How to build product taste and become a top 1% builder, with Ravi Mehta

https://youtu.be/cGT_r4Z__EE

---

Ravi on LinkedIn: https://www.linkedin.com/in/ravimehta/

Ravi’s website: https://www.ravi-mehta.com/

Top 1% Builder Substack: https://blog.ravi-mehta.com/

Ravi’s free workshop: https://luma.com/9x2hnhd0

Newsletter: https://intelligentartifice.kit.com

FULL TRANSCRIPT BELOW

Shamanth: Welcome to Intelligent Artifice, a part of the Mobile UA show.

Shamanth: Intelligent Artifice is a deep dive on how AI is transforming growth and performance marketing.

Shamanth: Since 2018, over 250 plus episodes, we’ve been deconstructing how the best growth marketers actually win.

Shamanth: And every week, I sit down with one operator shaping the future of growth in an AI world.

Shamanth: My guest today is Ravi Mehta. Ravi is a senior product executive who has led product organizations at Tripadvisor, Facebook and Tinder, where he was Chief Product Officer. He was also Executive in Residence at Reforge, where he co-created the AI Strategy course. He now advises companies on product strategy and writes the Top 1% Builders Substack.

Shamanth: I thought this would make for a great episode because AI has made building very cheap, and every team now builds much more than it can ship. Ravi’s view is that the scarce skill is no longer making things, it’s deciding what is worth shipping. In today’s episode, Ravi explains why AI is not going to end slop, it will only make it prettier, and why taste and judgment are the antidote.

Shamanth: He shares how he built his own product in a few hours and then spent six week on craft and refining. He talks about why the best AI teams should really throw away so much of what they build, and why fast companies minimize latency instead of maximizing velocity. There’s a lot to think about. There’s a lot to chew about and learn in this episode.

Shamanth: Let’s get right into it.

Shamanth: I’m excited to welcome Ravi Mehta to Intelligent Artifice. Ravi, welcome to the show.

Ravi: Thank you so much for having me. I’m excited to be here.

Shamanth: Yeah, Ravi, I’m thrilled to have you also just because you’ve, I’ve looked up to a lot of your work and writing for a very long time. And as I was telling you before we hit record, I’m excited for your workshop tomorrow and I don’t think we want to, you know, open the curtains on what’s coming, but I’m excited for that tomorrow.

Ravi: Nice.

Shamanth: Yeah, and you’ve done, you know, you’ve worked on leading products. Now you advise a number of fairly scaled products just now as well. You lead, you’ve led Reforge’s courses and you know, again, you’re all round all star. So I’m excited to have you here.

Shamanth: We can start by talking about, you know, some of the work you did for Tinder, where I know you had unlock a lot of growth.

Shamanth: You’ve written about this that, you know, at Tinder, what you were really reaching for was a price personalized for every customer.

Shamanth: Is that sort of ideal, platonic ideal if you will, is that closer with the help of AI today than it was before?

Ravi: I think it is. I think if you think about any business, whether it’s B2B or B2C, there’s a certain amount of customer value that you create and there’s a certain amount of that value that you get to capture as a business in terms of revenue.

Ravi: This is like a classic economics concept.

Ravi: You can never charge more of a price to a customer than the value that they get.

Ravi: They just won’t buy. And so there’s always some surplus value that you’re delivering to the customer where, you know, the price is lower than the value that you’re creating for them.

Ravi: I think the goal with personalization is to be able to get those two things, the value that you create and the value that you capture to be as close as possible.

Ravi: The, and that’s referred to as price discrimination.

Ravi: The least, effective way to do that is with a flat price for everyone.

Ravi: And so if you charge everyone the same thing regardless of the value that they get, you’re going to have certain people who won’t buy because that price is too high for them and you’re going to have certain people that will buy because, you know, that price is less than the value that they get from the product.

Ravi: That model has some advantages, which is that it is simple and that people perceive it as fair.

Ravi: But you can get to a point where you can provide different products at different prices and capture more of that value.

Ravi: And I think from an app monetization standpoint and even from a SaaS standpoint, that is the ideal.

Ravi: And so structurally, the way that Tinder does that and many companies do that is that there’s multiple different subscription tiers.

Ravi: So there’s Tinder Plus, there’s Tinder Gold, there’s Tinder Platinum, each at a different price point.

Ravi: Those tiers are aimed at, different customers who have different willingness to pay.

