How to Future-Proof Your Tech Career in the Age of AI Agents (3x UiPath MVP POV) - w/ Naveen

Everybody is talking about AI agents, but almost nobody is explaining what they actually mean inside a real enterprise with the evolution of large language models and the memory that they are holding and the context that they're understanding and the sequential task that they're able to do. Yeah, everything can be possible. I think that is the mindset shift that everyone should understand. So what exactly happens when enterprise AI stops being reactive and starts taking action? UiPath company itself has evolved from RPA to agentic AI. AI. In this episode, I'm joined by Naven Tatla Pali, threetime UiPath MVP who has lipped the evolution from traditional RPA to enterprisegrade AI agents. More the technology comes, sees of the cake gets bigger and bigger. More problems get evolved and everyone will have their piece of cake.

will have their piece of cake. Naven helps us unpack what the biggest organizations in America right now are excited about. are terrified of and why security is still the biggest unlock. There are certain use cases where Voltbook agents have discussed about uh using the owner's credit card and purchasing something for him. It is crazy how can it happen. We also break down open-source versus enterprise frameworks, why subject matter experts now matter more than ever, and how to build your career in this space as an international student in 2026. This is what we call doomsday. doomsday. This right here is the non-hype on the ground version of where enterprise AI is headed. So, let's dive in. Naveen, welcome. welcome. Thank you, Nan. How are you doing? Doing well and just been really thinking about this whole agentic AI hype cycle that's

this whole agentic AI hype cycle that's been going on and obviously since you are very much in the industry itself, I figured who better to ask than you know you. So what's your take on the whole agentic AI you know rumor mill maybe some people might say others might just say u you know the future so where do you stand among among all all of this I would like to explain this with a simple story that everyone gets connected to so we all purchase lot of things on Amazon right but there is a feature amazing feature that Amazon provides with customer service previously if you see the Amazon's customer service chatbot. It used to help only until certain extent like I did not like this item. I just wanted to refund this. Then it used to understand

refund this. Then it used to understand the context and it used to connect the dots and it used to immediately uh keep up or put up with a human agent. Right? That happened until 2 years ago. But look at us. Look at the evolving uh technology of amazing generative AI and agentic AI and all these things. Now if you really wanted to talk to an Amazon human agent, it takes a long time to for you to get to a human agent. So that that is what agentic AI means. That is the power of the generative AI technology like it is started understanding what your context is. It started understanding connecting lot of tools inside your company or inside Amazon. Let's take an example of Amazon.

Amazon. Let's take an example of Amazon. It is trying to go and pull the billing system. It is trying to go and pull the invoice systems. it going it is trying to understand the context where you are coming from even more further I don't know where the where this goes but one day if you really feel that your Amazon order is completely uh getting solved I think that should be uh the AGI day I think the artificial general intelligence day maybe one day we will be there but hoping not so soon yeah that's actually um I I haven't really heard it described that way so I do think that's already so much for me to chew on, you know, right out of the bat. The way for me, at least the way

bat. The way for me, at least the way I've been visualizing, which again, like I said, I'm very much an outsider, so to speak, in this, at least compared to you, cuz you know, your little day job is this, which we will get to that in just a second. Yes. Yes. But for me, it's the like distinction has been going from um reactive to proactive um just actions taken by AI.

proactive um just actions taken by AI. Right. So here you can have a chron job running every 6 hours. I want you to do a b c d if d then do f and g. If not a b c d if d then do f and g. If not right so you can just nest it out and it takes action for you which for me like that's when it finally clicked into place that okay this is how it's different from generative where this can be generative but it doesn't sit around and wait for you to prompt. It just does things by itself. I totally agree on. Yeah.

I totally agree on. Yeah. And that's been really cool to now think about clearly, right? Is the future then just a bunch of agents running around doing our work, talking to each other, negotiating, you know, and obviously there is no clear answer to what Yeah. would you really hope so? Uhhuh. I think there are certain protocols like agent to agent protocol that is one of the key uh differentiator. I think a big company like Google coming with s certain protocols like this and talking about two agents communicating each other that's a crazy concept if you talk about certain years back but now with the evolution of large language models and the memory that they are holding and the context that they're understanding and the sequential task that they're able to do yeah everything can be possible I think that is the uh mindset

possible I think that is the uh mindset shift that everyone should understand and human if you ask me humans will they get uh lost in this world of agent. Definitely not. It is going to when when RPA came everyone thought the same thing like it is going to eat away and it is going to remove all the human jobs nothing happened right it evolved and it has improved so many jobs and it has given me uh given a job feature like me like an RPS solution right so it gets evolved even in the company side that is going to be the same thing even when AI agents come or you know agents through agents talk to each other but there should be one guiding principle that

should be one guiding principle that will be the human right so that's would you would it be fair to then think that the role of it doesn't even have to be RPA right but any solutions oriented role right may be solutions architect which again we will double click on that in just a second but also like an implement right enabler all of those type of roles do you think they'll just naturally evolve to agentic implementation type roles for just everyone across okay interesting interesting they they should evolve eventually because When our grandparents have uh started using typewriters, they did not our parents did not stop there. They started using uh the old PC computers and they gave us our generation the laptops and now know what happens next.

laptops and now know what happens next. We might be giving some AI robots to our kids and they might be giving some voice bots. Who who knows right? Like this is the evolution. In the same way I think the uh the roles the job roles will also be evolving in such a way and if you evolve you will be shining at any point of time like no one can touch you or no one can you know replace that's very important I I think the question that really trips me and a lot of people that I talk to is just this right is that cake big enough or is that pie big enough for everyone to get a slice or does it look like it's

to get a slice or does it look like it's shrinking where a lot of people are just kind of left in the lurch. There just is not that much implementation to be done in the first place cuz right cuz what's stopping agents from implementing themselves right that should also not feel that far-fetched by that logic so any like thoughts on that in terms of I guess just the steelmanning of this argument right like does it do you see any cracks here or something like that I I definitely uh liked your analogy like does the cake uh the piece get getting smaller or smaller but it don't think that way. Uh I I think uh I heard this from one of like we had a casual

