The Builder's Blueprint: Pratiti Raychoudhury on Leading Teams, Democratizing Data, and Sundial - S02E05

The Original Source Is A Copyleaks Podcast

Welcome back to another episode of The Original Source! In our latest episode, host Shouvik Paul engages in a fascinating conversation with Pratiti Raychoudhury, COO and "Builder" at the new-generation analytics stack company, Sundial. With a career spanning over 15 years as a product executive at Meta (formerly Facebook)—where she was instrumental in the company's shift to mobile and led a team of over a thousand researchers—Pratiti brings a wealth of experience to the table, from her start as the very first market researcher at PayPal to her current role building an analytics solution for the AI era.

In this episode, they discuss:

• Pratiti's journey from leading massive teams at Meta to becoming a "Builder" at a startup, driven by her desire to innovate in the AI era.
• How AI is flipping the data analysis model, with Sundial becoming "insights-first" to dramatically 10x analyst productivity by automating grunt work.
• The essential need for companies to treat data as a compounding asset and the best practices for building a long-term data strategy in the new AI-centric world.
• The risks of AI adoption, including the fear of the unknown and hallucinations, balanced by the immense benefits of integrating tools like ChatGPT and Claude into daily workflows.

Follow us on social media to stay updated on all things Copyleaks.

Linkedin: https://www.linkedin.com/company/copyleaks/
Instagram:  https://www.instagram.com/copyleaksai/
X: https://x.com/Copyleaks

// 'The Original Source' is a Copyleaks podcast /

Transcript

SHOUVIK PAUL: Welcome to the Original Source. I'm your host, Shouvik Paul, and today I'm really excited to have on my show Pratiti Raychoudhury.

PRATITI RAYCHOUDHURY: Hi, Shouvik.

SHOUVIK PAUL: Hi, Pratiti. You know, you are one of the few people who said my last name correctly. There you go. That's one of the benefits. Yes.

PRATITI RAYCHOUDHURY: I was gonna say probably the first time anyone's done that, right? Yes, probably.

SHOUVIK PAUL: What I'm really excited about, Pratiti has spent over 15 years as a product executive at Meta, and she was a key leader behind the scenes helping to guide Facebook's massive shift from the web to mobile, shaping their strategy for their popular platform features. Prior to Meta, Pratiti started her career as the very first market researcher at PayPal. And currently, she is really the COO, and her official title is Builder at a new company called Sundial, which is creating the next generation analytics stack for the AI era. So welcome to the show.

PRATITI RAYCHOUDHURY: First of all, excited to be here. Thanks for having me.

SHOUVIK PAUL: I gotta say, so your official title is Builder, and I know the function of COO. What made you choose that as your official title?

PRATITI RAYCHOUDHURY: You know, when you join a startup, you wear so many hats, and I was talking to the founders when I was joining. I'm part of their leadership team, and I'm just like, what do I call myself? And you don't want, because you wear so many hats, including the founders, right? You know, we do everything. I'm doing things like marketing or sales, or things I've never done in my past lives. But then I was like, it's both me and Chandra, who's one of our co-founders, like Builder, because at the heart of it, we are building a product, a company. And also at the heart of it, I'm a builder. That's who I am. Hence the title builder. It might change over time, but when you know you're in a startup, that's what we are all doing, especially as leaders.

SHOUVIK PAUL: And the title builder, I'll tell you what, it sure is better than some of the titles I used to see a few years ago. Remember the titles of Ninja, Chief Samurai? So I like builder way better.

PRATITI RAYCHOUDHURY: I may have considered those ones too and then just went with it.

SHOUVIK PAUL: Amazing. Good choice. Good choice. So, Pratiti, let's start with your career, right? So you've spent the last 15 years at a company that fundamentally really changed how we use and look at data, right? And what did that journey look like from starting as a researcher there and leading a team, ending up, like, we're leading a team of over a thousand researchers?