Ravi: And then that gets you to sort of a broad brush stroke, better match between the value that you’re creating and the price that you’re charging.

Ravi: And then in addition, there are, microtransaction products, which are consumables that you can buy on top of your subscriptions.

Ravi: For Tinder, those are boosts and super likes.

Ravi: There’s other products as well.

Ravi: And so that allows a person to purchase more, based on what their needs are.

Ravi: This is a structure that a lot of games actually use and it’s very effective in terms of getting to that perfect, more perfect price discrimination.

Ravi: And that’s not necessarily a bad thing.

Ravi: I think, you know, people tend to think if you charge someone more, it’s bad because, you know, you’re charging them more.

Ravi: Nobody wants to pay a higher price.

Ravi: But some people actually do want to pay a higher price to get more value.

Ravi: And sometimes the price that you pay reflects the value that you get.

Ravi: There’s a concept called Veblen goods where the more you pay for something, the more you value it.

Ravi: And this is often what you see with luxury goods like Ferrari and Rolex and other things.

Ravi: And so to your question, your original question around AI, I actually do think that AI makes it easier to determine that perfect price discrimination so that people are paying what they value within the product because you can use algorithms, in order to do targeting and also, to find the right moment for the right offer for a user.

Ravi: And so AI is opening up these possibilities to have a more complex, and a more effective and lucrative pricing model, but with the tension that you want to do it in a way that feels fair.

Ravi: There’s a classic example of, Coca-Cola having vending machines that change their price based on the temperature.

Ravi: And so, you know, the price of Coca-Cola would go up, when it was really hot and go down when it was less hot.

Ravi: People really reacted poorly to that.

Ravi: So price discrimination is something that you need to do with the underlying consumer psychology as part of the equation.

Shamanth: Yeah, and I would imagine, you know, with Coca-Cola, the problem was it was not perceived to be fair.

Shamanth: Whereas in Tinder, you know, if you are charging people and providing value that’s commensurate to that, I would imagine it’s a different thing because they would think, hey, I’m actually getting more out of this, because I’m paying more.

Shamanth: I’m assuming that’s how you thought about it, but please correct me if that’s not the right characterization.

Ravi: That’s correct.

Ravi: And there’s a couple of elements there.

Ravi: You know, for Tinder, the use case is, I want to meet people and I’m willing to pay more to meet people and I understand that if I’m taking a faster or more effective path to meeting people that might cost me more versus, you know, the Coca-Cola case is I’m buying a Coke one day, you’re buying a Coke a different day.

Ravi: It feels unfair for us to pay different prices.

Ravi: It’s interesting.

Ravi: In certain markets, price discrimination is something that is okay.

Ravi: You know, you can see that in the airline ticket market or in the hotel market, for example, but in other markets, you know, people just view it as unfair.

Shamanth: Yeah.

Shamanth: And I think the other example that comes to my mind is Uber where they had, you know, surge pricing and that had a bit of a backlash.

Shamanth: So that makes sense.

Shamanth: And, you know, just continuing to pull on the thread about your work with Tinder, so there were network effects and, you know, the more people there were on the platform, the more powerful the network was, the more valuable it was for everybody.

Shamanth: For AI products, you know, what are the typical moats?

Shamanth: And obviously, I imagine this depends on what kind of AI product it is.

Shamanth: So a social product, probably it’s the network effect, but how would you think about the moat or the competitive advantage for AI products as distinct from what you saw in Tinder?

Ravi: Yeah, I think the moats are really shifting and sometimes in unintuitive ways.

Ravi: A few years ago, maybe almost two years ago now, Brian Balfour and I released a course, for Reforge called AI strategy.

Ravi: And the reason we released that course is because there’s lots of people focused on how do you build AI products, how do you use AI products, but the question that we felt is really important and wasn’t getting enough of an answer is today is the most intense competitive environment that I’ve seen in the history of tech.

Ravi: Things are moving really quickly.

Ravi: We’re in the middle of a platform shift.

Ravi: That typically creates a lot of blue ocean opportunities for new startups to emerge.

Ravi: But instead, we’re seeing that there’s a lot of red ocean here.

Ravi: There’s a lot of markets that are intensely competitive.

Ravi: There’s the big frontier AI labs like OpenAI and Anthropic that are providing the technology that we’re using to build our products, but also competing with us for the most valuable use cases.

Ravi: There’s, you know, big companies like Microsoft and Amazon and Apple, who understand that in a platform shift, it can threaten their established position.