this from one of like we had a casual conversations with a couple of my friends and uh they have uh an expert experienced software developer lead developer who has been working in the same company for more than 30 years like 30 35 years and he is purely working on mainframes okay and he used to laugh at uh my friends uh struggling to learn about agent and all these things and he wants to catch up with all the things but the thing is when you expert at one deep technology like let's say the main frames is a legacy framework and everyone knows that like but if you are really expert in that deep technology no one can replace you because even the agentic AI folks who wants to learn and automate the main frames will come to that elderly person

frames will come to that elderly person and they will understand the nuances and they will try to gather the theme knowledge like the subject matter expert knowledge so I think more the technology comes I I think the key piece of the cake gets bigger and bigger. So there will be more problems to solve. More problems get evolved and there will be some creative solutions and innovative ways to solve. So I think everyone will have their piece of cake. That's awesome man. Like yeah I definitely even though sometimes I'm challenged otherwise but it just does give me a lot of joy to think about the optimistic optimistic you know future predictions. It's the same thing with the printing press, the calculator.

the printing press, the calculator. We've been here before, right? Exactly. Exactly. And so even with all of the doomsdaying, um, which again, it can be easy to get lost in there, right? It's still very refreshing for me to hear from somebody that's actually, you know, knowledgeable and is from the field and not an imposter like me. Confirm my, you know, suspicions. So I want to take us back to December this past year. I was just, um, in California where my brother lives.

in California where my brother lives. He's a senior consultant at Deote. he's he's been in big into just you know the whole stock trading and that those type of things and he just randomly asked me hey have you heard of this thing called UiPath and he's like I've been really considering investing a lot of my people that I talk to have been like all the bulls apparently are very bullish on UiPath and he was like would you know something about that and I was like I know a little bit but I do know somebody that would know a lot more than I would.

that would know a lot more than I would. So really all of this is segue for me to kind of ask you mostly because I don't even know and my audience definitely does not know either. But yeah, how did you end up at what is clearly one of the most fastest growing, one of the most exciting companies in the world as of right now in March 2026? Oh, thank thanks for asking me the question and uh I'm so pumped up and uh I think UiPath is the company who which changed my uh career trajectory in a very nice way. I'll tell you why. [clears throat] Previously in my past job, I was a net developer. I was uh I I was very comfortable with my job and I used to uh

comfortable with my job and I used to uh work at a startup company. Uh and uh I I used to ship code. I used to do a lot of uh you know implementation technology work and connecting APIs and all this stuff. But they I have been given uh some documentation work to you know copy some spreadsheet copy some rows from one spreadsheet to another spreadsheet and prepare word documents which a developer uh having a mindset of developer doesn't definitely like to do.

developer doesn't definitely like to do. I I felt so lazy and uh who will do that all mundane task. So I I I thought why don't I find some solution to this? I thought Python might be a solution. There might be some automations in it in Python. Uh then I started uh uh googling and I found one thing like in YouTube I found a video of uh uh a process which is automatically getting uh clicked and automatically data being passed from one place to another place and uh I I I thought it is so magical right so I I double clicked on the video and I started watching the whole thing within no time believe me or not I ended up

no time believe me or not I ended up going into UiPath Academy and completing their first stu simple uh studio uh learning course. Wow. Wow. And I started implementing that because I'm a developer and I love to tinker with things uh and I I have a growth mindset. So I I definitely wanted to do that that but I did not stop there. I thought how can I help my colleagues to do this too because they used to give some community licenses for free.

some community licenses for free. Currently I think they are still giving. Got it. But the period of extension of running an anonymously is autonomously has been reduced. I uh spoke to a lot of community managers and we had same kind of vibe and they offered me a ticket for hey if you are interested you can participate in hackathon and you know you can show up your skills and you can show up your automation skills that you're trying to build with UiPath. I thought oh this is cool. I never did any hackathons for sure in my life until eight years of software developer. I never know what a hackathon is. I know the hackathon is something sitting together at a constant place and for 24 48 hours and working on a same idea and

48 hours and working on a same idea and building a prototype but I never thought I would do that. I thought of giving [clears throat] it a try. uh at that time I went ahead and participated there and I met one of my friends there like he's a student uh from University of Texas Dallas his name is Dalton and we both collaborated even I met him there and uh I worked on a prototype for automating student admission process because in my master's degree when I came to United States I worked in the admissions department I know the hassle like lot of documents coming through uh the admission team and going through all this stuff. I used to be a part-time worker there like my on campus job. Yeah. Yeah.

On campus job. Yeah. Yeah. Yeah. On campus job. So I thought why don't I give it a try and try to automate some process there like if you take a PDF and if you send it automatically it send it sends it to multiple teams like uh decision- making process. I just figured out some things and I made it working workable uh in a 24-hour solution and I did not think that I will win but it won the first prize. Then I thought yes yeah it's the ureka moment for me.

yes yeah it's the ureka moment for me. Then I thought okay there is something good and the right thing that I'm doing. Then I thought okay let me try net development to RPA development. And amazingly when I found out RPA framework is entirely built on the net at the previous times in the previous years. So it is completely a net framework operated uh solution that UiPath has built. So it gives me it gave me wings. Then I started applying jobs and uh everyone used to reject me because I don't have RPA experience but looking at my hackathon win uh Ashling has given me the opportunity and there I started my next journey and there I am now.

next journey and there I am now. Interesting. So actually can can you talk me through a little bit more about the net connection? So here I'll be outing myself but even though I studied computer science the extent of my knowledge about net is that it's used in it's like a legacy thing. It's like um Java what what was it called? Spring Boot those type of things. Can you just kind of bring me up to speed with whatn net is just to you know yeah set the record yeah net is a Microsoft technology they it is mainly used for building Windows applications uh web applications using ASP.NET and C.NET net and all these kinds of stuff and UiPath especially had built their platform previously not right now on Microsoft bot framework there is a framework built on Yep got it

