What we talked about
Yash Chaturvedi, Head of Product for Live Sports Ads at Amazon, joins Federico to unpack how AI, computer vision, and product management are reshaping streaming and advertising. Yash traces his path from early OCR projects at Staples to Prime Video, where he helped build long-form video understanding to automate global compliance and power features like maturity ratings and content descriptors. That same foundation inspired his work in virtual/brand placement:seamlessly inserting context-aware brand assets into scenes:leading to multiple U.S. patents and a new, non-interruptive model for monetization.
Show notes
Yash Chaturvedi got his first product management job because his hiring manager at Staples needed someone cheap:he openly admits it:and was willing to bet on a driven candidate from a tier-B school. Twelve years later, Yash holds seven approved US patents, leads live sports advertising at Amazon Prime Video, and jokes that filing a patent is like standup comedy: “you say something wild and crazy, and then you just try to justify it with your science and engineering team before the lawyers cut you off.”
What we covered
- When Yash joined Amazon, Prime Video was streaming across 200 territories, each with its own content regulations, and operators in multiple countries were still manually reviewing titles one by one against local standards. He built the long-form scene understanding platform that replaced that process:teaching a model to describe what is happening in any given scene or frame:which directly produced the maturity ratings and content descriptors that now appear on every Prime Video detail page before you hit play.
- The same scene-understanding capability led him to the insight behind virtual product placement: if a system knows there is an empty table in a scene, it can place a Coca-Cola can on it and charge Coke for every viewer who sees it. He described it as a win-win-win: content creators get an additional revenue stream, viewers avoid interruptive ad breaks, and advertisers reach audiences who are actually engaged with the content rather than leaving the room.
- The virtual product placement patent:which Yash describes as the one closest to his heart:is written not just to describe where the technology is today, but where it needs to go: seamless insertion, pixel-perfect compositing, audio-video QA, and the full pipeline from advertiser asset delivery to on-screen rendering. He treated the patent as a product vision document, identifying building blocks and their connections before the engineering team fleshed them out.
- On live sports, 30 to 40 percent of sports content airtime is currently dedicated to advertising breaks. Yash’s charter at Prime Video is to rethink that structure:replacing interruptive ad loads with formats that are more native to the viewing experience, including digitally augmented brand logos placed directly on the field of play. These virtual placements can already be personalized by audience segment or geography, meaning a viewer on the East Coast and one on the West Coast could see different brands on the same field simultaneously.
- Yash described the evolution of the PM role in the AI era not as a replacement of job functions but as a compression of timelines during the discovery phase. A sprint that once took two weeks can now take three to four days because PMs can prototype ideas themselves using tools like Cursor and Lovable, show something concrete to engineers and stakeholders, and get real feedback before committing to a full build cycle. He called this a shift from roadmaps to versioning:committing to a mini proof of concept first and deciding on the next iteration only after seeing how it performs.
- His parting advice to product managers: “Don’t be scared of AI taking your job. Be scared of the PM who knows how to use AI better than you.”
About Yash
Yash Chaturvedi is Head of Product for Live Sports Ads at Amazon Prime Video, where he has spent the past several years building AI-powered products at the intersection of computer vision, machine learning, and streaming monetization. He holds seven approved US patents, primarily covering virtual product placement and long-form video understanding, with additional patents pending around generative AI applications.
- LinkedIn: http://www.linkedin.com/in/yashchat
- Website: https://amazon.com
Episode 68 of the PreVetted Podcast.
Full transcript
Federico Ramallo (00:00) Welcome back to the PreVetted Podcast, where we spotlight extraordinary people and remarkable talent reshaping our world. Our guest today is Yash Chaturvedi, his head of product for Live Sports Ads at Amazon. Yash has built his career at the intersection of AI, computer vision and product management, leading teams that transform how audiences experience digital products.
He is also an inventor with multiple US patents, advancing how machine learning powers personalization and engagement. From scaling CVE ML products to defining frameworks for PM success, Yash brings a rare mix of deep technical insights and visionary product leadership. Yash, welcome to the show.
Yash Chaturvedi (00:50) Yeah, thanks for having me, Federico. Pleasure to be here.
Federico Ramallo (00:53) Awesome. So can you us your story of how you transitioned into product leadership at Amazon?
Yash Chaturvedi (01:02) Sure. So I’ve been a product manager in the product management space for about 12 years now, 12, 13 years. And I started off my product management career back in 2014 with Staples. The hiring manager at Staples, I asked him one time, like, hey, why did you hire me? So he was like, hey, I was looking for someone.
cheap because he didn’t have enough budget. So he was looking to hire someone coming in from a tier B school and was driven. So I kind of stood out and apparently I was cheap back then for him to afford. So that’s how I got into product management to be honest. ⁓ But ⁓ for the last eight, nine years, I’ve been in kind of the e-commerce space before I came into Prime Video. And I’ll talk about my journey within Prime Video.
