentiment analysis in the source material you shared, the anxiety is, I mean, it's palpable. Yeah, it really is. People aren't just curious anymore. They're genuinely worried. They're sitting there asking, is this the quarter I get automated away? You know, is this the beginning of the end for my role? So today on this deep dive, we are doing something a little different. We aren't just analyzing the AI landscape from a high level.

1:51:38Raj Shamani:We are effectively prepping you for that specific.

1:51:42Vaibhav:So it's still setting context right now.

1:51:43Raj Shamani:Consider this your rehearsal. Exactly. We've come through a match. Okay, let's say it's taking too much time. I can just join. Oh, hey, our listener wants to join in.

1:51:52Vaibhav:Hey, I really appreciate you guys setting the context. I really appreciate you guys setting the context, but could you guys jump to the meet right away, please?

1:52:04Raj Shamani:oh wait someone wants to join hey go for it hey i i really appreciate you guys uh trying to set

1:52:11Vaibhav:the context but can you guys jump to the meet right away please because i don't have the time

1:52:18Raj Shamani:oh absolutely we hear you yeah we totally get it and we know time is valuable right we're just about to jump in most people are back and most of us do zero preparation for meeting We can skip the broader framing. No problem. We can jump straight to the tactical advice. Because the core of this is about those three pillars. We're talking about prepping you for that specific five minutes.

1:52:38Vaibhav:Someone can talk to me in the podcast.

1:52:40Raj Shamani:Right, the all-hands meeting where everyone is worried. Let's focus on the message you need to deliver. So we'll dive right into the run of show for that speech. We want to make sure we cover the three key pillars and the specific actions to take. We're cutting straight to the meat. We've distilled the noise down to the core mandate. So you walk up to the mic. What's the very first thing you say to cut the tension? You have to address the fear head on. Right. No dancing around the anxiety. The best reframe we found from C.D. Arnovitz and Brendan Foody is this. It's the quote that stops the whole man versus machine debate.

1:53:13Raj Shamani:The line is, you won't be replaced by AI. You will be replaced by someone who's better at using AI than you. I love that line because it shifts the locus of control. It's not the machine coming from me. It's about competitive equipment. It turns fear into a competitive mandate. Apathy is the real enemy here. Once you establish that rule, you move right into the roadmap. The three specific pillars you need to communicate to your team. The builder mindset, the mandate to play, and the shift to taste. Let's dig into that first one, the builder mindset. This fascinating concept comes largely from Julie Zuo.

1:53:44Raj Shamani:She observes that traditional corporate silos are crumbling. When we talk about silos, think of the classic assembly line. The product manager writes a spec. The designer makes a mock-up. The engineer writes the code. that's the classic slow waterfall method and execution is incredibly expensive in that world

1:54:01Vaibhav:so yeah i mean what generic but that is what we got from the insights what will this model do so what we got from all these podcasts is a generic this insight right so it was able to convert that and prep me look it's also is generic because you and i and all of us have been on top of AI. It is not so generic.

1:54:23Raj Shamani:For other people. Most of the people. Most of the people. But the idea is that you can do it like that. You can get to the tech. The tech is the fact that you're able to condense any information. Plus I think the better would be if you just we just put it in the prompt that instead of summary give me three exact mind-blowing insights which no one in the world is addressing head on those insights are going to address and only talk about that absolutely

1:54:53Vaibhav:so then it'll be like much better yes i mean we tried to simulate an environment right now so i think uh it's just about the founder example yeah that is probably not relevant to most of the people who are watching this is to give you ideas on what you can do with this thing true true right

1:55:11Raj Shamani:and here comes the video okay welcome today we're going to distill the ai first company playbook think of this as your essential briefing. I got it. The final prep, you just start with the stakes, right? Because this isn't just another tech trend. No, this is a fundamental test.

1:55:28Vaibhav:So it's made a whole video. Nice.

1:55:30Raj Shamani:Your team takes its cues from your actions, not your memos. If they see you in the trenches.

1:55:36Vaibhav:Nice. I like this insight better. That insight will be done. We didn't have the patience to listen to the 10 minute thing.