PRATITI RAYCHOUDHURY: Oh gosh. I mean, it's been so long, but you know, first of all, I will just start off by saying that I feel incredibly blessed. It is a once-in-a-lifetime opportunity to work in a company like that and build something meaningful in terms of functions or products. Back in the day, it was like being a scrappy researcher. I remember like we didn't even have a lot of tools. We were getting people in our user research labs. We would run in the morning to Save Way to buy gift cards. Like, that sounds scrappy. We were, we would, there were like just four of us. We are like, "Does anyone else need gift cards?" To thinking about the future, you know, like one, starting from like, how do we define our value, not the role—like UXR was an established role, so it's not like—but how do we define our value? And two, like, who do we hire? How do we scale and grow? Like all of these things were, you know, kind of what me and, you know, as we grew, some other leaders spent our time thinking about. I think the thing that we did really well was actually defining our value. And that's probably one of the reasons why we grew so much, because we really, I think, demonstrated that we were an indispensable part of product building versus an afterthought or a nice to have. And that I think was the pivotal shift.

SHOUVIK PAUL: That, that's a really interesting point. Right. With a lot of companies, you're absolutely right. You focus in on the product, you focus in on the sales, and data is sort of like a thing you do on the side, right? Or it's an aftermath. Yes. So what you're basically saying is like, from the very beginning of Meta, it was very clear that data was gonna be sort of at the table.

PRATITI RAYCHOUDHURY: At the table. Yes. Yeah. Yes. And I also give a lot of credit and value to the leadership at that, in the company, because I was relatively junior at that point. It's not like I, you know, they really valued both data and people and insights and what are people saying? Like they just deeply cared about it. And, and I think so there was, otherwise, there's always this like tension between push and pull. And I think like both sides kind of like did their jobs.

SHOUVIK PAUL: Interesting. Yeah. Which is sometimes a hard balance to find, right? Because what you may be looking for or need and require from a data perspective may be very different from a product perspective.

PRATITI RAYCHOUDHURY: Yeah, exactly. Yes and no, because I think that was like the shift we made, because at the end of the day in UXR, we are not doing research for research sake. And we are doing it to inform product decision-making. And really like, I think understanding that, that we are all in this together. We come at it through different lenses, and I remember when I joined, my manager said this thing, which I still hold very dear to me. She was like, "If you don't understand the product and how it works, you won't be successful here." That was the best piece of advice she could, she gave me.

SHOUVIK PAUL: That's really sound advice, right?

PRATITI RAYCHOUDHURY: Yes. Yes.

SHOUVIK PAUL: So let me ask you, look, you, you rose to a very senior position at Meta, and in a field that's like pretty heavily male-dominated. In general, tech, right, is is male-dominated. And can you just talk about your experience navigating the landscape? Like what were some of the challenges you faced being a woman trying to reach like the higher rungs of leadership?

PRATITI RAYCHOUDHURY: Yeah. It's um, first of all, I was thinking that I, maybe, because I've always been in tech, I'm like, which field is not male? Good point. Honestly, to be very fair, I did not find it very challenging. And I will tell you why. One is right from the beginning we had Sheryl as our COO. And you know, having a female leader in that position was, it's not just like, see, you know, one is of course, you know, seeing is believing. When you see people in those levels, you're like, yes, I could be that, right, or someday. But the other thing is she, and as a result, like everyone else around the company, they really invested in us as leaders. So I think like, that's why it, it, like, you know, I obviously read a lot of articles and stuff about that. I'm, I feel, knock on wood, like very blessed that I have not had those experiences where I felt I had to like work harder or, you know, like, or you know, the things that you hear. Because I think everyone worked equally hard. The second thing I will say is also my mindset. And, you know, this is where I think everyone doesn't have the experience or the luxury that I went, experienced at Facebook. But the other thing is the mindset. And for me, I never let like being a female or the color of the fact that I'm an immigrant, like hold me back. I didn't just like, I don't, one is hold me back or let it like bother me. I don't think I ever thought about it day to day. "Okay, I'm a woman," or "I'm an immigrant, a person of color," for example. I'm opinionated. And I would just share my opinions. Like I, I didn't care who was the audience. I have an opinion, I'm going to share it. Right. And so I don't think I ever let stuff like that, things that are not in your control, hold you back. Like, this is who I was born as, is not in control. And, I just like, never let that hold me back.