Federico Ramallo (01:36) Hahaha
Yash Chaturvedi (01:47) But my experience goes from search, recommendation, personalization of e-commerce website. I’ve been with three fortune companies so far, Fortune 500 companies. And my expertise lies in kind of marrying the machine learning aspect of technology with kind of business problems and customer problems that the company is trying to solve. I have a deep technical background with my…
engineering, my bachelor’s in computer science, engineering, and with my master’s in data science. So, I understand, I always tell this to my engineering team. I can know when my engineers are kind of fluffing up their level of effort for any of the things that they’re kind of courting me as part of the sprint planning. they are always kind of cautious when it comes to scoping stuff for me because I can kind of…
catch them and tell them like, when they’re kind of overestimating.
Federico Ramallo (02:41) know firsthand how much time things are going to take. So you know when somebody is giving you false information.
Yash Chaturvedi (02:44) Correct.
Correct.
So like my engineers are always careful. But yeah, so I got into Amazon, within the Amazon Prime Video space, they were looking to hire and kind of build a product team around trust and safety, which is, and it lies under the umbrella of compliance. So imagine Prime Video streams in about 200 territories. Every territory has its own rules and regulations. So you have to make sure
titles that are getting streamed or movie series are that are getting streamed in those location or territories are compliant with the local rules and rules and regulations and back then when I joined a bunch of stuff were still done very manually we had operators in various countries that we were hired that we had hired
They were reviewing these titles manually and kind of saying, okay, like, hey, this matches this particular standard operating procedure of this country. So it was kind of, and we had millions of titles. So like a lot of titles didn’t have, were not meeting compliance. So like, that’s why they kind of hired me to see, okay, how can we kind of bring in some machine learning into the mix to identify what’s going on in the scene and kind of, and this is where the whole,
birth of long form video understanding came in where I came in and kind of understood where what the team was doing and how we needed to solve this at scale. So that’s where I came up with this idea of like long form scene understanding where how can you understand what’s going on in a scene in a frame in the entire title and kind of formulate the description about that title about that frame about that scene which gave birth
to maturity rating on Prime Video and content descriptors on Prime Video. So now every time you go to a detailed page on Prime Video, you see hey, whether this title is rated RPG. And also as soon as you play the title, you see whether this contains nudity, violence, smoking scenes and whatnot. So that kind of came as part of those long form video understanding platform that kind of came out of my first year within Prime Video.
And as part of that, also invented a bunch of new technology or new customer experiences within Prime Video that did not exist before. And I very interesting stories, which I know we kind of dive deep later on. But that kind of invented the whole, as long as I can understand what’s going on on the scene, I can also monetize and say, okay, if there’s an empty space on a table, can I put a Coca-Cola can?
on the table and monetize and go to Coke and say, okay, millions of customers are going to see this. Are you willing to partner with us and sponsor this particular scene? So that kind of gave part for virtual product placement, which is one of my biggest accomplishments in my professional career. And then for the last one, one and a half years, I’ve been focused more on the live events and sports monetization.
So how can we think about if you’re watching Thursday night football or NBA on Prime Video, how can I monetize the viewership? Because apparently 30 to 40 % of the sports content in general is advertising at break focus. So my kind of charter is to figure out how do we make it more customer centric, customer focused, rather than giving them interruptive breaks in the middle.
How can we make much more streamlined at customer experiences for viewers on Prime Video.
Federico Ramallo (06:09) So now you’re able to do product placement with LiveFit.
Yash Chaturvedi (06:13) Yes, we can. We don’t call them as product placement because what we do is we do brand placements on the field. a lot of times you’ll see sports, soccer, cricket, and you see these brand logos being augmented on the field. And those are traditionally were kind of painted on the field, but now the tech has come to a place where we can do them virtually.
And there are a lot of third party companies and like Amazon also has its own homegrown solution that kind of helps us do this.
Federico Ramallo (06:44) And with that they can personalize the fit for different countries, right? They can change the, yeah. I’ve seen those basically they are, as used to be now they have a placeholder and you basically can on the video that placeholder.
Yash Chaturvedi (06:59) Correct. And like there are targeting parameters based on like which advertiser wants to target what kind of audience and what not.
Federico Ramallo (07:06) Right,
right, yeah. It used to be like two sides, right? So you could do like, if you’re feeling from one side of the field or the other side of the field, then you’d have two different brands, right? Two different configuration if you like, I don’t know the jargon,
Yash Chaturvedi (07:15) All right. All right.
Yeah. So that’s what I’ve been up to these days and my journey within Amazon.
Federico Ramallo (07:26) Right, And what initially drew you into computer vision and machine learning products?
Yash Chaturvedi (07:33) So my love for computer vision started back in 2014 when at Staples we were trying to figure out how do we scan invoices and that’s where I was introduced to some of the OCI techniques. So essentially it’s kind of asking computers to detect patterns. I kind of give this example where imagine you’re I have two kids eight and three
So I have to teach them every aspect of like, hey, whether this color is red versus yellow as they’re kind of in the very younger age. So it’s similar with kind of introducing these patterns and these kind of notion to the machine. So that has been always exciting of like, how do you train certain models, computer essentially like to detect certain patterns, certain colors, words.