Ravi: They’re moving quickly to fortify their position.

Ravi: There’s VCs that are investing a significant amount of money in startups.

Ravi: And so there’s a lot of competition.

Ravi: The other interesting thing is that AI is the most horizontal technology that we’ve ever seen, and that’s led to a single product competing equally well in online therapy as data analytics, which is ChatGPT.

Ravi: Both really common use cases on ChatGPT.

Ravi: Never in the history were, these two very different, market, very different types of products catering to very different markets competing with the same competitor.

Ravi: And so that’s creates a completely different landscape in terms of how you think about moats.

Ravi: And one of the moats that’s evaporated very quickly is any sort of moat around content.

Ravi: Stack Overflow, Quora, even Reddit, Tripadvisor, Yelp, Chegg, these were all companies that had built up these moats around content where they had a flywheel of users coming in, generating content, then they distribute that content through SEO, more users come in, more people generate content.

Ravi: That created a very durable flywheel, but what also happened was all that content was crawlable, and now that content has been trained into the models.

Ravi: And so rather than going to

Ravi: Stack Overflow to get a question answered, you can just go to Claude to get your question answered or even better, you can just tell Claude what you need to get done and it’ll write the code for you rather than you having to, you know, write the code yourself and ask a hard question if you hit a roadblock within that process.

Ravi: And so I think today the moats that people need to focus on have shifted.

Ravi: The biggest moat is around something related to people, such as contributors. It might be hosts on Airbnb or drivers on Uber.

Ravi: If the advantage is actually the relationship with the person at the other end, not the content that they create, then that is a really interesting moat with an interesting network effect on it.

Ravi: There are still some data and content moats, but only for data and content that is proprietary to you.

Ravi: So by definition, it cannot be crawlable, it cannot be integrated into the training sets for these models.

Ravi: Brand is a very elusive moat, but it is actually a moat. When you have brand power, when you have trust with the customer, that is something that is hard to replicate.

Ravi: High switching costs is a moat. If it’s hard to move from your product to another product, that sustains you.

Ravi: Habits built around your product, which is part of these switching costs, sustains you.

Ravi: Really good unit economics, so economies of scale and scope is a moat.

Ravi: Google has an inherent moat around search because they have the ultimate way to monetize search.

Ravi: So even though there’s some competition coming in from ChatGPT and Perplexity, they are in you know, a relatively good position to be able to defend that.

Ravi: So, super important for companies to think deeply about their durable competitive advantage, advantages, which advantages sustain in the face of competition from AI native products, and which advantages might erode or evaporate in the face of those products.

Shamanth: Yeah, and I think that’s quite a big list, right? So which

Shamanth: Yeah. You know, is a good argument against

Shamanth: AI is destroying every industry and they’re all going to be out of jobs.

Shamanth: So there are very many different ways to have a defensible advantage as a company if you are one.

Shamanth: And of course, the first one you mentioned was really people, which I again goes back to what you spoke about with Tinder, which is people was the thing, right?

Shamanth: People is a super set of network effects and I imagine that’s what you were referencing.

Shamanth: Yeah, you know, again, coming to working with AI, you’ve written that, you know, taste is a skill.

Shamanth: And slop happens when, you know, someone’s just not, someone doesn’t really exercise that taste.

Shamanth: And I know you’re going to talk about this tomorrow in your webinar. You know, that’s if I recollect, that’s the habits of top 1% builders.

Shamanth: And how does a team build taste if you’re like, I kind of suck or I’m new to product management or

Shamanth: Yeah. We are all new to AI.

Shamanth: You know, so if someone that’s completely new to AI, that’s been like vibe coding a bunch of stuff, and suddenly finds themselves a product manager without being formally designated as such.

Shamanth: How should they think about building good taste?

Shamanth: And also how defensible is that because what if Claude Design just swallows your taste tomorrow?

Shamanth: Right? Yeah.

Ravi: Yeah, I think everyone is talking about taste and judgment and why it’s important.

Ravi: And if we zoom out, it’s important to ask, well, why is that become so important? Why is that where things have shifted?

Ravi: And the reason is that AI is very powerful, but AI operates on context.

Ravi: It can, you can give it a little context or you can give it a lot of context.

Ravi: And without the context, AI is an inert technology. It doesn’t do anything.

Ravi: And with a little bit of context, AI actually makes most of the decisions for you.

