How To Kickstart Your Singing (& Music Composition) Career In Bollywood - w/ Trishita

Let's begin background question. Tell me about your journey into HR and why you're excited to lead as an HR manager here. It's a great perspective. Valuing people as the heart of the organization is key to HR leadership. Let's move to the next part. part. What if I told you that was not a real person over the phone that VJ was talking to. VJ was being interviewed by an AI that he built and that was actually a demo. Why go through this?

actually a demo. Why go through this? Well, not only does it solve scheduling, but also it removes the bottlenecks that recruiters face and it also autogenerates summaries and helps you build a dashboard for relative scores of all the candidates that were interviewed. There are bunch of AI providers, right? You have Gemini, Brock, OpenAI, Anthropic. What we have told ourselves is not to get overly stuck with one AI provider. You want to keep it AI agnostic.

keep it AI agnostic. I'm Nam Pande. This is the ready set do podcast and in this episode my guest is Vijay Swami Nathan. Vijay is a bit of a serial entrepreneur, the host of the pathfinders podcast and co-host of the authentic hustle podcast. In this episode for us exclusively VJ pops the hood on his AI recruiting tool recruit.AI and quite generously shares the entire tool stack as well as the build process that he followed. If you're trying to build anything using voicep powered AI, I can promise that this episode will be immensely beneficial for you. As a side note, I've also been a guest on Vijay's podcast, Pathfinders, and I linked the episode here and below the like button if you'd like to check it out and show some love.

like to check it out and show some love. No one can win at being you, right? No can be VJ or no one can be N. Your authenticity is your only you. That is a marketing play. No one can compete. Did you buy code this or did you write code by hand? So latency is a challenge. The lowest is like 600 to 750 milliseconds and it goes up to as high as even,500 millconds which is sometimes a lot. In line with our theme of learning from somebody who's just a few steps ahead, my goal with this episode is to spotlight voice AI powered apps and give you all the tools and resources that you need to start building out your own. Subscribe on YouTube and any of your favorite podcast apps for weekly episodes featuring high agency individuals and

featuring high agency individuals and daily clips from those episodes on YouTube and Instagram. And now without any further ado, here's BJ. Welcome to the only podcast in the world featuring stories of high agency individuals who are just a few steps ahead of us. VJ, welcome. Hey Nam. So excited to deep dive into the cutting edge of where you've been operating, right, with all of these various AI initiatives, you know, such killer tools and resources for all of, you know, our listeners to look at. And I know we we'll be covering a few different things today, but to start off, do you want to um demo the use case that you built out for recruit AI and just, you know, maybe tell us a little bit about just to set the stage what it

bit about just to set the stage what it does, what problem it solves and, you know, we can then go from there. Absolutely. Uh so before I uh show you what the demo is and I'll just paint the picture just to give context. Yep. So uh there are two things that that we are working on. Uh one is for helping companies with sales and business development. Uh where if you have one person say one or two person sales business in the sales of business development and you want to increase the capacity by 10 or 20 times right. So how can you do that with the power of AI? So pretty much if you have like a spreadsheet and phone numbers uh

like a spreadsheet and phone numbers uh we have the capability to make hundreds and thousands of calls. So that's what we're building with conversion.com. Uh so which is started with my co-founder Brian Hayes out of Dallas, Texas. So that's one initiative. Uh so it's all about sales and and whatnot. And on the other side I'm also from the tech. I have spent a lot of time on the tech and I see constantly recruiters reaching out to people Monday to Friday 9 to5 and to screen candidates to get like future candidates and to probably sell a push into them and they can it's constantly happening and if we we wanted to help and we want to bring AI to help and if you could uh the recruiters can

and if you could uh the recruiters can do those screening with those candidates with the help of AI uh for different variety of questions. So uh that's what we are doing with recruit.ai AI uh alongside my co-founder Avina Sandal Kumar. He's out of Vinsor, Ontario. So that's these are the two initiatives I'm working on besides my software consulting firm Zadwell. So that's uh that's just an intro. Uh maybe I can get started with a uh like a recruitment uh demo if if you're okay with that.

demo if if you're okay with that. Yeah. Yeah. I would love to see that in action. Yeah. I'm so excited. Let's let's do it. So what I'm going to do now is going to uh I'm going to simulate. So AI will be conducting an interview. Amazing. Uh for a for a human resources manager question. So I'm going to pretend I'm a HR. I'm not a I don't have a HR background, but I'm going to pretend I'm in HR. So HR, if you're watching, uh excuse me for the wrong answers. So uh hello, I'm Laura, a senior HR director at Lootard. Uh hey Laura, thanks for calling.