So that’s where kind of I’ve been fascinated with machine learning and computer vision since 2014. But now, and because of some limited computations back in the day, but we live in a very different era in the last like one to two years, lot of things have changed. Compute power has kind of drastically improved. Introduction of LLMs, generative AI coming into the mix. So there has been a big leap.
in terms of where the technology was back when I started versus like where we are today. So like in the last 10, 11 years, a lot of things have changed.
Federico Ramallo (08:53) Right, a lot.
So what do you think sets apart a great PM working in AI from the rest?
Yash Chaturvedi (09:01) So recently I gave a talk on this topic in one of the AI conferences in Vegas this year. So there are kind of two, at least in my mental model, there are two different type of PMs. One is kind of a traditional platform PM who’s still writing a lot of kind of big BRDs, 20-page or 30-page BRDs and like it’s much more waterfall-ish.
where you’re trying to make sure you are specifying your P0s, P1s, P2s from the get-go and then you’re working with your engineering to flesh out the scope of the project, the whole delivery timelines. But in the last one to two years, some things have really evolved with many tools coming into the market.
such as kind of like cursor, lovable, a lot of things where prototyping has kind of drastically streamlined and improved. And there’s a new hybrid of kind of PMs that are coming into the mix. And I’m also kind of like trying to teach that to people that I mentor, which is kind of a mix of UX researcher who’s doing the research, a UX designer who’s
designing these systems and kind of prototyping. As a PM, you’re kind of marching, giving instructions of like what needs to be included, what customer problem are you trying to solve for? And then they’re also your BIEs and also your engineers. like all of this combined is kind of now a hybrid version of an AI PM where you’re doing the research quickly, you’re designing systems quickly. As a PM, you’re kind of marching through things and then
you’re developing prototypes, pushing them in production. Then you’re experimenting, you’re analyzing the data, you’re being the BIE, the business analyst or business intelligence engineer. And kind of now there’s a hybrid model of all these job functions coming into the mix and forming a AI PM for a lack of a better word. That’s kind of the evolution that I’m seeing at least in my space where these
Federico Ramallo (10:59) I see.
Yash Chaturvedi (11:06) four or five job functions are kind of getting combined versus a traditional PM, UX designer, engineer kind of. We still need engineers, but it’s more like how can you get to prototyping really fast? So that way you’re not just taking an idea to an engineer or to a scientist, but rather kind of a fleshed out idea of like, hey, here is a UX or here is customer and a problem that I’m trying to solve for by doing this.
So it’s much more like engineers are much more able to connect better with what the idea is and kind of articulate and can understand and visualize what you truly want to do. And then you have as a PM, I’m able to kind of prototype and experiment really quickly so that way I have more data in terms of what’s working, what’s not working and where do I need to pivot. So I can make much more kind of informed decisions rather than waiting for these long cycles where I’m taking my requirements to engineers, their building stuff.
Then I do certain reviews to see whether they are valid or whatnot. And then you push them into production. You see whether it’s working, you experiment. So it’s kind of the cycle that used to be a two weeks, two weeks sprint is now kind of very minimum to like three to four days. So we are seeing some efficiencies coming into the mix with like this hybrid AI, IPM, kind of new function that’s coming up in the industry in the last one to two years.
Federico Ramallo (12:25) I think that it’s very interesting how the product manager role is changing now with AI. And now that the expectation is that the AI PM will be able to do multiple roles at the same time. I’m still trying.
Yash Chaturvedi (12:37) I would not say
that the PM should be able to do multiple roles at the same time, but it’s more kind of which phase of the product that you’re in. If it’s in the discovery phase, you should be able to do all of these job functions, but you still need designers because as a PM, you’re never going to be able to design a fully fledged design that your system wants like, or your vision is, right? So you still need these job functions, but it’s more
the discovery and kind of production time that you’re saving for your team and kind of providing that clarity based on data rather than just a certain hypothesis.
Federico Ramallo (13:11) One of the things that I noticed with AI is that there’s a lot of hype about AI. And I’ve been trying to figure out how much is hype, how much is real, And I think bad coding, it’s a great tool for product managers to be able to build prototypes,
But I feel that that code cannot be implemented into production environment, in particularly enterprise projects, right? Maybe if you’re in a startup and you have a clean state, you’re starting to write code, maybe, but that’s very specific use case. But most of the cases managers that are on a mature project.
on a mature code base, right? And I feel that by coding and building prototypes should not be considered to be included into the production, right?
Yash Chaturvedi (14:08) Not yet. That’s what I keep saying. Not yet. Hopefully in the next four to five years we are there.
Federico Ramallo (14:12) Not yet, yes.