SHOUVIK PAUL: No. But it's, it's a remarkable journey, by the way. And it's a lot thanks. Women and, and everyone, like you mentioned immigrants. It's something for everyone to, to look up to. You didn't you make some, also some most influential list or something? I read somewhere.

PRATITI RAYCHOUDHURY: I did, I did make the most influential list last year. It's the Gold Gala, API 100.

SHOUVIK PAUL: Yes. Congratulations. That's, that's awesome.

PRATITI RAYCHOUDHURY: Thanks.

SHOUVIK PAUL: So great. What a, what a cool journey. Okay, so cool journey over 15 years at Meta.

PRATITI RAYCHOUDHURY: Close to 15.

SHOUVIK PAUL: Not okay. Fine. Fine. Close to 15, but yes. Long time. Long time, yes. Long time. Over a decade, let's say. What, what then made you leave and go to, you know, do the startup?

PRATITI RAYCHOUDHURY: Um, even I, I remember my mom used to tell me this, timing is everything, but it's, there's never a good time and there's never a bad time. It's how you feel. And I remember, and you know, you and I know each other for a long time, since I moved to the US it was like early two thousands. And I was actually working at an agency in the city. And when you're in Silicon Valley, I'm like, I have to be in tech. It's like in the air, it's in the ethos. I'm not saying it's for everyone. That's just who I am. And that's how I felt. You know, with this whole AI, the emergence of AI, I'm just like, I have to build something myself in AI, in an AI-related space. And that's, you know, it's hard to say. Like, like I think that's what it was. It's less, yeah. It's just like what I really just like wanted to do at this point. And to your point, I had been at Meta for almost 15 years. It was a wonderful, wonderful company. And I knew I didn't want to go to the, another big company because then I just stay here. It's a, it's a great company and it's just that I was itching to do something. Right. I mean, or building something. Hence, builder. Yes.

SHOUVIK PAUL: And was that, 'cause I'm assuming even in these near 15 years, like there must have been other moments where you're like, oh, I should go build something else. Like, was AI, the fact that AI just sort of blew up?

PRATITI RAYCHOUDHURY: I think AI is the one. I actually, interestingly, when you said in the last 15 years, I actually didn't have that urge to like go and build something because I wasn't excited enough about any of the innovation that was happening to make, you know, give me that itch. Yeah. Which I felt now. Interesting. Yeah.

SHOUVIK PAUL: So you finally sort of like, like this is a really solid reason to do this.

PRATITI RAYCHOUDHURY: Yes. Yeah. Totally makes sense.

SHOUVIK PAUL: So in this new company, which is called Sundial, I was reading that your, the company's mission is quote, "to be the data product you need." So in your view, how has AI already starting to change the way companies like big or small approach data analysis?

PRATITI RAYCHOUDHURY: I think specifically to AI, because there are a lot of data tools in this space. One is a lot just even outside of oh my, even outside of AI, I think there were things Sundial was doing, which was extremely exciting for me. One was you know, just automating insights or being insights first. Because there are lots of tools in the space. No one starts from insights first. They are all in the job of fetching data rapidly to enable you to do the next level analysis on the next. No one is, and you ultimately get to the insights, but there's a lot of things you need to do to get to that. Whereas like, Sundial was always like started from this like notion of, "Hey, we need to automate insights." Because at the end of the day, the service of like data is in the service of decision-making. And for that you need insights and how do we like turn that around. But coming to AI specifically, in my opinion, I think we have just scratched the surface. And by we, I don't mean Sundial. I mean all the players are, obviously trying to be AI first, you know, starting from the big players like Snowflakes, Databricks to other like, companies like Omni and HX and so on. I think everyone is, we are just like scratching the surface and I think the reason being AI, doing something around analytics is way more complex. We can talk about that in more detail. Um, and that's why it's not yet been cracked. Like there are companies doing coding successfully and you know, other stuff. There's a lot of emergence around like say legal or design, but analytics is kind of, I. So I think like, so we, I will say, to answer your question, we have just scratched the surface. Um. What but one thing I will say what I'm seeing in new, especially for newer companies or like the new emerging AI companies, especially because those are newer companies, right? I think they're embracing even when it comes to analysis, like embracing AI much faster. And this is no critique of the big companies. It just takes longer to make shift in big companies for sure. Like even just like tactically to, they probably have a number of like, contracts in place with current providers, right? Like even to, like, they have to fulfill their obligations. So I think because newer companies are not coming and with that debt, there's, there's more nimbleness to adopt these things. Plus they're built as an AI company. That's the other thing for sure.

