Abhishek Pinninti: Hello everyone, ⁓ welcome to VTLS pod. Today we are with Arun. I would say he is an AA enthusiast. ⁓ When I was going through his profile, have seen him talking about AA a lot and the trends about it. I wanted to put it out to you guys that it can help the community. Hi Arun, how are you doing man?

Arun: Hey Avershik, thanks for having me. I'm doing good today, how are you?

Abhishek Pinninti: I'm good. So yeah, tell us about ⁓ why you know you've been creating content around AI, what is the motive about it?

Arun: So I am a robotics engineer. I work as a robotics engineer. I currently live in the US. I did my masters in robotics and I'm very passionate about it. But recently the advancements in technology and AI that we have been seeing is massive. And I felt we can't it easy or take it light because it's got a huge potential to revolutionize almost every industry possible. ⁓ If you had told me five years ago, that AI agents can code, detect, diagnose diseases, can develop websites, can write software. would have just laughed at your face. But today it's all possible. We able to do it. We have AI system diagnosing diseases. We have AI agents writing code, building websites 24 per 7 around the clock. So I felt this transition is not so smooth because five years is such a short time. Even five years is too much because chat GPT was exposed, was released to the outside world in 2022 around November or December, guess, memories, right? So it's less than five years. It's almost just four years, three and a half to four years. And the massive growth that we have seen in the last three and a half to four years has been so massive since the transition is not smooth. I feel it's been very difficult for people to accept the fact or digest when ⁓ AI systems or agents are being ⁓ used to replace human workforce in their job. The main reason I started doing content in AI was I wanted to explore this technology, learn and also share my awareness, share whatever I know to the outside world and help people upscale themselves in AI so that they also can adopt the new technology. and survive in the new technological era.

Abhishek Pinninti: Right, and I feel there is that the thing you're doing is really good because with the technology evolving every with new tools coming in there is no particular I mean there's no space that people can be educated about it, you know.

Arun: you

Abhishek Pinninti: ⁓ It's all the transition is there every day some of the other tourists coming up some there will be new launch from one of the other market leader and It's I mean, it's good that you're putting it out. So when you say that, know, the transition is not smooth Why is that why is that statement that you have given that the transition was not smooth and what is that you feel that you know ⁓ I would say that there is a sudden boom.

Arun: That's an interesting question. So I would compare this to the era of computers when computers became so common, they came out for mass adoption, let's say in the late 90s, earlier or late 90s, we saw the same trend. People were ⁓ protesting against computers. People are not ready to accept the technology. ⁓ because computers were back then used to replace a lot of human workers. For example, bookkeepers, ⁓ lot of clerical posts, clerical jobs were replaced with computers. For example, a bank would have needed 10 people to manage the books, but all those 10 people could be replaced with just one computer and just one operator for it. So people were going really wild when they realized computers were taking away their jobs. I see the same pattern happening now, but... These days, computers have become inevitable. Irrespective people protesting, they were ready to accept it or not, it became the new normal. Right now, computer has become an essential thing. All of us need it, and we are all using it in our everyday life. So there's no point in protesting against the new technology coming, and it is replacing humans. So the right thing to do would be to adopt the technology, learn how to use them. So. The reason why I said the transition is not smooth is nobody saw this coming. if, let's say, if it had happened in a gradual pace, step by step, people have got perpetuals. ⁓ Getting used to these tools or getting an idea on how to use these tools or how to navigate it is a little challenging. I don't feel everybody is exposed to it. Not everybody can do it. Let's say somebody was being... ⁓ I'm not saying you've got to be tech savvy to use them, but still it needs a little bit of guidance. When these tools rolled out, maybe the Gen Zs and Gen Alphas, they were quick enough to adapt to it, but the millennials, baby boomers, and the previous generations, I feel they find it a little... ⁓ difficult adapter because of the conversation that I had with the 40 years and 50 years old. They still didn't have much idea on what the technology could do, what it is possible whenever they see an image. And when I tell them it's A-generated, they're just taken by awe. And it's very hard for them to believe that A is able to generate such authentic, such realistic images. This is just one example. just imagine ⁓ somebody

Abhishek Pinninti: Yeah.

Arun: who was in their 50s and they get replaced by AI in any profession. So it's very hard for them to digest it and it's very hard for them to even get use of the technology. That's what I meant the transition was not smooth. Let's say this entire advancement happened over a period of 10 or 15 years because the computers we use these days, the mobile phones that we use these days, they didn't happen overnight. The cell phones back in 90s were very bulky and for us to get the first touch screen phone, get the first iPhone, it took almost 10-15 years.

Abhishek Pinninti: Yeah.