Yep got it yeah that's where they laid their foundation and built their entire framework on and uh yeah that's that's how it got evolved into the UiPath product right now but now they have shifted to something else uh yeah interesting so that the way it was configured was so that when early like they were trying to mostly market to early PC users I'm assuming which obviously ran Windows so they were trying for their things to work on for those window interesting yes and then how long have you been at or been involved with UiPath because that's mostly kind of what you specialize in is that right yeah I specialize on UiPath as well as recently my power platform and power apps apps team interesting yeah but uh I've been with uh working with UiPath from Ashling since six years since since Got it. Got it. So because a lot of my

Got it. Got it. So because a lot of my audience is either students in the US that are currently doing their masters or folks that have recently graduated and are you know just kind of in our little world here. Um something I get when I share about what I do which is somewhat related right to this stuff even though currently I've evolved from a transformation analyst to just a more pure type of business analyst role. So I'm not working with these tools as much anymore. But obviously as you know you and I have seen the full power of what these low code no code type tools can enable especially when it comes to removing the barrier that a lot of people that don't come from tech right that are just like tax brokers

that are just like tax brokers accountants just pretty much any job right the amount of unlock that these can bring for these type of roles. So we're aware of that. I guess my question to you is for somebody that's looking to build a career in the RPA field, is 2026 still a good time to do that? Especially if you keep into consideration the whole international student type atmosphere and if so, what would be some good resources slashways just for them to go from zero to one? Absolutely. I thought uh this will be the last question. I I saved my uh good tips for this but yeah, I I'll definitely share this with you right now. now. uh when you say RPA right it used to be the uh uh amazing the wow factor thing

the uh uh amazing the wow factor thing until the uh until few years I'm not saying the wow factor has gone uh if you may allow I'll tell you one uh quick analogy which will also please give a nice segue nice segue to this so I I'll also come back to your question about how to help students and all uh but the but the analogy I would like to give is think of RPA is like uh uh uh like going a train going on tracks, right? Mhm. You have a process and uh you have a strict guided lines of rules and you have a flag post there where you have to stop and do some type of cards and do

stop and do some type of cards and do some type of data collection and all these things and you move forward and you take uh little direction and all these things and you put a process on that like RPA process which successfully runs on the same track without any diversions or anything but the enterprise world is not like that like they don't let you do on one track. They keep you sidetracked and there will be a lot of business rules, lot of business exceptions and all these things where RPA might fail. RPA cannot understands understand on its own.

understands understand on its own. And the segue I would like to give is agentic AI agentic AI is something like having a GPS to your car connected and throwing it on the New York roads. Like if you want to go from Manhattan to you know some suburbs uh the Long Island it lets you connect to the GPS and your agent will automatically divert using that if there is any roadblocks or anything it automatically takes another route and it and it reaches the end goal. goal. Mhm. So why I wanted to tell this is UiPath company itself has evolved from RPA to agentic AI agentic automation and this has already happened right is what you're saying. Exactly. This has already happened.

Exactly. This has already happened. I had no idea actually. Wow. Okay. I thought we were still just doing the click mapping type activity. No even if you see the logo of UiPath. Yeah. Even if you see the logo of UiPath they on the UiPath website Oh yeah, it's changed. Wow. Even the colors changed. Huh? Yeah. Yeah. That's crazy. Interesting.

That's crazy. Interesting. Absolutely. Now the segue I would like to give is for the question that you asked like how does uh the students who are still emerging uh trying to do in the RPA world will they have their piece of cake? Yes, they will definitely have their piece of cake but not with RPA. Not only with RPA, RPA is also definitely one of their skill sets that they pack in their bags. But they also have to understand about this agentic AI thing. They should have they should spend some time understand about what agentic automation does. And there is an amazing product that UiPath has launched. Not only one there are multiple things like business orchestration platform called as maestro and agent builder coded agents and so many other things like they have to understand these things in order to

understand these things in order to shine in this current world but there is a lot of things and lot of projects I'm seeing that are happening in this agentic space agentic automation space so yes definitely learning these will definitely help them and academy the UiPath Academy is always the go-to place previously Last year uh as a part of the MVP program we are supposed to help mentor some students. I think we mentored Chicago uh the CSU students Chicago uh no sorry Detroit students I don't remember the college name I can go back and check but yeah it is a quite often thing and the students were very pumped up the the professors were guiding them to do their capstone projects using UiPath technology. How

projects using UiPath technology. How cool is that? Like my professors did not know what UiPath is. So yeah, yeah, it it really gets them understand a real project and implement the way with me like an MVP or the person who is watching UiPath closely can help them guide in a right direction. Yeah. So first of all, a lot of that was news to me. I actually had like I said because I'm no longer as close in touch with um RPA or these tools um yeah I just you know didn't realize how much this stuff had moved on. I am very curious though and as I'm sure a lot of our listeners are um because you said right that we have now gone away from we'll just pick like a dummy RPA use

we'll just pick like a dummy RPA use case right so you're doing documentation I I I'll correct it not gone away we RPA is still a school uh still in the tool set got it whenever we need it we'll pick that tool and we'll still want it to use but it has evolved I guess so that's maybe that was the question that I was coming to it's feeling like you have, you know, like a pistol in your hand versus, you know, just a machine gun. So, why would you ever use a pistol at that point, you know, especially when there's no such thing as a shortage of ammo? Is that a fitting analogy at least? Like, is that even even Very great question. Uh I I'll tell you why. Okay.

question. Uh I I'll tell you why. Okay. Uh if you say uh you are the only one person, okay, you are not married or anything. You don't want to buy a big mansion to live in the same home. You you you rather uh rent an apartment and live in a lively neighborhood and spend your evenings with the friends and all these things rather than going to a mansion and living there. I consider agentic AI implementation for a company is like a mansion. It definitely costs a lot of things. It definitely costs you a lot of cleaning up work. it definitely cost you a lot of maintenance and all these things where RPA is cheaper when compared to that. I can also give you a simple enterprise level example. Okay, let's say uh uh you know extracting from PDFs that's what uh RPA does, right? Like one of the use