Ravi: But it’s going to make decisions based on the average of everything that it’s seen.

Ravi: And slop is really the accumulation of unexamined defaults.

Ravi: So slop is what comes out of the AI when you don’t give it an alternative.

Ravi: And my argument is that even as AI gets better and better, what’s going to happen is it’s not going to eliminate slop, it’s just going to create prettier and prettier slop.

Ravi: Because the reason is that slop and is actually something that humans read because they have certain expectations of what is good and what is great.

Ravi: And so as the AI gets better, we’re going to get used to the quality of the output from the AI.

Ravi: And what was magical just a couple of years ago is going to read as slop if it’s something that is just completely saturated in the environment.

Ravi: And so the way that you create exceptional products is you use judgment to set context working with AI as well as doing things yourself to build something that is unexpected.

Ravi: And when you build something that is unexpected, it might be unexpectedly great, it might be completely different in certain ways.

Ravi: People read that not as slop, not as an average, but as something that deserves their attention, as something that, you know, Seth Godin says is remarkable.

Ravi: And so my argument is that slop will always exist and that the antidote to slop is taste and judgment.

Ravi: And the way that you implement your taste and your judgment is that you use it to set better context for what you want to create so that what you’re creating with the AI is not what the AI has decided to build, but is what you have decided to build.

Ravi: And in that way, I think you can develop taste and ultimately what taste is it is your opinion about what is good.

Ravi: And so there’s no such thing as good taste or bad taste.

Ravi: There’s just taste that resonates with an audience of one, yourself, versus taste that resonates with everyone.

Ravi: You know, which is commercial taste.

Ravi: And so if you’re making a commercial product, you want something to resonate with thousands or millions of people, then you need to understand the taste of the audience and you need to align your taste with that audience’s taste.

Ravi: But if you’re creating something just for yourself, you’re creating a personal piece of software, you’re creating, you know, you’re a woodworker, you’re creating a table that’s going to sit in your living room or a painting or a photograph that’s going to sit on your wall, then the only taste that matters is your own taste.

Shamanth: For sure. And the article, the one thing I would also call out is, I think personal taste is perhaps a prerequisite for catering to universal taste, but feel free to disagree if you will.

Shamanth: Right? Because if you kind of know what feels good to you, chances are you are on your way to building something good for other people.

Shamanth: Yeah.

Ravi: And you know, I totally agree with that. I think about taste as kind of like a conversation between you and the world and understanding what do you like, what do other people like and rationalizing why those things are the case.

Ravi: You know, if a song is popular, you know, you could say, I don’t like this song. That’s totally fine.

Ravi: But if millions of people enjoy it, then part of developing your taste is understanding, well, what is it about that piece of content that people really enjoy and do I

Ravi: Yeah. Once you understand that, you can decide, oh, now I understand why people like it. I’m going to use that to create something that those people also like, or you can say, no, you know, I don’t want to cater to that audience. I want to create something really different.

Shamanth: Yeah. Yeah, I like that characterization, right? And it’s from what you’re saying, it’s like, it’s taste is being able to crystallize why something is good, if I understand correctly.

Shamanth: You know, and as someone that’s vibe coded lately for the last two years or so, I think I can resonate with that because I’ve been like, my early prototypes were terrible and now I’m like, with the benefit of experience, I can do a better job of articulating why I like a certain minimal interface or why I want to do something.

Shamanth: So I think that characterization makes so much sense.

Shamanth: Again, staying with yeah, go ahead, please. Yeah.

Ravi: And I think to build on that, it’s helpful to distinguish between taste and craft.

Ravi: And so the way we could define that is taste is your ability to have an opinion about what is good.

Ravi: Craft is your ability to make something that actually fulfills the bar of your taste.

Ravi: And so think about a restaurant critic. A restaurant critic might go to a Michelin-star restaurant.

Ravi: That critic is someone that people look to determine is this restaurant good or bad, but they could never, you know, sit in the kitchen and make a meal.

Ravi: So that’s a distinction between craft and taste. The restaurant critic has taste, but they don’t necessarily have craft.

Ravi: And so as creators, we want to have both. We want to have both taste and craft.

Ravi: And so we want to develop our taste, we want to develop our craft. And ultimately their goal there is to give us the power to create the things that we think are good.

Shamanth: Yeah. You know, with AI, there’s a lot of products that have been adding AI features.