1:55:45Raj Shamani:This is good. Does the CEO use chat GBD? Your team feels safe and empowered. Now what? How do you actually build stuff? Well, the playbook is super clear on this. You've got to build like a scientist.

1:55:58Vaibhav:The days of just vibe coding and hoping for the best, they're over. And the absolute cornerstone of the scientific approach is something called evals or evaluations. Seriously, this is it. Before you ship anything, you need a system to measure if it's actually working on the specific task you designed it for.

1:56:17Raj Shamani:This is without a doubt.

1:56:18Vaibhav:You know what evals are? In simplest way possible, you said what I know, what it has done is wrong or not. Those are in the simplest form evals. And it's saying set up evals for everything.

1:56:31Raj Shamani:Because Bro, this just makes me a top level expert and CEO and founder every time before I address my people. Because then I will be like so prepared. Because obviously when I will study this, I will put my context in what I want to communicate in. And just I'll get framing to say certain things in a much deeper. Wow. Very. This is good. It's quite insane. This is good. This is obviously like, this result is not good for me. But this tool and the way the process that you have made. I'm gonna exploit it a lot. I will use it a lot.

1:57:12Vaibhav:It's all about what we entered in this box. We put two lines and expect that the world is not.

1:57:21Raj Shamani:This is very good.

1:57:23Vaibhav:But I like how you have expectations from AI. I do. It's funny how the world is moving dude. like I think it can do better is what also scares me

1:57:36Raj Shamani:okay tell me last two things okay I want to talk about these two things specifically this open claw and this mold book what are they doing open claw founder one person made a whole company billion dollar conversation what is all of that

1:57:55Vaibhav:yeah man I think the world overnight changed from Jan when this open claw thing happened so open claw again is a scaffolding like some new technology someone thought about how should an assistant work these were a lot of small things that our assistant should remember what you've said about memory all these models are not so great with memory how can memory be better AI is supposed to execute tasks for us proactively now we have to say something then AI but I want AI to work proactively right for me knowing what I do in the calendar in the calendar it has a podcast if it has a calendar access so it has a podcast Raj's so we have to do research so that Raj has enough material before the podcast so it will only do the research and send it to you before even you ask for it is proactiveness nice with all the data it has.

1:58:59Vaibhav:So proactiveness is baked into this product called as Open Claw. And on top of that, this is an open source. Because of which people love that even more. Right? So it was scaffolding of Cloud Code. Cloud Code was made for developers. He built the backend of Cloud Code. He built memory, skills, and all these things. And built this product called as Open Claw. Which was an open source thing. and when people start using it their brain is broken and for the right reasons because it can do so many absolutely crazy things like I'll tell you the things that I use today I use Open Claw to read all my slack because per day I get 200-300 messages it's impossible for me to do it so Open Claw reads my messages every 3 hours and I give a summary send someone in his slack that these are things that have happened the last three hours, these are things that you should care about or sometimes there is nothing that I should take action on it'll be like don't worry about it, everything is fine I just let all your slack it does the same thing proactively on emails and calendar and all of that it can manage my calendar effectively, I always say something it can make phone calls on my behalf it has become my alarm also so I just tell OpenClaw I have a podcast with Raj at 8am wake me up at 7am and call me till I lift and say I'm up.

2:00:28Vaibhav:And this happens on Apple too? It will happen.

2:00:33Raj Shamani:Do you book flight tickets? There are many books. Not just randomly. I'm going to go there and I said before I'm going to get on. My very impromptu trip. It'll ask you for the payment link.