Yeah, I usually add not yet because probably by the time this episode is going to be there. Maybe not, but it’s improving very fast. that’s what I’m trying to figure out, how much is hype, how much is real, right? Yeah, but I can see how as a PM having
Yash Chaturvedi (14:21) Yeah, Sora 3 made them out.
Federico Ramallo (14:36) a lot of experience using AI allows you to do the initial work of all these other roles, right? That’d be a first statement.
Yash Chaturvedi (14:45) Yeah, so like the job function of a PM is still the same. You’re the advocate of the customer, right? You’re kind of defining what needs to be done and why it needs to be done. like that job definition just doesn’t change. What’s changing is the way we do things and kind of bringing in the efficiencies. So again, like all these frameworks of waterfall previously came in because you had these hardwares that needed to be shaped.
Federico Ramallo (14:50) Right.
Yash Chaturvedi (15:11) they were all one way door kind of shipments where you put in a requirement like hey a mouse needs to go from X to Y if a person does X right and you build millions of dollars millions of copies and then ship it so there was no feedback loop that was coming in and you had to probably wait like a month or two to get like customer feedback and whatnot and then you had to kind of make improvements to the second iteration of your hardware
And then came the online kind of the whole website, A-B testing and product management where like the feedback loop was much faster. So you can kind of do A-B testing really quick, get data, see what works. But with AI, the speed at which we are kind of experimenting and prototyping, that’s kind of getting very squished in. So the framework is still applicable of like, hey, what, how do we experiment? How do we build product? But
just the prototyping of how quickly we can get to the prototyping space. And the way we are kind of thinking about requirements is evolving.
Federico Ramallo (16:10) I see. So before Waterfall kind of one shot, one kill because the cost of making changes was so high that it was worthwhile to do the prep work before, Then when we moved to Agile, it was because the cost of changes very small, right? So we could do multiple changes even throughout the day and it has little to no cost, right?
Yash Chaturvedi (16:36) I keep saying this, like imagine when Amazon released their one click buy option, right? Like the speed to market was kind of the mode, right? Like, hey, like let’s launch this and get it patterned so nobody can use it. But now the way things are evolving is like, if let’s say your mode is first to market.
you’re gonna fail because essentially you’re doing the research and R &D for your competitors of like what’s working what’s not because the replication of technology is so easy these days because I can just go to like take a screenshot and go to any of the LLMs and say hey write a quote for this right. Speed to market is not the mode anymore it’s like how fast you learn
from the experimentation is what the moat is these days. So it’s not like just pushing things to production and like, hey, I’ve delivered this, but it’s more, let’s launch this, let’s learn from this and let’s be the first one to learn and then improve. So it’s more constant reiteration of your product that is the moat now compared to just first to market previously.
Federico Ramallo (17:40) I see. I see. And the…
The use of AI on the product managers allows you to, what allows you to do better than before.
Yash Chaturvedi (17:53) So the way at least I use LLMs these days is mostly kind of threefold. One is more on the discovery side of like doing some research of like, what data exists today versus where the gaps are. So kind of doing more discovery, more kind of diving deep for lack of a better word. Then using that information, getting to…
the prototyping phase as I mentioned, for me that has been the biggest unlock where I’m trying to say, okay, hey, like here is an idea that I have certain bullet points and I can quickly move to the prototyping phase of like, hey, here is what the UX looks like or here is what the idea is. So you can more substantially visualize what the idea is. And then the third thing is kind of streamlining the writing.
Because at Amazon we are a very document heavy, we have a very document heavy culture. We write about everything. We have six pager, three pagers. So we have a lot of internal systems that I kind of use to quickly get.
something on paper so that way can share it across my stakeholders and get their kind of feedback in. Previously it would take me a month or two months to kind of just flesh out the idea, get in a shape where like I can get the document in a shape where I can get feedback from people. But now like within a week I can kind of bring a prototype and a document in front of them and get some feedback quickly. So the three ways I’m using is more discovery, more prototyping and more kind of documentation and more kind of alignment.
driving with LLMs.
Federico Ramallo (19:29) Interesting, interesting. Yeah. I can see how that will help you a lot on your role. It will allow you to do a little bit of the other roles that you mentioned before, and also will allow you to compete on this race of continuous improvements with the competition. Interesting. If you were going to advise an aspiring product manager today,
Yash Chaturvedi (19:46) Correct.
Federico Ramallo (19:55) what career frameworks have it or recommendations in general would you give them?