Shamanth: You know, I’ve seen Slack, I’ve seen project management tools that have added AI features, Notion’s added AI features.

Shamanth: And certainly a lot of the smaller companies have as well, which and my perception is that there’s a little bit of a pushback around, okay, we actually don’t want as much AI in everything.

Shamanth: How do you think about where an AI feature is essential and where it’s probably putting lipstick on a pig?

Ravi: Yeah. I mean, I think it turns out customers don’t want AI in any features, right? The customers don’t think

Ravi: In those terms. If anything that they associate AI with data centers and existential risk to humanity and climate change impacts and it’s really expensive and it’s really slow. So I think if you throw the words AI especially on a consumer feature, it’s actually going to create it’s going to repel users rather than attract them.

Ravi: I think one of the best examples of how to use AI is Descript, which is the platform that we’re using right now to record this.

Ravi: You know, they have a whole palette of AI powered features. Almost everything that they do is AI powered, but they frame those features in terms of the problems that they’re solving for creators. And what do creators ultimately want?

Ravi: They want to with as little time as possible, look and sound their best.

Ravi: And so one of their features is eye contact and eye contact solves a pretty common issue for creators, which is you might be reading from a script and then you look down at the script, you look back up at the camera.

Ravi: And that creates a really hard ability to edit the footage because then your eyes are darting around when you’re splicing the footage and they use computer vision to be able to identify where a person’s eyes are and actually make it so that they lock with the camera.

Ravi: That’s an AI feature. It’s actually using very advanced AI capabilities, but it’s solving a very human need, which is, you know, I want to maintain eye contact with my audience.

Ravi: And so always start from what is the customer problem or the unmet customer need and how can AI allow me to solve that in a way that was not previously possible.

Ravi: And then you can create AI features that are actually valuable. If you instead you start with the technology and ask what’s possible, then in that case, you know, you’ll end up with a solution looking for a problem.

Shamanth: Yeah. And I think the other dynamic which I’m noticing is even if you add AI features, you’re you have the token cost, which I’m noticing a lot of companies haven’t accounted for in their original planning.

Shamanth: And they’re like, this is a serious line item in our P&L.

Shamanth: You know, versus any other feature which technically would have been free pre AI.

Shamanth: And I’m just curious how you think about that additional twist in the dynamics.

Ravi: I think that goes back to the first thing that we were talking about is what is the distinction between value that you create for the customer and value that you capture.

Ravi: You know, the problem with a lot of these AI costs is that you know, people are spending money on the tokens because from an organizational standpoint, they feel like they have to do that to stay ahead of the curve, but they’re not seeing the ROI for their business right now.

Ravi: And so we have, you know, I’m sure the AI industry is going to be fine.

Ravi: But we have this really concerning situation where people are paying more for the AI than the value that they’re getting out of it right now, which either means that, you know, the value’s got to increase, or the prices have to come down.

Ravi: But at the same time, you know, a lot of the token prices are being subsidized right now, by invested capital.

Ravi: And so, we are in, you know, what a lot of AI skeptics, characterize as a bubble.

Ravi: I don’t actually think it’s a bubble. You know, if you think about everything that’s possible today versus a few years ago, there is a massive amount of value being created, but we need to track very carefully, how that value is priced in, and where it makes sense and where it doesn’t.

Shamanth: Yeah.

Shamanth: And, you know, I would imagine part of the waste, if you will, that you’re referencing is because there’s just so much output now.

Shamanth: So I’m curious how you would think about, you know, when you’re working with a team or when you’re noticing a team or advising a team, the output’s clearly gone up, the quantity of PRDs they’re putting out, the quantity of prototypes they’re putting out.

Shamanth: How do you think about like, look, AI is actually enabling diverse new novel ideas versus there’s just a lot of slop that nobody has time to read.

Shamanth: How do you make that distinction?

Ravi: I think this is something that we’re seeing with a lot of teams right now.

Ravi: You know, teams, whether they know it or not, are climbing an AI maturity curve and the first part of that maturity curve is adoption.

Ravi: And once you started to adopt and adopt well, everything moves faster and then there’s just a lot more created.

Ravi: There’s a lot more specs and prototypes.

Ravi: And so a lot of teams right now are drowning in things.

Ravi: They’re drowning in unshipped features, they’re drowning in specs and prototypes.

Ravi: There’s PRDs nobody’s going to read, there’s status reports nobody’s going to read.