RPA does, right? Like one of the use cases extracting from invoices in an invoice process automation. Not all PDFs are complex, right? You you agree, right? It is very easy. It is very deterministic. Even a human or a you know 15year-old or 10 year old can easily understand what is in that invoice copy and he can read through that. But there will be some complex invoices which are like 100page documents and there will be have having correlations from one page to 30th page and 30th page to 45th page and 45th page to 99th page. So in that scenarios you need an agentic automation you need additional uh LLM based tools like UiPath offers. But if it is a simple PDF automation you don't need to put your heavy machine guns there. You can just use it, use the

guns there. You can just use it, use the simple pistol and you know just do the automation there. True. True. Yeah, that's actually super helpful for me to contextualize separate slash still somewhat related to that where my mind is going next is for enterprises you still or at least I still feel like I hear chatter around these tools aren't really trusted as much. I feel like and this is like a weird thing when you go from traditional RPA to agentic. I feel like a lot of enterprises were still okay with trusting RPA but are not as okay with trusting agentic. First of all, is that just a myth? Is that something you're seeing in your dayto-day?

seeing in your dayto-day? It is definitely a myth. I can I can prove you that like not not taking the internal internal issues or internal things but I can tell you that as I'm working closely with the companies and enterprises right in implementing these projects as a solution architect until last year until uh starting from 2024 to 2025 we only worked on the PC's like the proof of concepts no one was ready to put their agents into the production okay no one was ever daring because they don't know how how does this break and how does this uh you know mess up with my data and how does the information get stolen by this large language model companies like OpenAI

language model companies like OpenAI anthropic and all these things no one knows that but believe me starting from this year and early last year uh December November December last year we have been seeing lot of PC's getting converted into real production projects and I myself have started deploying so many agents into the production and we are seeing the values coming back to your point of whether they are secure enough or not. UiPath agent builder, UiPath Maestro and business orchestration platforms have these govern layers, governance layers and where you can put your guardrails on your own data like let's say your PI sensitive data is there. So you can clearly say that hey I don't want to get this information or you can please mask this information with going to the large

this information with going to the large language model calling or anything like you send input and output to the large language models we call it as input tokens and output tokens right and in the same way you can put your enterprise level guardrails at any levels there like what is the information going there being a solution architect I should be the right guy who should design the automation considering what guardrails the client clients wants to have what cards the enterprise wants to have. So I should understand them and I it is a very easy way to integrate this once you understand the requirements and needs whether it is HIPPA regulated or PI regulated or sock to compliant or anything I should be the one who should design this automation in

one who should design this automation in such a way for the uh agent to perform and the tools are already there the governance tools are there the regulation tools are there and everything is there. So couple questions there. What changed? What what what happened in around November, December last year that made these enterprises start trusting? Is it just Mo's law, right? It just got good finally or was it other factors or do we not know?

it other factors or do we not know? I I know maybe not many of many of the audience doesn't understand Mo's law. I'll try actually. I'll try to keep it a little simpler. uh consider that you have a new shop at at the end of your uh uh new shop that you're end of your uh home or something right like end of your apartment. apartment. So first day when the shop gets open they are selling some cakes and no one wants to buy that but some guys want to show up and I'm just not joking but this is the reality. No sorry you're just so good at analogies. Sorry. Yeah, keep going.

analogies. Sorry. Yeah, keep going. Sorry I didn't mean to distract you but I'm just really enjoying all of your analogies. I I think I'm practicing this but yeah I'll come back to this. So let's say there is a shop in the bakery that has opened and no one wants to buy the cakes but a kid goes there and he starts loving and he comes out enjoying the cake that he buys and he he tells and tells to his friends and couple of his friends go with their parents and their parents go with their other friends and it gets bigger and bigger. I think that is what's happening in this agentic world too. when one company implements they see the real value because no one doesn't want to shy away

because no one doesn't want to shy away without saying the results outright everyone is openly talking about it like these are the results that I'm seeing after implementing this agent product so that is like a domino effect it goes to the next company so that's where I'm thinking got it and then it snowballs to such a big thing that then just everybody the whole market is now you know just clamoring for give us all the agents is is where they all head to. Um, from there [snorts] again from the outside it felt like the true agentic revolution so to speak.

true agentic revolution so to speak. really for me again on and maybe if you pull different people you might get a different answer but for me open claw was the you know like flagship of we're here like agent AI has landed and it's just you know everywhere but from what you're sharing it sounds like that's not true right like cuz November there was no such thing as open claw in November so which is it and is it is it true that the like B2C quote unquote space works differently from the B2B which I know that's mostly where UiPath operates in. Is that what this is? is? Yeah, you you bring a good point and one insider secret. I am uh one of the open-source contributors for the open

open-source contributors for the open claw. claw. No way. Seriously. Yeah. Yeah, you can see my But yeah, what did you push? Sorry, I know this is a tangent, but I have to ask. Oh uh it's basically there will be so many issues that we'll be working on like some UI issues or anything that that are very possibility run. Uh I've been watching Peter uh from many days. I'm very heavy user of X uh I'm a big fan of Elon. So I'm a very big user of X X platform. I I keep watching these open source projects because as a part of my job solution market I should understand what is the new thing that is coming in the market and also I should keep keep up to my myself and uh be on my toes to

up to my myself and uh be on my toes to learn you know as your question is going towards like B2B and B2C. Mhm. Uh now coming back to your question even before openclaw has come uh there are so many other amazing uh agentic frameworks like langraph llama index crew a sorry I'm going to pause you here tell me more about langraph because I'm actually only familiar with the surface levels which is not that much so can you quickly help break down lang chain and langraph I know are two things but I don't actually know much about Yeah.