2:00:44Vaibhav:It'll give you the payment link to pay. Without it, you won't be able to do it. In this case, you don't have access to credit card. You'll give it. Exactly the point. Exactly the point. RBI will not allow you to do India. OTP will ask SFP. So no AI can work in India without. But yesterday Razorpay launched agentic payments by the way. Less than x value payments can happen on the go by AI. Using MCP. Yeah. Using UPI something. Wild. Yeah. So that's also happening. So this open cloud is running its main system? So there are two ways that you can do this. I will not recommend this to run on their own system while I do because I live bravely I guess but most of the people should not do it because one is that technically it can be very complex for people because they have to install it in the terminal just the process is overwhelming for people but if you want to set it up quickly and you want to start using it if you want to safely or securely you can use a platform like Hostinger okay right hosting there is something called as VPS which in the which in the simplest way possible if I have to explain you it will run in your computer where you can run open claw because it's running on a cloud it is safe secure and all that good stuff comes with it so it doesn't get into my system it doesn't get into your system it runs on cloud because it runs in cloud it runs 24 7 you don't have to when I play in my own system I have to keep it on I have to keep the system on otherwise it won't work but it's in the cloud so I don't have to think about it my Mac mini is on or not or my laptop is on or not or it's closed and all of that is not needed it will run on the cloud and it will operate on the cloud basically

2:02:34Raj Shamani:so tell me two or three things that if I play on my PC on my Asus so what will there be a problem will come in versus I'm going to hostinger so what will I get safety? When you're running on your computer

2:02:48Vaibhav:you're giving access to a lot of things on your computer which I probably don't even know. Yeah so prompt injection can be someone can hack their way into it because it's available on the cloud on the open so it's basically you opening a back door

2:03:04Raj Shamani:to it. I'm inviting people to come.

2:03:07Vaibhav:Yeah yeah so basically that is what it is and there are a lot of safety and security issues coming in but I'll be very frank now open clock opened and acquired now there will be more investments that will go into it so it will eventually become secure right now it's been secure with plugins and all of that but forget about security also right while security is an important factor if I show you to install or show you the audience will show you most of them will drop off from the video because there is terminal you have to put commands in you have to copy and paste API keys just the UI if you are stuck in the problem then you'll be like you have no idea because you don't know what command to write of course if you're smart very user friendly at this point it's not user friendly at all then also it is so popular right but in hosting or if you run it in VPS

2:04:01Raj Shamani:it is very simple it's three clicks and you are done basically so tell me now what will happen with it? Which 2-3 things that will happen for me if I put open cloud messes? It will proactively work for whatever you want.

2:04:18Vaibhav:If you do whatever task every day, let's say you follow geopolitics a lot. Let's say you say these are 10 speakers that you're gonna do podcast with in the next one month. And you want to be sure that if there's information about anything, information about people, I need to know that first. Yeah. So you can just tell your open claw, hey, I'm going to do podcast. And you can initially, you can tell yourself in your own way. Right? That I'm going to give you context. Then you will say, these are the 10 speakers that I'm going to do podcast with. Every single day, overnight, you research about them on the internet.

2:04:57Vaibhav:And 9 o 'clock in the morning, you send me a brief on my Slack or Telegram or WhatsApp about anything that has happened with any of these people the other thing that I use it for is stocks I have a bunch of stocks and I forget I'm not on top of stocks and I think I can manage it but I can't so my open claw the moment market opens in India and in the US I've given it the stocks that I've invested in my zero-dha kite account via API integrated and in IND money which I invested in the stocks in the US I said the stock value I've invested in these stocks these are the values that I own once I dumped information and I said remember this everyday when the stock market opens in the US I say read about how these stocks are doing if the stock has gone up or gone down do a research on why it has gone up and gone down and give me advice on it so that by the time I wake up I have insights if I have to take action on if there's no actions to be taken don't bother me so basically it runs something called as a cron job so they're working in the back end and they're always bothering you when they're doing something they're taking action yes but people are using this in wild ways what do they do so you have a software product which is a 1 % billion dollar company that's how it will be this was a 1 % billion dollar company This is almost a 1 % billion dollar company Nobody knows the value of how much Peter was acquired but basically Peter was acquired, Peter is the founder of Open Claw He was the guy who was acquired by Open AI and I don't think that will be less than 1 billion because in Lex Friedman's podcast he was saying the number is something that you can guess in that direction he was talking right, he was probably the first 1 % billion dollar company So how can you install Hostinger?