Arun: ever since the first commercially available cell phone was made. But it was not the case with AI as a technology. We saw charge GPT rolling out in late 2022, but now we have so many competitors. It's going crazy. We have agents. The initial days charge GPT was terrible. It hardly got answers right. It's like a lot of human intervention, know. With any stuff you did, it hallucinated a lot. A lot of factual errors, but today it's not the case. You have so many other agents, so many other... tools with which you can do marvelous things.

Abhishek Pinninti: Right, when you're saying this I want to talk to you regarding a real-time you know an example that me and my friend were you know having this conversation. He works for one of the Fortune 500 as a developer right ⁓ from India. ⁓ So when this transition of coding being automated by now cloud code has become inevitable. In the last two months it has become inevitable. I'll tell you on another example. So I am building an AI tool to automate Meta and Google Ads and optimize them to improve ROAS for e-commerce business. So we start building this tool for the last seven months. The project cost and the project timeline was around 15 to 18 months to complete the project and there was cost for it. In the month of Feb, the project cost got down to 50 % and the timeline got down to 12. So I mean, now it's just one month we are going to launch. This is one of the example. And now he says that whatever stuff they're doing as a backend developer to his clients from his company cannot be replaced by AI. Can you, you know? Give us a brief about this what is going to be the future of back end developers in coding. Is it going to be the same when you say 10 people is needed now 2 to 3 guys who knows how to use cloud code and put it in right place and it's done.

Arun: Okay, so I'm not sure what stuff they're doing at your friends company, but if you ask me in general, if you ask me about developers or software developers, programmers, I honestly feel AI has got the potential to replace them. I'm not saying they'll replace them 100%. For example, before AI, if a team...

Abhishek Pinninti: Yeah.

Arun: in a software company, a team requires, let's say, 20 engineers to do something, now the same stuff can be done with, let's say, five engineers, five to 10 engineers using AI. So it does increase their productivity. So I've been seeing a lot of debate on this going on. Programmers also say, know, ever since we started adopting AI into our workflow, our productivity has actually gone down because it requires a lot of time for us to troubleshoot.

Abhishek Pinninti: Yeah. Thank you.

Arun: and figure out things because there's a lot of bugs that's happening. If the code was written by them, it's pretty easy for them to understand or debug what it is. But if they are writing the code, it takes a lot of time and understanding what actually is put out there. And once you understand the structure, only then you can just go debug the code.

Abhishek Pinninti: Yeah.

Arun: There's been mixed reviews going on, honestly what I feel right now might not be there 100 % but it's getting better day by day. We can't dismiss it. If we take it for granted, then we're going to be in serious trouble. it's high time we all start acknowledging that AI has got the potential to replace us all someday. Or instead of saying replace, I don't like the word replace. I would rather say it would make a life better. ⁓ it would end in your life. So, so software developers, because with myself, I've seen that improvement happen. ⁓ like recently I made a video on that ⁓ four years ago when when Chagipati was brand new, when I used it for writing code and I used it for building a website.

Abhishek Pinninti: try to evolve with it.

Arun: It was so horrible because a of bugs. I had to manually intervene, check for what stuff was going wrong. I've been getting a lot of, when I tried building it, building the project, running it, a lot of errors. I had to manually intervene and just look for bugs. last year, was just, I, Last year, I tried building a website for my friend's wedding. I just wanted to give him a present. And so I just was building a website. And it was pretty smooth. I was amused. I was actually using that for building a website after a very long time. And it was pretty smooth. I'm just talking about the free version. And there was not much of bug. And it was a very simple website. And last month, I was able to build an entire application, front-end and back-end. Over the weekend, I was able to build a complete full application and that's how good it has become. But I'm not saying AI is completely there to replace ⁓ a senior software developer in a company. I was able to build a prototype, but still it had issues. If I try putting it out in production, if I try deploying it, obviously it's going to break somewhere. ⁓

Abhishek Pinninti: Yeah.

Arun: you need someone who's got the experience and the knowledge to troubleshoot how to get a deeper, get it scaled up. So that's still there, but it's getting better day by day. So we are talking about cloud. So in the last couple of weeks, they have been putting out some crazy announcements. They've been building agents to do code reviews, do PRs and all those stuff. So it's been quite crazy.

Abhishek Pinninti: Yeah. I mean.

Arun: So honestly, the long story short, I feel it's not there yet to replace as all ⁓ 100%. But it is obviously companies have started taking it seriously and companies are restructuring their teams. are trying to incorporate AI into their workflow. ⁓ They are, because a lot of layoffs have also been happening in all the top companies lately.