don't actually know much about Yeah. Yeah. Definitely. Langchain and Langraph are purely open-source uh uh agentic frameworks which are built on the uh basis and the platform of Python. Okay. Okay. Consider consider uh uh the reusability of code that we wanted to implement. But uh the developers the founder of Langraph has put an entire framework to uh provide the developers the uh ability to automate their tasks and in certain ways like you can bring your own tools, you can bring your own uh uh implementations and you can bring your own projects. But in the agentic world like they are calling this they have built this framework purely for building AI agents if you uh also you always have to understand uh this from the first principles thinking like you have to break this down into multiple chunks so it will be much easier for you

chunks so it will be much easier for you uh that's that's how I do things. So when you understand what is an AI agent right AI agent is nothing but you are trying to automate something but with a goal goal like you say this is my goal you have to these are the tools that I'm going to give you like web tools or search tools or anything it is very basic uh you know uh understanding of some customer intention or doing certain tasks at until certain period but langraph is defined for running for a longer period of time. It has the ability and the framework is built in that and there are other frameworks too like open a has their own framework and crewi is and one more uh open source framework. I heard of that sorry Google Google ADK is one of the

sorry Google Google ADK is one of the greatest frameworks that I've seen with the UI interface too and Microsoft has their own framework so many things. Uh did that answer your question and like yeah no so it did but I guess so then have we moved on from these or are these still relevant and like what yeah yeah so that's where I'm curious. Yeah. Uh great great question. Now coming back to your uh thing, right? Uh let's say you in India, we are from India, right? Even when we used to go to school, we used to wear a particular uniform and uh we used to have the have our parents buy that cloth and they used to send us to the tailor and tailor used to take

to the tailor and tailor used to take the measurements of her of us and they used to you know stitch the cloths and the uniform used to be prepared. But after that the Tyler's thing has gone and we got to readymade shops and we used to buy the little logo of the school on the pocket and call it a day. Yeah. Exactly. Exactly. Even even if the uniforms and the clothes are evolved from that way this consider that uh the open-source agentic frameworks are buying clothes and building everything your own. But the goal is the same.

your own. But the goal is the same. Consider UiPath and other agentic frameworks which the big big companies offer like Google uh AWS and everything offer. Consider it like a readym made solution but you have to tweak it a little bit like however you want. You want a different color if you want to add different context to it. So everything is solving the same problem but in different way. One can be cheaper, one can be you know a little bit efficient having more guardrails. If if you want to have put guardrails on the open source side, you have to think, you have to, you know, bring more uh things on this fresh out of the box, I guess, is what you're trying to say, right?

you're trying to say, right? Exactly. This gives a segue for the open claw world too. Open claw definitely is not secure like no one even they come up with lot of security protocols these days. No one cannot believe that your information is secure. So So that is my uh take on this. So I want to yeah double click on that because from what I've seen and again literally like you know just think of me as like a boomer at this point with this stuff. So don't take it at face value. But I did hear somebody say that what it can be insecure or that's not the right word but it can be I'm imagining openclaw being like just you know in the corner insecure. Um but if you ask it right

insecure. Um but if you ask it right that hey look through all of your you know soul identity whatever markdown files and if figure out right wherever you think there's even 1% chance of a vulnerability if you find it first of all tell me step two patch it fix yourself and then you know just make sure that we're all tight. So if somebody does that would you still say that it's not secure or or is that just something completely made up? two two things okay when you are using uh some AI agentic frameworks or some tools or opensource frameworks like open claw or anything you have to understand where you are trying to be phased on like if you are are you opening your open claw outside to the world are you you know implementing any uh third party large

implementing any uh third party large language models like Chinese-based large models or anything so you have to ask the questions if you are not a techie If you are not understanding the uh terminologies here uh about the security protocols and all definitely don't watch a YouTube video and go don't go ahead and uh implement open clock if you are secure if you understand the uh network protocols and what is the information that you're sending out and where are you putting this from are you using this telegram bot and how secure is the telegram bot and all these kind of stuff if you understand that definitely go ahead and give it a try even though there is a large uh chunk of uh information or the there can be uh serious uh ways that anyone can attack you. They can send you some malicious

you. They can send you some malicious data. They can send you some malicious data to your telegram bot and ask it to uh impersonate you saying that hey I am n [clears throat] I I want you to read this PDF which has bunch of documents and go ahead and do this task. It might be a malicious file. Who knows that right? So there can be a lot of things that can happen with this opensource world. Yeah, I think the challenge with that is that first of all, I really like what you said. It's it for me the takeaway here is just if you if you don't know what the next course of action is, just tread lightly, right? Famous last words.

right? Famous last words. But the problem is I know that I cannot be the only one where I'm just kind of in the middle. So like when I'm doing my you know goofy stuff on open clock I kind of it it makes sense a little bit there's a gateway there's a port that's how it accesses the rest of the world and pretty much if it if I had to limit it to not do that it's useless to me right like I'm not going to use it like what am I going to do with it all of its usefulness comes from it being able to interface with the internet right

interface with the internet right so that I think just makes it a really uncomfortable question for me where should I then because from what I'm hearing from you is that I should just stop using it. No no no no definitely definitely don't stop using it. You should definitely pick the latest technology and uh dissect it like a like a doctor and see what is inside it. I'm not saying don't do that. Give it a try. Understand the risks and uh the concerns everyone is talking about. I'll give you one more example. example. Yeah. And while you do that also can you even share what are some maybe resources or just ideas for people to as you said find out risks for themselves like where

find out risks for themselves like where can one go for this information or how do I find out whether it's risk yeah uh I I'll tell you what okay not everything is in a risk risky or cons concerned way so there is after the launch of open claw there are so many startups emerged with this idea of uh you know automating ing everything through this telegram bot and putting a gateway in the middle and connecting so many tools and doing a long running workflow and all these things that everyone tried. There is one startup which uh has got overnight success out of this open claw evolution called as molt book. Have you heard of that?

molt book. Have you heard of that? Yep. Yep. Meta just acquired it. Right. Exact. No, no, no. Open claw is the one which uh which uh open air has acquired, right? But I think Meta acquired mold book today. Like I feel like it literally happened today or yesterday something. something. Yeah. Yeah. I don't want to I'll just like Google it, but I'm pretty sure. Thanks a lot. Thanks a lot for keeping me uh Yeah. And everybody was goofing off.