Uh hey Laura, thanks for calling. Thank you for interviewing for our human resources manager role where you'll lead talent strategy and foster workplace. looking for your background in HR specific and a leader scenario in this quick demo. Let's begin. Sure. Sure thing. Journey into HR and why you're excited to lead as an HR manager here. What inspired your HR career? Absolutely. I always loved working with people and in an organization. I think uh the the human beings make make the organization and they bring the vision to fruition. So I I love to help them on their journey uh and on their journey in the company and if they have any questions or uh any challenges along the way. Uh so I so that's why I chose this profession as a HR manager.

profession as a HR manager. A great perspective valuing people as the organization is key. How would you develop a strategy to improve employee engagement in Yeah. What was the question? Can you ask the question again? Of course. How would you develop a strategy to improve employee engagement in a hybrid workplace? If you'd like, you can share a few more times. Yeah. So, I think for a lot from what we are seeing is employees like to be heard uh and valued for their efforts and opinions uh and cherished along the way and when they feel valued and uh s and their opinion is sought from time to time through feedback and surveys and whatnot and also engaged through like different organizational activities. uh we see a lot of like positivity and then

we see a lot of like positivity and then like great ROI for those kind of activities. So that is what I have done in my role as a HR. That's an insightful approach. Prioritizing feedback, recognition and engaging activities truly resonates in a hybrid environment. Now on to the final question. Tell me about a time you res employee relations issue. How did you approach it and what was the impact? Uh the there was a pressing uh need for a massive uh recruitment but uh there was this employee who uh then he he was in a critical time of his face. His father just passed away and he was also approaching to a to a marriage and the project was also seeking completion. Uh so that's when we kind of leverage hybrid work and remote work even like 10

hybrid work and remote work even like 10 years ago when remote work was not as popular. uh so he was remote but at the same time was able to finish the project successfully uh and also uh finish his family duties and also finish his marriage wonderfully. So that is some experience I always cherish uh being a HR HR sounds like you handled a very delicate situation with empathy and strategic thinking ensuring thank you that's all for our demo. Do you have a quick question about the role? role? Uh yeah so how how soon can I join this position and what are the benefits in the role? We aim to fill the position promptly and our benefits package includes comprehensive health coverage, professional development opportunities and supportive work life with more

and supportive work life with more details. Excellent effort today. Thank you. Very welcome. Thank you for this time and we'll be you have a great day. So that was all AI, right? Uh it's not a person speaking. It's really incredible. Yeah, like the voice, you know, it Yeah, it was very very lifelike for sure. like yeah definitely just the voice piece right if you don't go off of the actual content and if somebody were to just focus blindly on just the voice um yeah I think that was extremely convincing if you ask me yeah and we were going to say it is an AI person talking I mean we were never going to pretend that it is an it is a human because it is that real it's like uh you yeah it can and we tried other

uh you yeah it can and we tried other languages like like uh Spanish and like a British English and like an Indian English and we tried a few different accents. So interesting. interesting. Uh it it it did come out well. Yeah. Yeah. Now, so I have obviously a few questions around this, right? So obviously I can appreciate the value of doing this because now you don't need somebody to, you know, do the scheduling. You don't need somebody to call some other person at a fixed exact time and obviously you can only be on one call at a time, right? It's not like a one person can have you know like three different screenings all at once right so this fixes that so obviously you can parallel process I don't know even hundreds of applications like that

even hundreds of applications like that so that part is super valuable obviously has a lot of value I think what I'm curious about is what kind of the like behind the scenes look like from from what you just did so what sort of analytics would the recruiter or the company be getting from that call uh yeah absolutely uh I just want to like describe what you mentioned there is absolutely concurrency parallel processing uh because given in the current day and age of a situation let's say 10:00 a.m. any given day, Monday to Friday, if a HR is making a call or a recruiter making a call between 10 and 11, right? Each call is probably 10 to 15 minutes roughly, give or take, right?