Abhishek Pinninti: I'll tell you another part of story. I've seen a company that has come and used that they have laid off some 500 employees by embedding AI into their systems and after one week their servers got down. So it's... As you said that it is not yet there but over the time it is definitely the same which computers have done to humans. The same thing AI is going to do to due to the humans and I think they have to be prepared about it rather than you know just complaining about it is what my view towards it.

Arun: That's true. I also like to add, since you brought that up, I also want to quote another similar instance that happened with Amazon Web Service, AWS. So last year, think it was in November, October, late last year, there was an AWS outage. There was a global outage. I'm not sure if it's global. It was just North America. There was a huge outage. the outage lasted for almost 11 hours, I guess. And the reason why, after the issue happened, they just did some investigation on why this happened. And the reason was them incorporating AI into their workflow. So when AI agents were given access to write code, so instead of fixing a bug, they decided to delete it all and start afresh. So that caused a huge outage and it costed the company almost millions. But that didn't make them go back on the decision. didn't. Let go of it. They worked on more processes to make sure this doesn't happen. For example, recently I read something on X and I'm not sure how authentic it is or if it's actually what's that happening in the company, but I would still like to share. So they have brought in new processes where a senior engineer has to approve any air written code by a junior engineer before any changes get pushed. into productions, a senior engineer has to approve, especially if it's AI written. So they're not giving it 100 % control, but still they are having it in the workflow. They didn't decide to go back on it, even though it just caused a lot of chaos. So companies are looking to adopt over it, make the process much smoother. They're not ready to willing to give up because I think they all know the potential of what AI has got and how beneficial it can be for them.

Abhishek Pinninti: Right, I want to talk about the biggest concern about this evolution is security and compliance. This is the biggest concern of the users that see when we use openclaw when we use cloud core when they when all these agent cares are you know embedding into our coming into our account systems checking out our payments you know completing the whole task where there is ⁓ data involved and ⁓ cash involved So how do you think that the security and compliance body should be evolving? Because there is no particular body until now. There's no body that we can go to an authority and say, hey, OpenClaw has to just wipe my card for $1,000. Who is going to take my complaint? And how are? You know, founders or the companies are trying to handle this. See, that is not fair that open pro just goes and swipes my card for $1,000. How is that fair for me as a user? So what do you think about this?

Arun: Okay, since you brought up OpenClaw, let me just give you some clarity to you as well as the audience. If you're planning to use OpenClaw, use it with caution because it has got a lot of security vulnerabilities. It is someone's weekend project. That's how it started, OpenClaw. Someone else's side project. They just want to wipe out something. And it was an experimentation. And it has got a lot of security vulnerabilities. It's not tested extensively. So when you're trying to run it, make sure you run it inside a sandbox, like a virtual box, or you run it on a separate computer which doesn't have any of your data, because there were so many stories where a lot of data was compromised. ⁓ But other well-written products or well-developed products, for example, other AI tools, ⁓

Abhishek Pinninti: Yeah. Yeah.

Arun: they don't have such vulnerabilities, but still there are some data risks involved. ⁓ So most of the companies, all the agents, all the AI agents, all the AI companies, there's something called an MCP, stands for Model Context Protocol. That is one framework which they use with which you can isolate your... ⁓ Your data stays local. It's not exposed directly to the LLM or it's not shared anywhere to the cloud or to any third party sources. So there are so many methods that are incorporated to make sure your data is private and safe. It's the same as using social media. So you're putting anything on social media. It's going on to some third party cloud. Your data is compromised. It's not safe. So that's the same with using any LLM you use, Gemini, HAT, GPT, Cloud, anywhere. So... It's just being shared into their, ⁓ the data is shared to a third party and most of them use it for training their models. So it's just to the users, the users got to be more careful in what stuff they're sharing. So don't share any private information like your social security bank account, data, birth phone number, and stuff. Make sure you're very careful with that. ⁓ And addressing the next point that you said, obviously there's got to be some governing body that has to overlook these features, safety and stuff. ⁓ Since I said this technology is fairly new ⁓ and the growth which it has seen is so massive and it's in a very short duration, nobody was prepared for it. So it's all happening in the back end and I feel very soon it's... ⁓ as it's going to be out there. Some governing body, maybe some international bodies. And above all, the company's got to be more ethical with this. They've got to be more transparent on how the user state is being used. I hope I answered your question.

Abhishek Pinninti: Yeah, I'm just you know taking that call of you that which is that governing body coming in again It's just you as body sitting down there and you know telling the world that you have to do this you have to do that but yeah When I say ⁓ Claude, G-P-T, Perplexity, Gemini How do you rank these and What do you suggest for someone who is confused of what they have to use which tool?