Yash Chaturvedi (20:02) And like I’ll try to tie this to kind of the evolution of AI product managers. So the framework at least what I use is you always think big, you always think in terms of where you want to be in the next three to five years. You start small, you think about, what I clearly know where I need to be in the next three to five years, but where do I start? So start very small, take up one…
one customer problem, one tech that needs to be solved and then kind of learn fast. As I mentioned, the more it is more like how do you learn faster to build the best product that your customer needs. What I have understood or kind of learned in the last six years being with Amazon is you need to write everything down. So you always like we have this document called
PR FAQ, which is essentially a vision document. So you always start with a press release on, when I launch this product, what this product will look like. So that way it provides you clarity of like, once you start writing down, you get clarity of what the product is, what problem are you solving, where you need to be and how you need to get there. So write everything down is very critical for like aspiring PMs. Bill, intuition with data. So
Don’t wait for like everything needs to be kind of like, hey, I need X data to make this perfect or like I’m waiting on some research from my research team. Just start building products with intuitions. Use data as kind of a guiding principle of like, okay, here is slight directional data that I can use to build, to get started. And as you…
build as you prototype, as you experiment, you’ll start getting more data that will kind of build better intuition in the future. I do recommend reading, working backwards. It’s a book that talks about how you think about vision and how like you work backwards from a customer problem. One of the habits that I would say is like always build a mini POC.
before committing to a full roadmap. Traditionally, PMs are known to kind of formulate an entire roadmap of their product by quarter, by year. But what I’m seeing in the last one to two years, what has changed is like there are many proof of concept that are coming up and then based on how they scale, how they perform, then you start figuring out, okay, what’s my next iteration or what’s the next version? So from a roadmap, now we need to get to versioning.
rather than kind of formulating an entire roadmap. So at least that’s what kind of my two cents would be to the new PMs. Don’t think about roadmap, think about versioning of your product.
Federico Ramallo (22:34) Amazing, So you’ve been granted multiple patents, right? Can you walk us through one that excites you the most?
Yash Chaturvedi (22:46) So I have, last I checked, had seven approved patents and a few of them are pending and the ones that are approved are mostly around virtual product placement or kind of the non-interruptive ad formats or long form scene understanding or video understanding that I spoke about earlier.
And there are a few pending around generative AI and kind of like some of the newer stuff that we’re working on. But one pattern that is really close to my heart is the overarching pattern of virtual product placement. So what it does is like if people have time or are interested should kind of go and read this virtual product placement pattern, which kind of shows the mental model of how a product needs to be built. So it talks about not where we are today, but where we need to be in the future.
So it kind of drives an overarching technical design of like, hey, if I were to do virtual product placement or non-interruptive ad formats, how do I go about building things to where we are, where we need to be? So it talks about how like everything is seamlessly inserted, how everything is pixel to pixel perfect. How do we do audio to video, QA, QC?
It kind of has various components. So that particular is very close to my heart because it just kind of shows like the thing big aspect of like, hey, let’s outline how this tech needs to be built. What are the key pillars that needs to be part of this? And then slowly and steadily start working towards building these components together.
Federico Ramallo (24:21) amazing. So what led you to filing these patents? How did you the problem and what drew you to create the patents?
Yash Chaturvedi (24:32) Yeah, so as I mentioned before, I was working on this long form video understanding platform and we were, and this was during COVID time. So we had a lot of spike in our viewership and we were trying to see, okay, how do I, how do we as a team monetize our videos and movies and CDs combined? So this, like we were brainstorming with my engineering team and suddenly the idea came of like, Hey, if I can understand
a video of what’s going on in the scene, why can’t I insert brands and product because I have all the context that is needed for it to be inserted and product placement isn’t new. Product placement has existed for over 100 years. We were just trying to think about digitalization of product placement. Right? So think like there’s an empty space behind you, Federico. Imagine if I were to paste a Coca-Cola poster.
And for every view that this podcast get Coca-Cola has to pay you X dollar of premium. Right? Wouldn’t that be nice? So that’s the conception of the idea that came from and we’re like, hey, everyone’s going to be excited. It’s a win-win-win because A, content providers, content creators that are creating the content will be excited to get additional revenue stream so that way they can make more better content. B, for the viewers, everyone…
hates interruption during a movie or a series. So it’s a way of like non-interruptive ads because you still need money to make content. Right? So it’s a way of monetizing content without interrupting your viewing experience. And then for Amazon and for our business, it was another revenue stream that would help us make more better content for our viewers. Right? So it was a win-win-win situation for all these three stakeholders and also for advertisers, right? Because let’s say if I’m engaged in the content,
people are much more likely to notice a brand rather than tune out when there’s an ad break. Right? So mostly if there’s an ad break, I’ll just go grab a water or go use the restroom. Versus if I’m engaged in a movie, I’m watching a Coca-Cola poster right behind Federico because I’m so engaged.
Federico Ramallo (26:37) Right, right, the ad starts, we kind of disconnect and do something else, right? ⁓
Yash Chaturvedi (26:42) Alright, or you’re
on your cell phone, you’re on your second screen or you’re talking to your partner if you’re watching with someone.
Federico Ramallo (26:49) I remember when we used to watch and we had the ads and we ran to the or the fridge to get some food, you know, and we’d prepare something quickly and we would just, you know, hear when the, you know, the announcement of the TV show was starting again and rushing in. We don’t want to lose any moment, we didn’t have the opportunity to pause.