Ravi: Everyone’s getting daily updates on their daily updates and it’s just like a ton of information, but not a lot of insight.

Ravi: And so I think the next phase for teams is curation.

Ravi: It’s curation of the context that really matters within the org.

Ravi: And for product development teams, it’s curation of what features do we actually want to ship?

Ravi: What’s worthy of our customers?

Ravi: And so I recently did a workshop for leaders in Berlin and I talked about how the product life cycle is changing.

Ravi: And one of the things I said is one of the critical new phases of the product life cycle is a curation step where we look at what we’ve built and we decide whether or not we want to ship it.

Ravi: And actually the most effective AI teams should be not shipping a lot of what they build and only shipping the best of it.

Ravi: And so it’s less about shipping more and more about shipping better.

Ravi: That sort of curation is something like Apple, for example, has had the luxury of doing for many decades.

Ravi: They don’t ship all of the products that they build.

Ravi: They only ship them when they’re ready.

Ravi: Because building software has historically been expensive, other companies have not had that luxury, but now we not only have the luxury, it is a necessity.

Ravi: You cannot ship everything that you build to the customer without overwhelming them.

Ravi: And if you do, your product will suffer.

Ravi: Me saying that was very controversial because a lot of product leaders are like, there’s sunk costs there, why would we waste time building something if we’re not going to ship it?

Ravi: And so let’s just ship it all, we’ll run experiments and then we’ll let the AB tests decide.

Ravi: And I think there’s lots of problems with that.

Ravi: You know, not the least of which is you just don’t have the traffic to be able to run reliable AB tests on everything that is being created.

Ravi: And two, we all have used products which are the accumulation of lots of AB tests and they feel like Frankensteins.

Ravi: And it’s much better to have a thoughtful intuition about what is customer worthy, and what, you know, makes sense to have in our product than to ship things to the customer and let them decide.

Ravi: You know, every decision we’re not making in the building is a decision our customers need to make and customers don’t want a product that’s constantly changing on them where they can’t depend on the functionality and where, you know, it’s an accumulation of winning AB tests rather than a thoughtful design.

Shamanth: Yeah, and I imagine that’s the hardest thing because you have to restrain yourself and like you pointed out in that workshop, people have sunk cost around, look, I made so much, do we have to throw it away?

Shamanth: But you’re right and that’s what requires a lot of the discipline in figuring out what you’re going to ship.

Shamanth: You know, and I think the Apple analogy also makes sense because I think historically, at least on our team and when we worked with arts, we had, we even pre AI, we would have designers come up with explorations and say, let’s come with five different directions to explore in, then we will pick one.

Shamanth: And from what you’re saying, that’s the kind of mindset you want to look at AI with where you’re like using it to explore, but using your taste and intuition and experience to pick out which of those explorations actually makes sense.

Ravi: Absolutely. And I think there’s an interesting analogy to artists and creators.

Ravi: You know, a movie director, there’s a lot more footage in the garbage than there is on the reel for the final movie.

Ravi: For a photographer, there’s a lot more photos that never see the light of day.

Ravi: For musicians, the best musicians, they’re creating a lot more than they ever put on an album.

Ravi: And so I think it’s very similar for us.

Shamanth: Yeah.

Shamanth: Yeah.

Shamanth: And, you know, you’re working on your own product, which is a bit of, you know, a bit of a shift from working on scaled products in within companies.

Shamanth: And you mentioned when we were preparing for this call that you made the launch trailer completely in Claude Code, no editor.

Shamanth: So, you know, you did this in Claude Code.

Shamanth: You didn’t use this script which we talked about or any of a video editing tools.

Shamanth: So, are we moving to a world where people use products without the product interface?

Shamanth: And what place do interfaces have in this world, if any at all?

Ravi: Yeah, it’s a really interesting question.

Ravi: I’m working on a new product, it’s called Copy Paste.

Ravi: And it’s an interesting example of this, idea of taste and curation.

Ravi: I one shotted the product when Fable came out.

Ravi: So, in about four or five hours, I had a working version of it.

Ravi: It’s essentially integrates AI into your clipboard.

Ravi: So when you copy something, you can run AI actions on top of it.

Ravi: It’s part of me thinking about like what are the different interaction models.

Ravi: And one of the key things that AI is great at is it’s great at manipulating information.

Ravi: And where does most of the information, transit happen on our computers?

Ravi: It happens on our clipboard.

Ravi: And so, the idea is, you know, how do we think about, an AI native clipboard?