15 minutes roughly, give or take, right? So they can probably make four to five calls in an hour. So with this you can scale it to probably you can make 50 calls in an hour. So so the scale is like what what they can achieve like probably they can do 20 30 calls with this they can probably even do 500 or,000 calls a day. So easily. Yeah. Easily. Yeah. So that's the power of scale. I wanted to explain that. uh and your question on analytics what is possible. So when conducting an interview in this uh recruitment setting uh each organization will have different kinds of parameters how they want to gauge a candidate right so okay how is their like uh leadership skills how is their like language skills how is their like experience on particular domain x y and z right so they may have different

and z right so they may have different parameters for each organization so that can be configured and uh so we are like like a beta stage we're trying early stage we are trying test and so there are parameters that we set uh and we can measure against those and we can put a score against 1 to 10. How do candidates measure on those skills? Okay, how are they how is their leadership experience?

they how is their leadership experience? How is their like domain experience? Right? So, so we'll give a score. Okay, their leadership skills is seven, domain experience is eight or things like that. So uh overall uh if they have like 20 candidates to call so the AI can go and out there and make those calls for you and it give out a beautiful report saying that hey this is what happened and this is this is what it looks like. So So yeah exactly and I think that report piece that you mentioned is kind of what I'm you know focused about focused on.

I'm you know focused about focused on. So for instance uh I'm assuming obviously that you would have baked in the uh the ability to condense what you said and you know have that shared with the hiring manager right? So what you said around your answer to the behavioral question which was how did you deal with like a difficult situation in which you said you empowered that employee to you know take that day off or whatever. um like do you have that pipeline flowing through in a way that so so basically that they don't have to refer to the entire transcript and oh oh yeah yeah no what what we will do is like actually with these both initiatives what we do is like after every conversation so there is a twoline summary of the entire uh call so they don't have to sit and hear the

so they don't have to sit and hear the recording of the entire tenants let's say they call 200 candidates in a day they don't have to literally uh listen to 200 calls actually is they they'll get a quick twoline summary of what happened in that call. Okay. uh and and they can definitely use that to leverage and all calls are also searchable n so every conversation that happened with the candidate right so they it is all searchable with the keyword let's say the candidate said Java for instance I don't know I'm just making it up right or they said like uh what particular domain explain like manufacturing for example right so you can search for those keywords and then maybe you can start playing from there onwards and see oh what did they say about manufacturing or Java for for for instance. Yeah.

instance. Yeah. Yeah. That would be really cool to just kind of see or maybe what the like the back end of that looks like. It's okay if you don't have that now, but if you can like, you know, maybe share any pictures, whatever the case might be, and I can like throw it up for our listeners to view whenever this airs, I think that would be really cool because yeah, I'm really excited to just imagine uh you know, like the back end of what this would look like for that company.

this would look like for that company. So like there will be a dashboard, I'm assuming. So you know like 70 out of 150 candidates called there'll be like a breakdown maybe 40 answered maybe out of those 40 30 will be green 10 will be red or something. Yeah we are we we we have built the MVP and we're actively building it out.

and we're actively building it out. Currently this R&D is still like on onward but that's that's where we are headed actually. So the core uh I I showed you the demo of the core piece that we just saw. Exactly. Exactly. Uh but uh but yeah the conversation can be searchable. I'll send you a screenshot of like how we can search that conversation and things like that. Yeah, that would uh the actual report that's something we're still building, but it's it's coming. It's all coming together.

coming. It's all coming together. Amazing. Yeah. Sorry. And I have this habit of getting ahead of myself, right? I'm just, you know, yeah, really very easy to get um excited for me at least about stuff like AI. Um next, obviously I'm curious in understanding how you built this. So obviously I understand there is some IP at stake here. So without revealing too much uh you know just for anybody that's interested in building something similar or you know making like an interactive realtime AI voice uh system of some sort what are some of like the core tools and technologies that you use if you could go over that that would be obviously this is voice based intelligence and so something that we are leveraging uh some tools that we leveraging is called lean labs so for

leveraging is called lean labs so for for different voice and if you want to do different accents and cloning and things like that so we 11 Labs we use Cartisia. So those those are some of our voice providers out of the box that we we do we do use that. Uh and also now that now I think people are all familiar with chat GPD open a open AI. So there are a bunch of AI providers if you notice right like like telephone network providers there are AI network providers are like large now you have Gemini you have Grock you have open AI and uh like you have perplexity anthropic and the at least there are like 12 to 15 different AI providers so what we have told