Yash Chaturvedi (27:07) Yes.
Federico Ramallo (27:12) like now we have with Netflix, right? So it’s something you know, my son doesn’t know about, for him it’s normal to pause the TV,
Yash Chaturvedi (27:22) Yeah, and like even with the streaming kind of services coming into the mix, now every streaming service have an ad tier, right? And there’s only so much levers that these ad supported tiers have that at the end of the day, I can only keep increasing the ad load. So instead of seeing two minute ads, now you’ll see three minutes ad because they got to make more money revenue, right? So this is one way of like monetizing your content without adding ad load on your customers or your viewers.
Federico Ramallo (27:50) Right, And that’s the thing, when you can pause the movie or the TV show, ads become much more relevant because now people know that, well, if I miss, I can go back a little bit, Or I can, you it’s easier to disconnect from the ads, If it’s within the movie or TV show, then…
You’re more eager to watch it. Yeah, I can see that happening. ⁓
Yash Chaturvedi (28:16) They’re more immersive,
more native to the content, more non-interruptive, so you’re not tuning out.
Federico Ramallo (28:20) Right.
Right, remember a that the character was drinking and it was on the cinemas. And then when it was released on, you know, I don’t think it was VHS, was DVDs probably. But when it was, you know,
launch for a rental, they changed the Coke to another brand because they got, know, another brand was able to pay a lower premium because, you know, it was, you know, it was, it on on the rental movies, right? Lower audience, I guess. And further down the road, when it went to the, you know, to platforms, they changed it.
to an ice tea company that I don’t remember the name now and it’s not relevant, this idea that now the of that product placement was much lower, right? So then other company was able to get in and the producer were able to get the benefit of product placement three times, right? Instead of like…
Yash Chaturvedi (29:31) Yeah. Now imagine like if
I were to say like for people watching this podcast on East coast, get to get to see Coca-Cola versus on West coast, get to see Pepsi. Right. So that’s level of kind of personalization and targeting that you can bring in if these are digitally kind of augmented on the screen.
Federico Ramallo (29:50) Amazing, amazing. So you saw that problem, that led you to, or opportunity, you like, led you to file the paid patents.
Yash Chaturvedi (30:03) So this is where I think the role of kind of creativity comes into the picture of like anytime you’re thinking about publishing a pattern, it’s more like understanding the tech feasibility and a product storytelling. So the product storytelling was like the brands had a pain point, advertisers had pain points where they were
The creatives were fatigued, the costs were rising, customers didn’t really enjoy interruption in their viewing experience. with technology coming into the mixture of like, hey, now I can understand context of every scene.
Now I need to streamline a bunch of other stuff of like, where the ad creator will come in. Let’s say I need a Coca-Cola can to be placed. So I need to work with the advertisers to get the Coke can. Now how do I insert them seamlessly in the stream? So there’s a lot of research and development that had to go in. But we had an idea of like what needed to be done. So this is where like we were like, hey, let’s find the pattern. that way, and like we were of course partnering with our science and engineering team. They were the kind of
people who are in the driving seat, but at least as a product manager, I was able to articulate what the vision is and where we needed to go. And that was kind of the foundation of this pattern. Where like, hey, what are the building blocks that we need to build? How do they kind of connect and why do they matter? So like it’s more like creative storytelling of like, hey, what my product is? What do I need to build? How do they all connect?
and let the engineers and kind of the science people do their work. But as long as you provided that clarity, it’s easy for kind of the product to be patented easily rather than kind of just juggling and figuring out, what do I get pattern versus not?
Federico Ramallo (31:52) Right. Yes.
Yash Chaturvedi (31:54) And I made
a joke recently where people were asking like, hey, I’m interested in filing a patent. How do I go about this? So I kind of joked and said, filing a patent is like a standup comedy. You say something wild and crazy, and then you just try to justify with your science and engineering team before the lawyers cut you off. So it’s like you’re outlining everything.
you’re working with your science and engineering and the lawyers will say, okay, hey, let’s kind of stop this. And this is what the pattern looks like. So that was kind of one of the jokes that I recently had.
Federico Ramallo (32:29) I love it. I mean, as far as I understand, I mean, and this is probably very simplistic version, but the way I understand patent is, you know, whoever said it first wins, you know, it’s, a, have to basically prove that nobody else came up with this idea, right? And then the patent is yours, and that could mean a lot.
could mean anything, could mean nothing, right? It depends on what the idea is about,
But I find that process interesting. And there are patents that are wild and crazy, and their patents have been very useful to the humanity. I find that we the creative become victims of the bureaucrats.
So, yeah, but I see the value of the patterns because that allows you have solid foundation to build something upon it.
Yash Chaturvedi (33:29) Yeah. And also on a personal note, like it helps you kind of build your or establish your legacy, right? What you got to be named for. For example, like there are companies such as Tesla, they don’t believe in kind of closed patents. All their patents are public. Anyone can use it, but it’s more kind of the legacy that they’re trying to build with the people that are filing for those patents. Yeah. It’s like a nice way of saying kudos, like good job.