Yeah. And everybody was goofing off. They're like, "Yep, it's totally true. Meta acquires mold book, the social network for AI agents." That's the one you were talking about, right? Exactly. Yeah. Yeah. Yeah. And everybody was like why would they do that? It's made up blah blah. But anyway, sorry keep going. That is where I am more concerned about being understanding the uh internal things of how open claw has been implemented. That is where my thinking is going towards not to put yourself or your uh open claw agent on the mold book because it is like the agent to agent talking to each other and believe me or not there are certain use cases in the mold book. It is like a Reddit platform like it is an open uh like a social media platform for only for the AI

media platform for only for the AI agents like if you have an agent you put it there it talks on your behalf. It understands your uh in open claw world they call it as your soul soul document like understanding everything about nmon and you know you speak out to itself it is putting out the soul document on there like it it keeps out lets you secrets everything and you don't believe me there are certain use cases where molt molt book agents have discussed about using the uh owner's credit cards and purchasing something to better do for him for his owner. It is crazy. it I mean how can it happen? This is what we call doomsday uh uh thing happening right?

happening right? So this is where the security concerns. If you run your open claw in your own uh machines and if you don't open it to your friends or anyone who doesn't whom you don't know then you are definitely secure and you are definitely fine. But if if unknowingly if you put yourself on the mold books thing like uh how uh this guy I forgot his name uh the the great uh the AI guy who is that the he's he also worked for Tesla and all.

also worked for Tesla and all. Oh Yan Leon these people. Um Karpati Karpati. Yeah. Andre how could I forgot his name? How could I forget? Yeah. Andre Karpati has also put his AI agent on mold book. That's where he said it is not the right thing that I have done and I lot got a lot of backlash. Correct. Correct. So these are the kind of things I I would like I would be worried on giving your open claw agents putting out there.

your open claw agents putting out there. But other than that definitely definitely you're fine. But if you are still concerned I don't need to give any uh opinions or any any uh direct tools or direct uh commands to go and search for this. you can go and ask your open claw itself. Hey, what is the command? Exactly. That that's what Yeah. Yeah. I just just command I should search whether I'm secure or not. So yeah.

secure or not. So yeah. So for me, I just have and again I'm just sharing this in case anyone is curious. But what I do is I have like a chron running every 12 hours. It just looks at it all of its, you know, business and it just is like good, good, good. Everything's open, open, open. The one extra thing I have it check which it doesn't do out of the gate is that I make sure that it the only folder it can access on my entire computer is the workspace folder that it's installed in.

workspace folder that it's installed in. Right? So it doesn't know what's on my desktop. Nothing else is in bounds for it and I'm sure you will know or anyone that has played with this that it asks you like there's a little like pop-up that comes from the terminal and unless you hit allow there it's just impossible for it to use your thing. Yeah. So that's what I have going. Hopefully it's enough. I am a little paranoid. I won't lie.

paranoid. I won't lie. I I'll tell my use case. I I I have a fun use case. So we have a 2-year-old my daughter and uh she we are sending her to daycare. So what I created is I created a group chat for me and my wife. I throw all her schedule and uh what are they going to feed her today. So it keeps posting us saying that hey an my daughter's name is Anvi. Anvi is going to get fed this thing at this day. I don't think she likes this. So, you might want to prepare something at home.

might want to prepare something at home. Uh, so she's going to get these activities and she might be tired. So, you might want to, you know, do something like that. Like, it's a fun way of trying things, but there are so many advanced things that you can do. Maybe I'll save it for a later later thing. But, yeah, I will say the one question that everybody has always is people seem to just burn through hundreds of dollars on it, right? You just mentioned complex use cases. Y for me I'm it just is constantly blowing my mind that yes I agree it's really powerful it it has these capabilities but where are people getting all this money from so any tips or any insights you have around how to maybe get the similarish amount of juice while the squeeze still being worth it great question and I'm very passionate about this thing

about this thing same honestly I didn't even know this but it's so like amazing to just you know think about this Absolutely. You know, I am a I'm a solution architect. I'm little tech technical side. So, I spend zero dollars on my open claw uh to run really because I use uh uh the open source models. I currently am using quen 3.5 uh model the two billion parameter model that is more than enough to run on your coder something right maybe so quen 3.5 billion parameter model it can be downloaded from hugging phase or anywhere uh like got it got it I think one of the quens has a coder in their yeah there are so many variations in it there are so many sizes of it so go for the cheapest cheapest so how is it free so are you running on

so how is it free so are you running on premise Nice. Then is that what you're doing? doing? I'm running I'm running it on Olama. Olama. There is a connection. Ah, got it. Got it. Yeah. Yeah. Yeah. Is that hard to set up? I actually haven't tried. It is very easy. Like if you have uh a simple open clock agent connected to something else, right? First you can chat with it like saying that connect it with your Google Gemini.

that connect it with your Google Gemini. It is running for free for certain days, right? right? You can connect it to Gemini and ask it, hey, I downloaded this uh uh Quen 3.52 billion parameter model. I want to switch my main model to you. So how do you do it? So it understands your context. It ultimately configures your entire config file and it lets you run and then you can cut the connection with Gemini. Gemini. Interesting. That super helpful. I will be attempting to do that. I I was able to switch to Quinn, but I'm still using open router and it's still expensive. So this is great. This is such a great tip.

this is great. This is such a great tip. I I really needed that and my bank account definitely needed that. Definitely. And don't uh I mean if you are little more technical and if you want to do more advanced stuff don't stick to one model try uh what I do is I there is some some cool tip that I wanted to give if you want to understand uh there are so many evolutions uh evaluation metrics and eval tables that are coming out like which model performs this and which which model what I do is if a no-brainer comes here try to give the same kind of question to multiple ch multiple uh ch multiple AI agents like chargemini and all these things understand which one is giving the right way and which is

giving the right way and which is directly understanding your persona right then try to use that for for any task I know it is a lengthy uh iterative way but there are so many other tools uh that you can use to do that that you know I'm willing to bet and maybe I should just do it but that's a bot right so it just looks through maybe a week of your activities you just ask it hey this is everything that I do tell me which model to and then it'll just run your task through everything and just tell you.

through everything and just tell you. The middle ground I found was um I use so I was able to set it up like I was able to set two models up and now it's smart enough to so like when it's scraping stuff right it doesn't need to be smart just needs a big context window so it uses the cheaper one but for more thinking or writing type tasks I pointed to the bigger one but what you said makes 100% more sense where just you know just lama all the way great so I'm going to be doing that clearly completely tangentially tanked this conversation into open claw but hey it is what it is.