Ravi: So I built the product initially, in that one shot, and then it took me about six weeks after that to get to a point where I actually felt like the product had the level of craft that I really wanted it to have in terms of good user experience, working the way that we want, nice

Ravi: And fast, feels like crisp in terms of all the animations. I’m getting ready to launch it, so I’m going to launch it in the next few weeks and of course when you launch something you need a launch trailer. In the past, a launch trailer would take several weeks if not months, it would cost tens of thousands of dollars and it would be very hard to actually create or iterate on.

Ravi: I initially thought about doing it in Screen Studio and so Screen Studio is a screen recorder. You can record using the product and then I would clip things. And then maybe use something like Descript. But instead I created the launch trailer with a set of Claude code skills and a an open source project called Hyperframes, which is from HeyGen.

Ravi: And in addition to using that allows you to create video, I also just turned on computer use and had Claude not only you know, build the video, but also actually record all of the footage on my computer that it needed. So I didn’t do I didn’t record a single piece of footage.

Ravi: It got all the sound effects. I just gave it some background music. And so I was able to, you know, it did take me about half a day or so, but go from no trailer whatsoever to a finished 90-second launch trailer with background music, beautiful animation, something that feels highly polished entirely in Claude code by prompting it and having it do the work to get the footage, to integrate the sound effects, to do the clips.

Ravi: And what was interesting is that it both created the underlying footage by doing screen recording, but it also was able to use motion graphics to highlight different parts of the product, to make it all feel really tailored, something that feels really professional that was done, you know, much faster, at much lower expense, but still to the same level of craft as I would expect a really nicely done launch trailer.

Shamanth: Yeah.

Shamanth: Yeah. No, I think that’s so impressive.

Shamanth: And I think what also stood out to me from what you said is you one shotted it, but you also took six weeks to polish it. And I think that’s also our experience, my team’s experience that a lot of creative work, videos and stuff, right? Because I think one shotting is easy to go from 0 to 80%.

Shamanth: And from in the framing that you’ve mentioned, the remaining 20% is taste and really and I would also imagine it’s also fixing things that shouldn’t be shipped in the first place. And without that, the product just can’t ship even if it’s 80% of the way there.

Shamanth: You know, yeah.

Shamanth: And switching gears, you know, and again, just staying with, you know, your own product,

Shamanth: How did the approach to your own product differ from some of the things you’ve advised scaled companies to do?

Shamanth: And what part of your own advice has been hardest to follow?

Ravi: Yeah,

Ravi: You know, to some extent, and I know that there’s always complexity to deal with, but I think the companies that get the most from AI, even if they were big, they are big, they’ll start acting more like startups. And so, Yeah. Within a large company, things take a long amount of time, not always because the work took a lot of time, but because they have communication processes and approval processes, and quarterly planning and resourcing and all of these different things that add time.

Ravi: And so, one of the ways I like to think about speed is important. We need to move fast, especially in this environment, but most people misunderstand speed. They think about it in terms of velocity. They think about it in terms of the number of things that are getting done.

Ravi: But actually fast companies are not maximizing their velocity, they’re minimizing their latency. They are decreasing the time from when they have an idea to when they can actually get something in front of customers to validate that idea.

Ravi: And so the litmus test I like to use for customers is if you know that the name of a button, the copy on a button is the wrong thing and you want to change it, how long does it change take you to change the copy on the most important button on your site? So that might be the buy button or the save to trip button or something like that.

Ravi: Some companies that’ll take weeks. You know, the actual time to change the text is, you know, moments, but it’ll take weeks to be able to approve that and ship that and get that in front of customers.

Ravi: Other companies can do it later that day or the next day. And that’s not necessarily a function of scale, that’s a function of the internal processes.

Ravi: And so I think big companies need to minimize their latency.

Ravi: What tends to happen in large companies is big companies always get more done. They have a higher velocity than smaller companies. You know, 30,000 people are going to do more work and have more output than 3,000 people or 300 people or three people.

Ravi: But they do that work with a high degree of latency. And it’s a little bit like driving a car. When you are going really fast, your turning radius increases. You can’t be as agile. You can be much more agile when you are moving slower, when you don’t have so much velocity.

Ravi: And so I think startups have always had the benefit of being more agile, of having lower latency and big companies need to do the same. And they need to be comfortable with we can have an idea, we can throw that idea into Claude code, we can get an output and we can ship it.