AI providers so what we have told ourselves is not to get overly stuck with one AI provider and so we want to keep it AI agnostic in a way so that let's say let's say if anthropic is not working we want to switch to grock or we want to switch to Gemini or to open AI chat GPD right so uh uh we kept that layer uh agnostic so that we don't have to be facilitated because what might work in English doesn't necessarily have to work the same way in a different language or in a different accent right so the models are different uh we really see that and deepseek of course we do support deepseat all as well so we can change the provider uh so

well so we can change the provider uh so that is one part of uh how the conversation is actually consumed. So the way it actually happens nmon is like uh person speaking it's actually we are converting that to a text speech to a text uh so all of that text is transcribed. So we use some tool called deep gram it is it is used for transcribing you might have heard about all these like different calls that people use for like go uh the Google meeting and so many things where the transcribing happens. So deep gram is one such tool which helps you in transcribing. So uh from speech to text and it transcribed beautifully and from and again once LLM has processed and has an answer we have to process back from text to speech again to speech.

text to speech again to speech. So so both these of these needs to happen. So we use deepgram for that. So those are some of the tools uh uh that we that we use to put this uh solution together and of course there's some IP involved but I think that's if if anyone has a question that's that's how it actually works. Yeah. No, and that makes total sense and thanks for sharing the names of those tools. Obviously, I'm familiar with 11 Labs. I think they are definitely the market leader. I feel like when it comes to anything voice, I think they're at the very cutting edge of that technology. I in fact, I think I recently saw um you can now type out a transcript, right? So you can have like nan colon and within like your double inverted quotes say why are you standing

inverted quotes say why are you standing there and in parenthesis you write surprise stone and when it generates that it will actually add that surprised inflection in the voice itself which is just that it just blew my mind to think that you know that's just possible to do now and it'll be my voice which is the craziest part of all probably but something that I noticed was just how quick it was. there was almost very little latency when between you saying something and the person you know like responding to your thing. So how did you make it to be so like less uh like how did you reduce the latency so much? So latency is a challenge. The lowest I believe I think covers is like 600 to 750 milliseconds uh latency and

750 milliseconds uh latency and sometimes it goes up up to as high as even,500 milliseconds which is sometimes a lot. uh so it depends on so for each uh AI that we're building for each assistant that we're building which is actually representing a certain job it is doing a task right this AI for instance is playing the role of a recruiter and it was actually recruiting candidates right so it's entire script is based around recruiting u and for companies for for different needs they will have different kinds of information that they want AI to learn right so so AI will be applying based on those knowledge bases. So, uh it has it should be care we should be careful not to overfeed the knowledge base just because we can

because we can uh we shouldn't feed like a one terabyte of knowledge base you're probably going to get like 100,000 millisecond response so not not get a response at all. So we have to play uh so like we have to play between those things not overfeed it but at the same time give what is only needed because it can handle pretty well any conversation it can uh is mature enough to handle that. Mhm. Mhm. Yeah. So speed is one thing and uh that is something we we exploring more on it and we want to reduce them response time as much as possible. Yeah. So yeah, that's you know already so fascinating what you said around um and like you've said this a few times where you you've

said this a few times where you you've said that you want to play around with it, you want to explore it. So I think one of the questions that I definitely have and I'm sure a lot of our listeners have as well is in this space right in the in this day and age and as of recording this is July 2025 it literally feels like there is a new earthshattering technology every single day right and I think one of the biggest um questions I get from just interacting with my peers friends is just that it's so hard to follow what is going on because by the time you catch up to a news that is now old news

catch up to a news that is now old news and there are three new models that do that thing better. So what is your method of you know kind of staying up to date with the latest and greatest with AI developments and to what degree do you do that if at all? Okay. Uh got it. I mean you you will not know everything. That's a fact. I have told my I understood that as a fact like probably eight or nine years ago and I told myself I will not know everything.

told myself I will not know everything. Yeah. Uh but when I want to know something, I go and dig for it and and I find find out. That's how it has always been. been. Uh yeah, I I still would say I don't I don't know a million things. So I'm happy to say that. Uh but but like I I I don't there's no secret hack actually because there's a uh there are different communities I'm involved in. I mean like for blockchain for instance. So people keep bringing new interesting stuff one way or the other. And of course you have seen us do podcast and things like that. There are different guests. You and I will get to speak to they bring different ideas like challenge challenge you on what you

challenge challenge you on what you thought was the ideal way. They come from a completely different world. That's another source of information. Um and when I what I when I learned the most is also when I'm preparing for a guest to speak for them, right? Because I try and dig in what they're doing and like prepare an introduction for them what they're doing. So that is also a learning moment. Uh yeah, every conversation is pretty much a learning moment actually uh to be very honest.