SHOUVIK PAUL: And, and also they don't have endless resources to do that data mining.

PRATITI RAYCHOUDHURY: Yes. Yes, exactly. Yes.

SHOUVIK PAUL: And if I understand it correctly, Pratiti, I think your, you know, when you say we, we start with the, uh, rather than looking at the data, we, you guys are basically flipping the model, right? Like, like ask the question, right? So I think in that example, rather than saying, hey, trends are showing that, you know, your ARR is decreasing here. It would be, show me why my, like, it'll basically answer the question. It'll say it's your ARR is decreasing, and it'll answer that question of why. And that frees up again, you don't normally it's like you look at, someone has to analyze it, get to the point that I think our R and it takes time to analyze, by the way, even for insight first, even to get to the data. Like what is your ARR? What was your ARR last week or last month? As someone who's not a data scientist probably doesn't fully understand this, there's a lot of work that the data science is doing to pull that data. Even that, we have made it very easy for a data scientist to pull that very easily because it, it like democratizes the field a little bit.

PRATITI RAYCHOUDHURY: In other words, you can by, by whether it's using Sundial or other tools, like the idea here, you can operate at the level of a very large company or an enterprise, even if you're a startup, essentially, right?

SHOUVIK PAUL: Yes. I would say Sundial probably has democratized it the most. Okay. Because again, we start, like if you look at our dashboards and you, we have this concept of pages. It starts with insights first. Whereas I think some of the other tools, they don't start with insights first. So the data scientist is still doing the work to pull that insight. Interesting. Yes. So, so like, you know, if I'm sitting here listening to this conversation as a data scientist or as a analyst, I should say, I'd be really worried about things like Sundial right now. And in terms of like losing my job. Should I be worried?

PRATITI RAYCHOUDHURY: Oh, absolutely not. I, I think what Sundial does is actually 10x's your productivity. You know, to break it down in, I think the very essence of analytics has four jobs. Okay. First is like, what is going on? Say what was my weekly active user last week? Right? Then what are the drivers for that? Or like, what drove that change? Um, you know, say a product or a feature or something. And then it's like, what will happen? Say if US contributed it, since, uh, that weekly active user change. And since you mentioned ARR, what will happen to my ARR if US whatever drove that change? And what should I do next? Otherwise everything is a data, but the analyst also spends a lot of time doing trying to say, tell their stakeholders what to do next. Right? Or what should I do with the data? So I think what happens with a lot of, you know, the stuff that we are doing or others are doing around AI is we have really automated the first two jobs. What is going on and what are the drivers. So, earlier, say an analyst would spend, I, and that's the thing, like if you, you know it, a lot of the work is being done by the analyst behind the scenes that a stakeholder, like, a C-level exec to a PM or whoever doesn't see. They're probably spending 60% of their time or more in the first two jobs. And 40% in the second and third that I mentioned, which is like, what will happen, what should I do next? Now if that is automated and it takes like a few seconds. I can spend much more of my brain power and my time doing the other two, which like is like really important and spend like, you know, see, I think it shifts. You know what's very interesting, you know, one of the reasons I joined Sundial when our co-founder Chandra was showing me, and we all by the way worked at Facebook together, so we all know each other. We like, not just know each other, we work together. So when he was showing me, you know, demoing Sundial to me and when I, when I was, you know, this whole notion of insights first and I was like looking at it, I was like, "Chandra, we would spend a couple of weeks looking at multiple dashboards, joining different tables to come up with this insight that now we, it is possible to get in a few seconds." This is just like life. Like I would've loved it if like when I was doing this work, if I had it, I would've loved it. Right. And the fact that I can do the other two now, spend more time doing that. Right. So anyway, to answer your question, I don't think their job is at risk. I just think one is, I believe it increases their productivity and their job changes. The flavor of it changes as a result.