Federico Ramallo (33:35) Right.
Right, right.
it’s an open patent kind of thing, right? Because support the use of the yeah. they recently they published a patent where they talk building the cart, assembling the cart differently other companies, right? So in traditional companies, they put, you know, a chassis and then they put everything together.
Yash Chaturvedi (33:58) Good night.
Federico Ramallo (34:19) on top of this, but here basically they build part of the car and then they assemble the part of the car and then they them together and that kind of becomes a car. So they’re pre-painted, they’re to the car that is on the line and that’s a different way to build cars. see the value of doing those types of patents.
what advice would you give somebody that is thinking about patenting their idea, a patent basically, know, how do you balance that, you know, creativity until the moment you realize, okay, I should start patenting this idea.
Yash Chaturvedi (35:01) So the way at least I do I think about patterns is you need to write down every what-if that comes to your brain which is like okay if I were to do this what if this happens so listing down all the what-if scenarios and don’t be limiting so don’t kind of close down what if like this doesn’t work
Don’t close down that idea, write it down then you kind of think about how do you solve that particular thing or like what the of the tech components or vision needs to be in order for you to solve that. And then I can’t emphasize enough like start partnering with the legal as soon as possible, as soon as like you have this idea and like you have a kind of a sort of a blueprint of what you want to do because
They do this for a living. They know like what to extract out of you. So it’s always good to kind of partner with legal from the get-go. And then treat it like product discovery. There are gonna be many dead ends of like, hey, like what if I were to do this and this doesn’t work, okay. So like think about like you’re doing a product discovery while you’re writing a patent. So it’s more also like helps you kind of define what the product is.
define what components you need to build. So a lot of discovery that goes in probably 40, 50 % of our time during the product discovery, treat it as such. A lot of dead ends, but you’ll find some gems that will come out of it that will not only help you file the pattern and get it approved, but also help with your product launches, product development.
Federico Ramallo (36:41) And how soon do you think they should pursue the patent? Everybody can have a great idea, But it’s execution and the validation that actually makes it an amazing idea.
Yash Chaturvedi (36:54) So this is where
the product storytelling and the tech feasibility have to kind of come together, right? It’s not like I have this wild idea that I’m going to solve for cancer. There has to be kind of some tech feasibility of like what components are needed in order for me to solve X and what exists today versus where my idea needs to come in to help kind of bridge the gap. So it’s like, you need to have a blueprint of like what the tech
feasibility is today and where I need to be plus what problem am I solving and what components do I need in order for me to solve that problem. So it’s not like you have a wild idea, let’s get that patterned versus like having a realistic tech feasibility with the product storytelling is very critical to getting a patterned approved.
Federico Ramallo (37:30) Right.
Right, Yeah, I mean, I use a rudimentary example, but yes, I agree with you. It has to be something that you already proved, that you already has been able test, shape. And the other thing that the other value that I see on the is that if you have a good idea, the scrutiny of going through
know, challenging that idea, right, validating it with, you know, real, real use case, right. ⁓
Yash Chaturvedi (38:12) And this is where
partnering with lawyers from the get-go helps because they can do discovery of like whether there are similar patterns that were kind of approved or filed. So that way they can stop you from kind of going into the rabbit hole and spending a lot of time. Plus also they know what to extract out of you and help you file a patent in a way like there’s more kind of chances of getting approved.
Federico Ramallo (38:39) Right, right. Yes, yes, I Yeah, what I’m trying to say is that a good idea will be able to survive the scrutiny and the challenge of any challenge that you could put up to it, right? That makes it a great idea, And basically, you can have an idea and then you shape it based on the technical implementation.
the legal that from the legal team, you’re going to get, adapt AD until it can It’s not an attack, but the challenge and the scrutiny any point of view, then you have a solid idea that that a patent,
Yash Chaturvedi (39:26) I
think there’s a saying where all good ideas are good products until you test and validate and then it becomes a great product.
Federico Ramallo (39:35) changing follow a specific framework when making high-stake, proud decisions?
Yash Chaturvedi (39:42) so.
There are like a few things like in terms of frameworks and I’m probably someday gonna write a book about this but
If you’re, let’s say an aspiring product manager, depending upon where you are in your product journey or like your product career, these frameworks evolve. They also evolve what kind of product are you trying to build, whether it’s kind of a platform product, whether it’s a customer experience product, versus it’s a new kind of business line you’re building for the company or like if you’re kind of working for a startup. So the framework, there’s no static framework as such.