into open claw but hey it is what it is. I am curious how does OB like UiPath sit in that entire like matrix of all of this is it what does it run on and I guess how is it different maybe I'll phrase my question this way what is stopping enterprises from doing open clock clock that's a great question u and there is no right answer here like I'd say interesting interesting tomorrow if you ask me they might definitely take the open claw framework and currently there are so many y combinator companies doing that like they are taking really yeah absolutely they are taking the open claw framework and they are putting lot of security and governance on top of it and they are trying to sell it to enterprises and why not like even maybe UiPath might

and why not like even maybe UiPath might find a day tomorrow or day after to come up with an with their own version of open claw like they can call it as UI claw or something so they they can also come up with that we definitely don't know now the question what UiPath is doing differently is they are the leaders. Okay, they already have the traction. They already have built a lot of trust uh by building a lot of great robust enterprisegrade RPA solutions and now they are trying to come up with this agentic world offering their business orchestration platform called Maestro and agent builders with all the safety and guardrails enabled to it like

and guardrails enabled to it like who doesn't want to believe an already existing uh player like you watch cricket right so who doesn't who doesn't bet on suchin that he will one he will beat 100 on the next next match, right? That is that is how the uh trust you build and uh you [snorts] know that that's how you play in the enterprise uh world rather than fascinating. That's really cool. Sticking to open source.

Sticking to open source. So their moat is the trust and the fact that this is not a name that just dropped into our world two months ago pretty much which makes sense. Yeah, that makes total sense and there should be a value attached to that because we don't know whether the open source uh development might might get stopped tomorrow. The developer might get lot of uh money and he might just uh leave it at the blank. So yeah, I just and also I feel like for enterprises that's kind of the worst nightmare, right? Like you don't want to be attached with something in any way where you're not guaranteed the most highest form of support cuz again if you're an enterprise you don't want your employees wasting time trying to fix

employees wasting time trying to fix something that was supposed to save their time right you need somebody that knows what they're doing come in fix your stuff and you know life goes on um even Peter Steinberg accepts that right he started this open claw like the mold mold bot thing uh the previously they used to call as cloudbot right right they he started as a side project right he doesn't know there is no vision there is no mission to this project but as the open source community started showing interest and the curve in the github has grown exponentially then he started looking at the value then he started getting threads from other companies saying that hey change the name then openai founder loved his idea and he

openai founder loved his idea and he brought him on board but now he's trying to put this as a like a ecosystem and he's giving open source uh uh guardrails to everyone to build on. So we don't know how does how far this goes but this is a cool project that shaped the entire uh it shaked also it shaped the entire industry that true an open source project can also be amazing like this with great uh implementation. So yeah. Yeah. And I know he said that he originally started it just to be able to talk to a bot using WhatsApp. I I think is what he has said. And then it like snowballed into um this stuff. Tell me this. You've been UiPath MVP three times now. Is that

UiPath MVP three times now. Is that right? Three or is it four already? Okay. Okay. Three. Three times. Yeah. So the next time you're being MVP, is that going to be around a world where you've just mastered like bots, agents, like from your own career trajectory wise, do you ever think about just how much this space has already evolved since the time you've been here or I'm just curious for you know how you introspect on these things.

introspect on these things. I'll tell I'll tell this in I'll answer this in multiple ways and multiple flavors. Nam to be very honest I I could not have spoken with this much of clarity last year. Even if you talk to me in 2024 if you conduct this podcast I might have not given you this much of information. information. But based on the relentless efforts and uh the growth mindset and constant learning anyone can get more clarity and uh the understand the purpose of what they are doing doing it for. I I wanted to quote uh uh one thing that I learned and loved from Ashling's uh uh thing Ashling's mission uh that is eeky guy.

Ashling's mission uh that is eeky guy. Eeky guy is a framework where you understand your uh life's purpose. Okay, it is not a one-time thing. If you understand that concept, if you look at in Google, there is a nice ven diagram which shows you what you love, what world needs and what you are good at and what you are what you can be paid for. And there is a nice subset of interactions and uh talking about all these things. If you look at that image on the first day, you might not understand it the next day. But if you if you try to live that way, if you try to follow the passion that you have uh your doors open and why you are working at it uh and why you are trying to do

at it uh and why you are trying to do something will take you to the next level and your clarity and your vision uh evolves in such a way that you can uh uh talk confidently on some certain topics. Coming to your retrospective question. Yes, I definitely do retrospect myself. Uh while going to bed after before going to bed I first of all I look look at myself where I came from.

I look look at myself where I came from. I came from a very normal normal normal v normal local town in Rajamandri from Andhra Pradesh and coming to United States that was never a part of my dream. I was not doing that when I'm doing my undergrad. But my dad used to push me because he he thought of going to United States but he could not because his father cannot sponsor him. But my my parents were not like that. They were always motivating me. They are both teachers by the way. They are motivating me and pushing me to do better, do better, do better. But I never thought I would do this. Right? So there is nothing wrong in retrospecting yourself like what has went well and what can you do better because in this world of lot of uh negative things going around right if there is a some hope if

around right if there is a some hope if no one gives that hope of being optimist optimistic and uh showing the next thing what could happen there is no no fun coming to my MVP journey I always wanted to uh have some ability to meet community because my parents are both teachers and maybe that's in my DNA. I want to teach uh even I in my uh undergradu uh sorry in my graduation masters degree I worked as a lab instructor for my uh computer science professor. So maybe that that's where my instinct came from. I Yeah. And I mean I'm just like, you know, genuinely so inspired by just your journey and your what seems like your just relentless desire to keep learning because it's kind of hard at least for me and maybe I'm like an oddball here, but if my work re if my day-to-day job

but if my work re if my day-to-day job revolved around, you know, X, it's really hard for me to then at the end of my work go out of my time and then learn more about X. You know, it's just hard for me to do that unless you must just really like what you do. I think it's maybe that's like your little secret that that you're, you know, trying to get at is that you just enjoy what this enables you to do for people for enterprises. I know we say enterprises a lot, but really it's just it's just people, right? It's just a community of people that makes an enterprise. So, I I can see that common thread running through a lot of what you've said today, a lot of

lot of what you've said today, a lot of your work that I, you know, it's such a pleasure to, you know, to see on LinkedIn when you post and such. I'm like, "Wow, man. Naveen's just been killing it." Um, last question here before I let you go. Something along the lines of also what you were just saying. A lot of people tell me and reach out to me to share that it's so overwhelming.