SHOUVIK PAUL: No, absolutely. By the way, we've through history, we've seen this over and over, right? Tools like Excel came out, you know, QuickBooks. Oh, accountants. They're gonna be out of a job. That never really happens, right?

PRATITI RAYCHOUDHURY: It exactly. Or if you think about industrialization when you mention history. That's right. No, it, in fact, what, if any, opened up more jobs, just different jobs than what people were doing. Right. And and like you were saying, I think it allows folks to focus in on what they do probably best or truly like utilize their skills. And the grunt work gets automated, in other words, right?

PRATITI RAYCHOUDHURY: Yes. Honestly, ideally AI should also, that's the thing even we want to do at Sundial, also get to the three and four, which is what will happen and what should I do. But what's interesting is we do, and Sundial does have a team of data scientists, who among other things, does this work? Because we need data scientists to think through these things. And because, AI, AI is prone to hallucinations as we know. Yep, yep. But in analytics it is, but it is opaque. It's very opaque. So you need users or experts to be able to use it and tell us what the errors are. Like that's an example of how the jobs will change.

SHOUVIK PAUL: Yeah. There needs to be, in other words, like a human in the loop. All we, right.

PRATITI RAYCHOUDHURY: Exactly. Yeah.

SHOUVIK PAUL: Well, what are some, so, so obviously clearly a lot of great benefits when it comes to using AI associated with data. What are some risks? You mentioned hallucination, but what, what are some risks?

PRATITI RAYCHOUDHURY: I don't think, first of all, I don't think hallucination is a risk. It is. It is part and parcel of where things are now. And as, like I said, as models get more and more advanced, which we are seeing is happening. Or as we like per, for analytics right now, there are lots of humans involved in the errors. I think as we develop like playbooks, which is kind of what we are doing at Sundial, those will be largely addressed. So I don't think hallucination is a risk. I'm just, I'm trying to think like what is a risk, you know, because it is so new. There's lots of unknowns. And I think with unknowns there's always a risk. Because you don't know how to like, you don't know what will happen. Like, I think even this whole thing around like jobs being at risk, I think it's, and you know, I shared my point of view, but I think it's the fear of unknown versus like anything else that it's, there's a risk, right? There's a fear of unknown. And, you know, we are human beings. We like certainty. We don't like uncertainty. Right. Um, but I, I, analytics specifically, honestly, I don't know the risk. I actually think it'll just like increase productivity. And it'll, you will be able to have analysis in your fingertips, which was like not even possible. Um, there could be other risk as AI gets better and better, which I'm sure you've been reading, I've been saying, saying like, being able to impersonate someone, it was happening. It just gets even easier and even harder to detect. But what actually, what on the positive side, when you, when I say jobs are changing, I'm sure there'll be a whole field that will emerge on like detecting things like this, right?

SHOUVIK PAUL: That's literally what we do here at Copy Leaks, right? So we, we help a lot of companies sort of authenticate content, whether it's published or, there you go. Yes. A lot of insurance companies are, are, are now, these are new problems, right? It used to be, uh, yes, you, you, you get into a car accident, what's the first thing that happens? Your insurance company says, well, exactly. Send me a picture of the car. Right? Not that you couldn't have photoshopped it there, it's just that. Now it's just infinitely easier in a matter of seconds for the average Joe to put a dent in a car. Using whatever. Using AI. Yeah, using AI. Right. And, and so every industry is now having a real reckoning with how to curtail fraud, curtail. We work with like news companies that are like, "Hey, you know, we're in the business of news, in the business of breaking news." They wanna be the first announce it. Yeah. Now. So they've always scoured information, Twitter, whatever. Now they're going, if I'm seeing it, seeing and hearing is no longer believing. I don't know. Did you see Sora two just got announced and immediately there was a video of Sam Altman stealing something from, from Target, which it looks so real. I mean, you have to also see the humor in this, right? You probably know that's not real and you just have to like see the humor in it.