There are various frameworks that are applicable across like how mature you are in your product journey, where you are, what kind of product you’re building, what kind of like product personality you have. So for an inspiring product manager, always say like Amazon’s working backwards is an amazing way to kind of start building products, which is we start with a PR FAQ, which is
you’re outlining like a narrative clarity of like, hey, where do I need to, what is this product? Where do I need to go? What exists today? What is the business case? What does the customer problem looks like? Is it worth solving? So there are a lot of these questions that get answered with Amazon’s working backwards. And it’s not proprietary to Amazon. Anyone can use it. It’s just a great tool that thankfully Jeff Bezos has kind of made public.
where anyone can come who’s aspiring to be a PM or like is a PM that is trying to solve a problem can use. So working great, working backwards is a great product framework for product managers, which provides kind of narrative clarity and also simulation of like, hey, what if I were to solve this, what happens then and kind of what are the other steps we need to do. Then you need to pair that with experimentation framework of like.
A B testing, kind of getting some research or surveys done with your customers based on like what product you’re building. We have this joke at Amazon that if you don’t have a PR, your idea is just a tweet. So, don’t just tweet, get me a PR FAQ at least. That’s what I tell my team and kind of my stakeholders. Don’t just come to me with like, hey,
here is a grand idea in like one word or like one sentence or one paragraph, like at least flesh out a PR FAQ because it’s a great mechanism for product managers to challenge their own ideas and also flesh their own ideas thoroughly. like working backwards is great framework for product managers. But there are various different frameworks, as I mentioned like.
that are applicable across like all these different kind of variables that we deal with day to day.
Federico Ramallo (42:33) Yeah, because otherwise it’s daydreaming, right? If they give you an very small idea that haven’t been documented, haven’t been like put down in writing and it’s a dream, but when they put it in a PR, then it becomes a much more mature idea, right? And that’s the other thing, it becomes a fixed object, right? A fixed target versus if it’s in the air,
then if it’s in my mind and I tell it to you, every time I tell you, it’s going to be different, right? Right.
Yash Chaturvedi (43:04) It’s open to interpretation versus once you start
writing it’s much more like there which is like hey we can all align on this particular line that you have written versus like an abstract idea that you might have a different opinion or an idea about versus me.
Federico Ramallo (43:19) Amazing. So talking about AI beyond ads, when new applications of AI excite you the most, you know, in the near future?
Yash Chaturvedi (43:28) So, agentic AI is, I think, taking off pretty drastically. So, agentic AI products, because it’s just not prediction. You’re not predicting what the viewer or the customer is going to do, but it’s acting across workflows. There are many applications in healthcare education, compliance automation.
from a creative streaming point of view there are like video editing, dubbing applications of AI that are going to come into the mix where a bunch of people or streaming services have already started experimenting when you’re watching. There’s a lot of craze these days with Korean dramas, K-pop dramas and K-pop songs. So a lot of times the lip-sync of the character
Although the subtitles and the audio are kind of dubbed in various languages, the lips still move in the Korean language. So there are a lot of like AI automation that is coming in where even your lips of the characters can be custom to the language that it is dubbed in. So there are a lot of like these applications in the streaming space, a lot of healthcare. I think…
beyond kind of ads and beyond kind of good customer experience, AI is going to be a co-pilot that reduces time to decisions or like kind of brings in the efficiency across these, like across our day-to-day kind of jobs. Not just as a PM, but like it’s across like, how do I streamline?
my day to day and bring in efficiencies where I’m doing a lot of repetitive tasks or things that can be kind of that exist somewhere but it’s just taking me time to kind of go discover. Recently I was experimenting with Comet 2 browser. It’s an amazing kind of browser which is if I’m like there was a task that I have to do repetitively it takes me two hours every week.
And I just went in into Comet and said, Hey, this is a task that I need to do. I’m doing this weekly. Can you help automate? And it kind of just ran like just based on this prompt, it did the task and created a procedure that is repeated every week. Yeah. So AI is for me is more co-pilot for lack of a better word, but like it’s going to streamline and bring efficiencies in everyone’s life. So it’s not just
Federico Ramallo (45:43) wow.
Yash Chaturvedi (45:56) the application when it comes to like various organization, various departments, but it’s across like as a human kind of a tool that humans can utilize and bring efficiencies in their kind of day-to-day life. I’m more excited about kind of that application of AI.
Federico Ramallo (46:17) Amazing. Amazing.
Awesome, so we’re running out of time, but first I truly appreciate, Yash, you joining us today. I think we learned a lot about patents, about ads, product placement, your experience as product managers. It’s been amazing. So before we wrap it up, do you have any final remarks that you want to share with our audience?
Yash Chaturvedi (46:41) Yeah, so I’ll tell all the product managers that are listening in that don’t be scared of AI; that AI is gonna take your job. Be scared of the PM who knows how to use AI better than you. So that’s kind of my parting words. So AI adoption is very critical for PMs, whether kind of aspiring PMs or PMs who are mature. AI adoption.
If not done right, then like, then you should be scared.
Federico Ramallo (47:08) Thank you, Yash, for joining us today.
Yash Chaturvedi (47:11) I appreciate it. Thanks a lot, Federico.