moment actually uh to be very honest. No, no, and I agree definitely as a also obviously, you know, we're both uh we we're both podcast host hosts and I can relate to what you're saying. There's definitely a great deal of just engagement, right, that happens in in a way that we wouldn't have done in our normal lives if we did not if we weren't hosts of our respective podcasts. And you have a few of them at this point. So you you're obviously exponentially multiplying um that thing already but yeah and I think I guess from there where my mind is going next is um what can you tell us about the build process?

can you tell us about the build process? So like how how much time did this take you to build? Um did you like vibe code this or did you write code by hand by scratch or I guess what are you willing to share about the actual build process? Yeah. So the build process is like so we are originally like software consulting company right so our original thing is like with Zadwell so we do software consulting it's a services company so we have our remote operations in India since I'm originally from we are from India so uh we have uh base in Bangalore uh so we kind of leveraged that uh we we

uh so we kind of leveraged that uh we we didn't we didn't wipe code uh any of the prototypes that we made so uh we did some for the website but mostly for the app and all the integrations and stuff like that, APIs and all of that. It was all handcoded. I'm sure like our our devs, our developers did use chat GPD. Maybe they used Oh, yeah. For for bug fixing and Yeah.

Oh, yeah. For for bug fixing and Yeah. But we didn't like AI didn't code it. If you're that's what you're asking. No, I didn't code it. Got it. Uh it was people who wrote the code. Uh yeah. Is there a reason? Is there a reason for that? That you went that way or or was it just not something AI could do? I'm just curious. Oh, okay. Fine. Now it's not the capability thing. It's uh it's like how can we quickly uh uh create that MVP uh that we can go out and raise money for and present the idea to someone and that was our primary uh intent and I think we were confident enough that we can we can do it uh with with

that we can we can do it uh with with the information that we had collected. So putting it together and then create those modules and selfcode it and with with the team u that that was the idea actually. I mean nothing against wipe coding and it's beautiful actually. I've tried it. It's fantastic. Yeah. Yeah. Yeah. It's it's it's maturing and and sometimes it goes into the hallucination mode too. It kind of doesn't know where it is and you point to a bug it's kind of creating a new feature altogether. So Right. Right. Yeah. And it just trips over itself and Yeah. Exactly. It has no idea what to do next. Yeah. Yeah. Just runs around in circles, right? I love that. Well, Vijay, thank you so much for taking the time here today.

much for taking the time here today. I've learned so much from you. And you know, as somebody that's somewhat also deeply involved with AI stuff, it's always really good to pick the brains of another person that also is. And I'm sure you'll probably agree or maybe you won't. I don't know. But I don't know many people that are this deeply involved with what's going on as we talked about, right? Every day there's a new development, there's a new breakthrough. So yeah, I think I really value where you are at with all of this and you're somebody that I look up to with all of your various initiatives.

with all of your various initiatives. You really seem to have more than 24 hours in the day for for whatever reason, at least from my vantage point and in a really cool way. So yeah, you know, really happy to continue to support you from the sidelines and probably collaborate down the road, but really appreciate you taking the time here today. Hey, likewise, Nan. Uh I I really had a great time uh today being a guest and and I also loved interacting with you uh months ago on Pathfinders. So uh uh yeah I try and like I've I've I uh yeah I try and like I've I've shared your links ready set do uh sometimes. So I think it's a fantastic podcast and whatever initiatives you're doing and I think you lead from the heart. I've seen that time and time again. Thank you.

again. Thank you. Uh keep keep doing what you're doing naman. So uh wish you the best man. you're always looking forward for something new. So, I like what you're doing in terms of the how to I mean you're it's based on how to, right? So, I I of how to Yep. I I love what you're doing with that. So, yeah. Congratulations and all the best. Thank you. Thank you. That brings us to the end of that episode with VJ. I don't even know why you're watching right now. Go start building. I know I have and it's definitely helped me. 11 Labs is just such a game changer. If you would like to support me, the easiest way to do that is by subscribing on YouTube and leaving me up to a fivestar rating on Spotify or any of favorite podcast apps.

Spotify or any of favorite podcast apps. Something that goes a really long way for me is if you by word of mouth share with your friends and family about how you found your new favorite podcast. Catch you all in the next one. New episodes every Wednesday.