PRATITI RAYCHOUDHURY: Sure. What, so what, what's your, I know this is not related to data at all, but what's your view on all of this? I mean, one is of course you have to see the humor in this. Like if it is funny, right? But there could be bad things that is, that could also potentially happen. You know, my view on this is, you know, of course I don't work in any of these companies. Sure. I think all companies approach things like pretty responsibly. Even if, I don't know what the stories out there, but you know, like they announced Sora, but like two days before they announced, slightly orthogonal, but like parental controls. That means they are taking things like safety very seriously. That means there's a whole team behind the scenes that is doing this work is what I would assume. Yeah. So I just like, and I fundamentally believe that like companies will approach this in a very, uh, responsible way. So I think that's kind of how I would, that's my take on it.

SHOUVIK PAUL: Yeah. Hey, coming back to some of the, you know, we were talking about risks earlier. One of the things that I think actually is a risk is not having that human in the loop, by the way, over reliance on anything that's pure play, LLM generated without having a human in the loop might be a risk, right?

PRATITI RAYCHOUDHURY: So yes and no. And this, and I, when I say yes and no is if it's a very tightly defined, well-scoped problem, then I don't know if you need a human in the loop.

SHOUVIK PAUL: Okay. Gimme an...

PRATITI RAYCHOUDHURY: If it's like I'm thinking, what is say writing a line of code, you know, you or you're developing an app. A lot of people like will use like Claude or something. You don't necessarily, I feel, need, and depends on, again, the complexity. I think it is. And that also goes to like, our jobs going away and or the repetitive task. It's less about repetitive, but how narrowly a thing is code versus something which is more ambiguous. So, I mean, even take like this image generation thing, a very tightly defined scope will be, "Hey, I want to create cat videos for this friend of mine who loves cats." Yep. It's a very like, tightly, for their birthday. Yeah. It's a very tightly different scope. Yeah. Versus hey, if, even if it is in jest, like I wanna create...

SHOUVIK PAUL: Yes. In that, in that yes example, for example, like you're right, saying something is, "Hey, create a image of a cat." And now what I'm saying is you would still need that human to validate in fact, oh, that it's a cat before you publish it straight to your website or something. I'm just saying that, right?

PRATITI RAYCHOUDHURY: Yes, probably. Yes. I, although I think, like, I think the models are getting pretty sophisticated now to detect these things. Fast forward data, we may not care. Right? We, we, we, yes. Yeah. So good. Yeah. Whereas like the Sora example you gave, but you know, I laughed about it because it's kind of funny, funny. But like the fact that you can create a video of a famous person, you don't know how it'll be used. I think those are things you would need a human.

SHOUVIK PAUL: Correct. That's right. That's right. Okay, so let's assume there's this, this, world where AI has completely really simplified automated data analytics. Then I'm putting together a let's start with like, a new company. You're, and how do I build, or let's even say mid-size company at this point. So I've, I've gone through the basic stages of growth. I now really need to start growing out my data team in the new world. With AI, you used to run a team of over, like, around a thousand at Meta. How would you build that differently today?