2:06:59Vaibhav:If you want to install Openclaw, you can come to this Hostinger link. We will share that in the description. So anyone can use it. This page created Hostinger. This is created so that it will be easy to install Openclaw. Because the problem with Openclaw like I told you is that To deploy it is very technical. And Hostinger has made it like a couple of clicks. which I really like right so here we are on the page you can either choose this directly or you can pick up the KVM2 or a KVM formation I'll tell you why I picked this up because I think it's a sweet spot it has the right processing right RAM which is fast it can scale you can run multiple bots if you want and if you don't want to scale later you can do more resources that too okay I'll just select choose plan quite simple

2:07:57Vaibhav:and now you can basically select the period that you want let's say 12 months okay i select the 12 months now we have a very cool code i spoke to the team and we got a code for people which is a 20 % off which i can apply it's figuring out ai let me see if it works perfect we've got a sweet 20 % off this is something that you can do right apply the code and again we'll put that code down too thanks Hostinger to make that happen but yeah next is something that is very cool I feel and this is only Hostinger so it happens that Open Claw has no AI Open Claw is a scaffolding like I told you so it needs an open AI or an anthropic to be connected with it and what happens in the usual setup What happens in the world?

2:08:46Vaibhav:You have to find the API. It's quite simple for the people who don't understand the technical API. So Hostinger has actually made that simple. They have connected the Nexos AI and it's ready to use AI. If I just tap this, it allows me to select a plan which I shouldn't give myself an API. So it works out of box for OpenAI, Antropoc, Google and XAI models. So that's what you do, which is like a differentiator. Okay, you can add whatever you want to and just simply click on continue. And that's it. You can quickly register with whatever email ID and whatever account you want.

2:09:33Vaibhav:And just pay for it. That's basically it. And let me show you the whole setup very quickly as well. So the payment is done. Right? Once I've done the payment, it's quite simple. You can do it from UPI. That is also one more advantage. because they have Indian payment gateways. Now, here comes the interesting thing. If you see this, right, there are three things that you're able to find. OpenClock gateway token, WhatsApp number, Telegram bot, you can do all these in later. You just have to copy this code and keep it somewhere handy because this code will be useful later. Okay, so I'll just keep it handy.

2:10:05Vaibhav:Of course, you can add your WhatsApp number if you want. If you want to connect Telegram, you can do that now. That's it. You just click on deploy. You're going to take about 5-10 minutes and a computer will open inside which is a VPS install all of open claw and the whole setup will be ready for you to start playing with it right and I'll show you a couple of things as well once we get in so let's give it some time so we have our VPS setup right now okay VPS has its setup now it's asking me a bunch of questions which I can skip okay this is the high level it's now running right Now you are showing this button here, this is the URL that you need to know.

2:10:46Vaibhav:As you click on it, I told you that you should copy a token and paste it here. So this is the connector and that's it. We are inside of Open Claw. For people who were like 2-3 hours technically, we are done in like a couple of minutes. Now, how do I connect Telegram? Don't break your head. Open Claw is smart. I can just come here and say, how do I connect to Telegram? You just talk to it. Everything that you want to do, you just chat with it. It gave you the list of instructions that you have to do to follow. Simple, right? Or let's say I want to say, what are the things I could do with you?

2:11:32Vaibhav:And here it comes. It tells you everything it can do and more, right? I can also say okay let's set it up now for best performance and it'll tell you a bunch of things that you need to do. So we can definitely do this much more bigger deeper dive on open claw but I'll tell you a simple rule of thumb because we don't have time now. If you have to do anything anything that you want to do don't think that this AI can do or not. can do it. That is the rule of thumb. If you need new tweet ideas from Twitter, you just tell me about this. Figure out. And Openclaw is smart enough to figure it out. If you have to write a code for yourself, he will write.

2:12:18Vaibhav:If you have to find an API, he will find. Whatever it needs to be done, it will do it for you. So you don't have to worry. Just go wild. Go wild with the ideas. Play with it. Next time we will meet, we will go deep into this. Done? Awesome.

2:12:33Raj Shamani:That's it. Sick. One last thing. I will watch your video, Sweet Pea. What is this? It looks fascinating.