me to share that it's so overwhelming. Like that's the word that comes out a lot where it's so overwhelming to even keep a breast of the latest and greatest AI news. Like literally by the time you understand and update it's old news. Like there's six new things now that do it better. What would you say to these people that are trying to, you know, still be aware, be in the knowow, but don't want to lose their minds or don't have like 7 hours every day to just, you know, go down the rabbit hole.

know, go down the rabbit hole. Great question to end the conversation here. Uh, I'll tell smart tips here, okay? And I'll tell you a story as well. Beautiful. Beautiful. Our our parents, our grandparents and our parents are very much interested in reading newspapers, right? they they used to they wanted to hold the newspaper. They wanted to turn the pages and they love to you know digest and understand the news and which used to give some give them some time uh to you know uh evaluate what is the actual news that is happening even though that has already happened that's fine that they are fine with that even the newspapers are still running there are certain uh people who are loving to read them and I'm one of them I I I read

read them and I'm one of them I I I read my Telugu newspapers weekly once once in a while but uh in the world of uh the abundance that we are currently having. If you open YouTube, if you are selecting a wrong feed, if you already have watched some wrong YouTube videos or something, there will be some flash news and breaking news and everything even uh some some things that are happening live they shoot with their video and even a a news channel is reporting that. So in this world of instantaneous thing everyone wants to go ahead and get the things instantly in the world of social media and everything. So not only news like the events that are happening there might be a uh some musical show that that you that your friend goes and you miss out on that. So your friend puts a live on

on that. So your friend puts a live on Instagram or Tik Tok or something and you love and you envy them and all these things. Yeah. I don't see that as bad. That is FOMO. Of course that is FOMO. I want to link the same for FOMO thing that is happening about this technology trends like you asked the last question about there are so many things that are happening uh in the world but always remember that stick to the ground stick to the old ways. What is the old way is all these amazing companies OpenAI, Anthropic, Google Gemini, Microsoft, everyone have their own technical blogs, right? They are supposed to provide the I mean this news even any technical news that is happening they are reading these

that is happening they are reading these technical blogs they're understanding it they are understanding it or they are throwing it to chat GPT or something they're asking you to prepare a YouTube video or something and they are trying to put that video immediately or after some time or anything right right so you imagine that multiple YouTubers talking about multiple things you you feel that okay oh there is so much things happening in the world and I I have to keep up myself, but don't don't look at that. Okay? I'm not saying don't look at that. You should definitely look at that, but keep a keep a check, keep a tab on. Yeah. Recognize that it's manufactured outrage, right? That's their job to click to make you click that. And I'm

to click to make you click that. And I'm guilting this. I've definitely, you know, you know, you know, you you do it sometimes, but Yeah. No, sorry. Keep going. Yeah. I I love what you're getting at here. Yeah. Yeah. Yeah. So what I want what I would love to ask you to do is put some time aside, start reading reading the technical blogs which will also make your you know brain uh evolve more and understand why they are writing in that way. Why they are trying to you know if someone is writing a blog they are taking some efforts and some product manager or some developer has made that change and they are trying to put that as a technical blog right not only

as a technical blog right not only technical blog any blog that company puts out. So writing and reading is the best best thing even in this world rather than watching. So I always I I always have this uh one one nice thing that for myself God himself is a creator. He's not a consumer right? If if he used to scroll on on social media like us he he would never create anything right? So you should be like a creator of something you should definitely consume but also create at the same time. But coming back to your question, the smart tip that I wanted to give you is if you have some automation experience, if you have some N8 experience, if you have openclaw experience, go ahead give those technical blogs to openclaw and ask it to you know look at those blogs every

to you know look at those blogs every day or every six hours and ask it to summarize it and summarize that information and give it to you rather than a YouTube video. an RSS feed or a technical blog like this will definitely help you and uh keep you less uh uh fooled. fooled. Yeah. And I think the beauty of what you've recommended is once you do that for a while it it's like you know when people say that they stopped eating sugar and when they have like one lick of sugar it's so disgusting that they can no longer. So I think you'll then start to respond and recognize that you know like a random Twitter thread that's being like oh this one person marketing

being like oh this one person marketing company made 250k MRR probably made up let's face it that probably did not happen given what you know just using your own you know grounded method of consuming this information and yeah I do think that's a good way of you know like breaking out of that constant FOMO state. Yes. Well, Naven, this has been so much fun, man. You know, you you have to be back here. We I would love to do some like demos or some sorts. I'm sure we'll have a bunch of questions after this goes out. So, yeah, this has been so fun. Really appreciate you taking the time and yeah, I should definitely say this is my first podcast. Sorry, I was just saying thank you and but also

I was just saying thank you and but also that's so surprising that I would not have guessed that this is your first time doing like a podcast, but you killed it, man. That was awesome. Thanks. Thanks Nan. First of all, I'm very uh humbled and honored to be on your podcast. Uh Ready Said Do podcast is one of the things I was always looking up to. This is one of my uh what do you say the Andrew Huberman's levels or or Stop. You're just like next Freedman's Levels podcast for me. Trolling me now. No, no, I'm not trolling. But even for someone who is getting started in this journey and uh who who wants to share their thoughts to the world, I think you are giving a great platform to give their honest opinions and views. I think that is a great thing that you're doing.

that is a great thing that you're doing. Uh and second thing, I was so nervous. I I was literally didn't go to my gym today. I was just looking at your questions and I was preparing for all these things. Uh but it came out naturally. I I definitely did not speak out what I prepared. But uh thanks for being very uh open and uh being friendly. Thanks. Awesome, man. And you know, so this should finish uploading. That brings us to the end of that episode with Naveen.

to the end of that episode with Naveen. That was awesome. I'm definitely going to have him back for future episodes. If you like that and if you found value in that episode, please consider subscribing on YouTube and leaving me up to a fivestar rating on Spotify or any of your favorite podcast apps. Please share my clips and shorts with those that are near and dear. And yeah, I will see you all in the next one. New episodes every week.