PRATITI RAYCHOUDHURY: Huh. So I think, I'm thinking, first of all, I think I would still say, because you, you know what you said, think about data as your compounding asset. So that means you need to think about it early on and not when you've had the growth. So one is like, just think about it. So, because, you need to invest in the proper data infrastructure and think about it like investing in proper data infrastructure as a first class citizen hire. You don't have to hire an army, but hire a good data science leader or or so who can do this thinking for you, especially if you are not a data scientist by training. Um, and the other thing is like, this was actually, I was going to thinking like think longer term, and this goes to like, don't think about it when you've had the growth. Because one is like even little things you need to understand early signals for your PMF or product market fit, right? Typically what, or even like, so the market is so fragmented and what I've seen play out is that even at that stage you're like, okay, I need to log my data. So you only think about like events logging. So there are companies that do just that. And then after a point you're like, oh my God, I have grown. Now what do I do? I need these insights or whatever. I need more conflict analysis. The thing is you to think you will, you should assume you would grow and invest in it. Like honestly. And like Sundial, we always is all in one and shameless plug. But this is an example. There are some others also in the market. Yeah. But if you're doing events logging, you will not get to this like advanced, like more advanced analysis or insights. So you'll need it. It's a question of when, but now you're suddenly like, what do I do with this tool that I have? And now I have to like have this other tool. And what ends up happening is companies end up with like multiple tools now.

SHOUVIK PAUL: A hundred percent. I, I, I think you just hit the nail on the dot. Like, look, I mean, it's a, it, it's this tricky game that I think every growing company plays, right? Yes. I'm too early, I probably don't know this. And then when it gets to a certain stage, the, you're like, I wish I had this data, but I, I didn't, I don't have the data set so I can bring in the tool, but I don't have the data set. Or you're like, preemptively bringing in all these things to your tech stack and, and someone usually the CO is going, what the hell? What's this money being spent on? No one's using it. What is this actually, what are we gaining? What's ROI off this? Yeah. And, and it's a complicated balance. And so I guess what you're saying is like finding some sort of a tool set which allows you to, even from the very early days, track it all. Even if you don't, right away.

PRATITI RAYCHOUDHURY: Exactly. And you know, we've had like, you know, we were talking about newer versus older companies. I am like, there's some new companies I'm seeing are doing that. They'll reach out to us. There's like, hey, our PFO is looking at ways to consolidate a, like a bunch of tools. And we looked at your capabilities, like you can do that. So I do think like some of the newer companies are already starting there, but that's like my one or few things is like, think about data as a compounding asset. So don't wait too long for it with that. Like hire someone who knows how to do that job. And third is like, really think long term. Like not what you need right now, but what you will need like even a year from now, two years, especially when you're a startup. And you might reach your inflection point very quickly.

SHOUVIK PAUL: Yeah, that's right. You just don't know. Right? You just don't know.

PRATITI RAYCHOUDHURY: Yeah. And then you're suddenly like, oh, I don't have the logging in place, or I don't have the right tools in place and stuff. So I think one is like also, that's why I said like this is the kind of like priority Yeah. List is like hire that data science leader who will then like, help you understand like for sure how to think long term. Because obviously if you're not a practitioner, you might not think about it that way.

SHOUVIK PAUL: That's right. That's right. Yeah. So, so well what, what kind of clients are you guys working with right now? Like, who are your clients?

PRATITI RAYCHOUDHURY: You know, it's on our website. We have a long list of a lot of gen AI companies and starting from Open AI, CamerAI, Character. These are all on our website. We also have companies like Venmo. We have signed contracts with couple of really big companies that because they're not official yet. Yeah. Sharing.

SHOUVIK PAUL: Yeah. Those are, those are some phenomenal names. Yes. Not convert. Yes. For startup to have. So moving to a different, different question side here. Like, we talked a lot about AI. What, what are some things that Pratiti you use, ah, on a daily basis? Or, or, or like, and this could be for personal use, this can be for professional use.

PRATITI RAYCHOUDHURY: Both. I use both for personal. Professional. I use Chat GPT a lot and also Claude. I have not used Gemini as much unless it's already integrated with Right. You know, their products, like I use, end up using Gemini notes a lot actually. That's one of the reasons now I use Google Meet much more than Zoom because of the note taking capability. That's right. Uh, but I think Chat GPT and Claude quite a bit for a lot of like both, like personal and professional, almost like everything. I'm just like, okay. Let me see what Chat GPT...

SHOUVIK PAUL: Are you using the speech functionality? Are you, are you one of those that talk to it all day long? Are you typing? How do you...