2:12:42Vaibhav:Bro, the world man. Now so far we've been talking about digital. I think real AI is physical. So again, right? We have bought it and released it. Basically because we are on top of AI. Right. But basically Sweet Pea or it was also called as Project Dime. Is a... What is it? What? Ear accessory. Okay. It works with bone conduction. So, it doesn't close the whole ear. It looks like it. Okay. Based on the patent that they have filed. Chinese leak everything. So, I read the patent paper. So, it works on bone conduction. It keeps it looking like it. Basically, it's an on-body device with AI all the time.

2:13:24Vaibhav:Okay. So, I can talk to it and can talk back to me effectively. Basically, it's ChatGPT on the go. is the phase one of this device so now you need to pull the phone and message you are just asking the question hey this is what I want to check can you quickly check it will talk it will find that information and talk back to you but because it's bone conduction so your ears are not blocked so you can keep that device on 24-7 that is the directional thinking is what I am able to understand from the device choice not the earphone that is level phase one but this is something that people are not talking about again you can say it's speculation because I'm connecting the dots recently Sam Altman became co-founder of a brain interface company okay it's Mirage Labs or Miyake Labs I can't remember Mirage Labs I guess something like that invested I think a bunch of millions into it and became a co-founder Mirage Labs does something Neuralink does which can know what you're thinking Neuralink, you have to do a surgery to insert it but this is not invasive this is non-invasive it works with electromagnetics you remember last time we spoke about alter ego a device that can sit on this it's an extended version of that OpenAI's co-founder ended up investing and becoming the co-founder of that company why do you think that is happening?

2:14:56Vaibhav:we are moving into the age of thinking to response now Raj has asked me something I don't know his answer but he's on camera what do I say? so I'm just thinking what should I answer back of my mind that device is listening Miryaki Labs is listening processing on the on cloud device and answering back to me through my ears now this is not crazy you know what has happened like the world is evolving Open Claw a new open source model Vision Claw which is some developer built it which you can hijack your Ray-Ban Meta glasses for if I wear those what I can do is basically I have Open Claw access through my glasses and it can see what I can see right now sick and Meta also recently announced that they are planning to turn on facial recognition and feedback on Ray-Ban Meta glasses via Meta AI because if they take a picture of you Instagram, Facebook even if you're not Raj there's an Instagram profile there's a Facebook profile and Meta has all the information about your profile already through glasses through bone conduction again I can know who you are without even meeting you or without even saying hi to you these are all things Raj we have spoken about in the past yes they're all coming into action right now People used to think we were speculating.

2:16:25Vaibhav:They're all happening. It's exciting and scary. And biggest of the companies are investing. Apple invested in QAI. I think something like that. Q-U-E, something like that. Which can understand what you're thinking by your facial movements, not by talking. On how your face is moving, how your muscles are contracting. This is wild. and they invested in it and they plan to integrate that into Siri. Sick. So next time you're in tension, your Siri will probably talk back to you saying, dude, it's fine. Why are you so worried about? You've seen worse shit than this. That's the world we're heading towards.

2:17:10Raj Shamani:Wow. Well, thank you so much for leaving me on spiral of thoughts again and teaching me so much. I love this thing about you that you've grown up. Thank you. You're a good teacher. It's just, I don't have to prompt questions or I just have to dig deep and try to understand certain things. I just have to ask you one question and then you don't have to say much. Like that's a very good thing. You're a good teacher. You teach me so many things and you make it easy. And you also know that when you wait, what do you want to say in that time?

2:17:44Vaibhav:I wish we could go deeper. I think the audience has to tell us.

2:17:48Raj Shamani:Right. Do you want deeper or no? Let us know in the comment that we should make all deep content or make it a base level.

2:17:55Vaibhav:We should make a little technical or make it a little agent building.

2:17:59Raj Shamani:Things like that. If people want, then they can comment and let us know and next time we will make it. Done. Perfect. Thank you so much. Thank you. Thank you so much for watching this episode until the end. Now, you have to do three things. Number one, subscribe to this channel because as much as you subscribe, we will get guests to you. Number two, please comment we have told you what we did wrong and what we did good so that we can learn from the feedback and learn more and create more better experience. And number three, this episode must be shared with someone else because one conversation is enough for changing someone's life.

2:18:36Raj Shamani:I'll see you next time. Until then, keep figuring out.

2:18:48Thank you.