Arun: That's a little complex question to answer, ⁓ but yeah, I think you can answer for that. So it comes down to your convenience, first point, whichever tool you're comfortable with. You can experiment, try out all the tools and feel whichever tool works the best for you. That's one way to decide. In my personal... in my personal experience. Right now, every tool is getting better because they are all out there in the same market. One has to compete with one another. So maybe at least in initial days, you can easily classify them saying, OK, this one is better than this one. It's so terrible, so horrible. But now almost everybody is getting better day by day with each update. If one tool is not performing to the mark, if it's not performing as good as the other one, obviously people are gonna just bash it out there in social media and the companies are seeing them and they just work on giving an update on that. So right now if you look at it, chat GPT, especially in the chat market, the AI chat market, chat GPT has got the largest market share. I think it's close to 50%, I guess. But. It has dropped. Initially, it was around 60 or 70 last year. And this year, has just dropped down. Gemini is catching up. ⁓ Gemini's market share almost quadrupled in the last year or so, I guess. ⁓ If you look at the enterprise market, Cloud has got a bigger market space than any of their competitors. So personally, feel Cloud is good at writing code. ⁓ I've been extensively working with Cloud, chat, GPD. I haven't tried much of Gemini, but there are some tools where ⁓ most of the time I just try to shuffle around the models. For example, if you're using Cursor's IDE, so you pay one subscription, but you get to choose from a list of models. It has access to all models. has all the models, top models of Cloud. has DeepSeq, has OpenAI, has Gemini. So I get to choose from a list of models. So it both, and of it just comes to your convenience. You got to try out and see which works best, which are comfortable working with the response that it generates. So I think it's very hard to pick just one favorite right now.

Abhishek Pinninti: Right. So I see that Claude is also catching up in the market share than the biggest competitors over the last two to three months. Does that come in? mean, did that come into your notice?

Arun: Yeah, that's what I said. So Cloud has got the highest market share in the enterprise space. ⁓ They've been doing really good. From the very beginning, they have been doing really good with enterprises, especially with businesses. So they've been really good, especially with their recent updates. And a lot of people have been boycotting and... ⁓

Abhishek Pinninti: Thanks.

Arun: boycotting and moving from open AI to cloud recent political things that happened here in US.

Abhishek Pinninti: Yeah, so tell me your favorite AI tools that you have come across and maybe some fascinating tools that you wanted to share that you have come across while doing anything about your daily life.

Arun: I use lot of NIDN for automating my stuff, especially with my content creation. I've developed an automation system where once I decide, once I write the script, everything is automated. So writing caption, posting them, scheduling them, and just sharing my team and stuff. NIDN is a really good automation tool if you want to get started. And it has also options to... If you want host N8n on their website, you got to pay a premium for it, but you can also host it locally on your computer, which is free. Also you can automate any of your stuff for that. And I use ⁓ Google's notebook LM for studying. If I want to study some new concept and if it's a lot of pages and I don't have the time, I just get an understanding or a quick overview of what it is, I just drag and drop the book. ⁓ content material into Modpakellum and they just summarize it. They got amazing features like you can get an audio overview, it gives you video overview. There's also a beta feature where it generates an actual video based on that content. I didn't test it out yet but I recently saw there was an update.

Abhishek Pinninti: Yeah.

Arun: That's one thing. And recently, I also read lot of research papers. I recently also saw a beta version with Google Scholar where they introduced an AI feature for searching for research papers. It was pretty good. ⁓ And there is an interesting website called, there is an AI for that. So you go to the website, it has a list of all the tools that exist. So you can just search there if you want anything. You can just choose for what purpose. For example, you want, you're looking for tools for generating images, click on it, it lists all the available tools. If you want something for audio generation, you want something for coding, lists everything. It's called, there's an AI for that.

Abhishek Pinninti: There's an A for that.

Arun: There's an AA for that.com.

Abhishek Pinninti: I haven't thought about it.

Arun: So yeah, that's like a search engine for AI tools. Go there, can just search for any stuff that you want. It's right out there. can just.

Abhishek Pinninti: Thank you. I think they have as well there. Thanks for letting us know. I have tool list as well there. My tool.

Arun: So I didn't get them. Yeah, I think you can add your tool. I think there's a form that you can submit so that your tool gets out of there.

Abhishek Pinninti: Nice. I mean, this is something which I didn't know. Thanks for letting us know. And I think we had a pretty long discussion about a lot of things about A.I.A. and I feel this conversation has helped a lot of our community of founders. And thanks for your time, Arun. And definitely with things changing around so quickly. ⁓ I think we can sit down after six months and talk about many other things as well which has happened in the industry and Let's do another one again after six months and talk about this

Arun: Sure, sure Abhishek and thanks for your time and thanks for the opportunity. was nice talking to you.

Abhishek Pinninti: Thank you man, thank you. I'll see you. Have a great day.