PRATITI RAYCHOUDHURY: I do both. I do both. Uhhuh, yes. Uh, but a lot of times I will, uh. I end up typing because I might be just like looking at Yeah. You know, reading something and then Yeah. I'll just type it.

SHOUVIK PAUL: Yeah. Tell, tell me. I feel like we all have these moments where I'm like, oh, wow, that's so cool. Like, I didn't even think to use it that way now that I did. I'm gonna, I'm hooked. Did you have one of those moments when it came to these LLMs?

PRATITI RAYCHOUDHURY: Oh gosh. Coding for sure. We were like having a hackathon and we were like, I don't think, like, I would've been scared of coding. Yeah. But like, Claude made it so easy, or any of these, right. Cursor is another one. I think it was like definitely one of those. Well, but you know what is very. Interesting is I do a lot of market research using Chat GPT Uhhuh, and I'm like, wow, this is pretty cool. Like, I don't have to read a bunch of market research, like reports or surveys and stuff. And I think that was like, actually, you know, because being a practitioner, I was like, wow, this, this is like pretty cool. You know, because I mentioned I'm also like doing some marketing GTM kind of stuff. I think the output you get is like very, like to get started is like pretty good. Yes. For sure.

SHOUVIK PAUL: Yeah. It, I feel like the more, like, that wouldn't have even been possible, like even three, four years back, I would've had to hire my first marketer by now, but I'm like, yeah, I'll wait.

PRATITI RAYCHOUDHURY: No, that's, that's so true. And also like, I feel like the more you use it. The better you become also at prompt engineering and the better the results are.

SHOUVIK PAUL: Exactly. You know, and, and the more you sort of, um, realize like the potential of these things, you know, like even for my personal use, I'll tell you one of the biggest things that I've done and it just now is in my day-to-day workflow is Sure. This is the same with you. People send me all sorts of stuff. Like, like I'm not just talking about articles from the very beginning. I'm like, oh, summarizes that was easy. Right? But people will send yes, like really interesting video. You know, but there'll be some Ted Talk type seminar or something and I'm like, super interesting. I really wanna watch this. And I found myself like saving them, hoping to watch it at some point. And I really never did. And now what I've started doing is just, I'll just basically get. One of these LLMs to go through the transcript, which now, like on YouTube, is available for them to go through. Yeah. Look at the transcript now, summarize it into three, three or four key points. And I find myself like learning a lot more from even videos that again, like there's a lot of good content in there and information. It's just who has time, right. To go through 10 videos, you know, like, yeah.

PRATITI RAYCHOUDHURY: Oh my God. I have to tell you my wow story actually. Okay. I think I had to send a rejection email to a candidate and you know, it's always like the hardest thing to do and I'm just like, what do I write? Like the tone has to be like, you know, and this friend of mine, my colleague is like, just give it to Chat GPT. Yeah. They'll write an email for you. I'm like, you're kidding me? They're like, no, that's what I do all the time. Yeah. And so I just. Get the prompt to Chat GPT and the email was amazing. And somehow you feel like less guilty, right? Yeah. Right.

SHOUVIK PAUL: No, you're, it's like getting HR to write the email or something, right? Not your problem.

PRATITI RAYCHOUDHURY: Yes. I'm just like, wow. But this is like, I'm just like, this is life changing. Just like the emotional burden was not there.

SHOUVIK PAUL: Amazing. Amazing. Well, listen, it has been an absolute pleasure speaking to you and just getting your insights today. Um, you truly have been, watching your journey and over all these years, uh, and becoming sort of this prominent person in the Bay Area in, in Meta. It's been a wonderful journey to see, uh, and, and really inspirational. So thank you so much for coming on the Original Source today.

PRATITI RAYCHOUDHURY: Of course. And thanks for having me. Yes. This has been a blast. It's so much fun.

SHOUVIK PAUL: Uh, and to our listeners, thank you so much for listening to this episode of The Original Source. Please do, uh, follow us on, Facebook, LinkedIn, X. Um, join us in the next episode, to stay updated on all things AI and of course, Copyleaks. Stay original, stay safe. Thank you. And see you next time.