Ravi: Yeah, there’s some risk with that. You know, there might be a mistake, but there also might be a mistake if a human developer does that. If you, I’ve talked to multiple companies now who are using all these AI coding tools and not moving much faster because the code’s getting done faster, but it’s working within the same process as before.

Ravi: And that just means, you know, they’re spending all this money on tokens but not actually getting any sort of advantage.

Shamanth: Right. You know, if you’re producing code faster and there’s still an approval layer which hasn’t moved at all, it doesn’t matter how much code you produce. That’s so interesting. There’s also

Ravi: And there’s a shocking amount of companies that are in that bucket of, yes, we’re using all these AI tools. Nah, things are not moving that much faster.

Shamanth: Yeah. Yeah. I’m reminded of this book called The Goal. I don’t know if you’ve heard of it.

Ravi: Yes, yeah. Absolutely.

Shamanth: This is the exact principle he talks about, right? Because you just produce so much and the work in process builds up, you just don’t have enough throughput, which is kind of crazy. And I’ve noticed and found a lot of parallels with the principles of the goal with a lot of digital work, knowledge work, which I find fascinating.

Shamanth: You know, and also just talking about organizational dynamics, you know, on our prep call, you said, you know, let’s just say five years ago, a lot of CEOs would have picked moving three times faster with the same team versus moving at the same speed with a third of the people. It’s kind of related to the organizational dynamics we’re talking about.

Shamanth: Now, there are companies that are cutting people to spend on AI.

Shamanth: So, you know, let’s assume we are like, hey, let’s keep the same head count, let’s get the speed.

Shamanth: There is excess capacity, again, in the language of manufacturing the goal, there is excess capacity.

Shamanth: Where does that go? How should leaders think about it if they’re like, look, we don’t want to cut staff, but we want to hopefully move twice as fast with AI.

Ravi: Yeah, you know, I think there’s this interesting, there’s sort of an assumption, especially with public companies that leaders have, you know, this deep causal understanding of their business and if they just had more resources, they would be able to grow faster.

Ravi: And we’re seeing that’s not true. You know, everyone has more resources, everyone can move faster, but companies aren’t growing faster, not because AI is not having an impact, but because they don’t know where to apply all of that productivity because they don’t have a causal understanding of what drives growth for their business.

Ravi: And so for companies that are in that boat, I think the important thing is not producing more, but understanding more. Talking to your customers more, maybe creating prototypes. This goes back to just creating things that you’re going to throw away.

Ravi: Create a bunch of prototypes, get them in front of customers, understand what your customers really need, what that means for your strategy, what that means for where you can grow. Build up that understanding, which is slow, right? That process is not as accelerated by AI. It’s much more gratifying to say, hey, we shipped 10 features instead of three with AI than it is to say, actually we’ve we haven’t shipped more features, but we’ve talked more to customers and we’ve gotten to a better customer understanding and now that’s going to enable a more sound strategy.

Ravi: And so, you know, ultimately, I think companies are trading talent for tokens. That to me is a mistake. You know, that’s something that Wall Street is rewarding because they doubly so, right? When you do layoffs, your stock price goes up anyway because your cost basis has gone down. And now, you know, if you’re doing it in the name of AI, then you get the double benefit of, oh, strategically they must know what they’re doing.

Ravi: But that’s Wall Street rewarding short-term behavior and the long-term behavior here is really now that we can move a lot faster, do we have enough of an understanding of our business to know where to apply that energy?

Shamanth: Yeah, and that ties to what I’m hearing the overarching theme from you, which is really quality and discernment is so much critical than speed in and of itself, or just putting out sheer volume.

Shamanth: Of features. And if I had to distill everything you said, really

Shamanth: Having that level of understanding of really what moves the ball, what moves the needle,

Shamanth: What’s good taste really makes the biggest difference rather than just

Shamanth: Token expenditure.

Shamanth: Yeah?

Ravi: Absolutely. And I would add to that like, you know, I love AB testing. I’ve run thousands of AB tests over my career.

Ravi: But what is an AB test really saying? It’s saying we don’t know what customers want.

Ravi: We’re going to ship a couple of different versions and we’re going to let the customers

Ravi: Decide. And so AB testing is often done from a lack of understanding of

Ravi: Customers. And I think it goes back to what you were just saying, which is the taste, the discernment,

Ravi: The judgment. If you understand your customers, if you know what they want, then it’s going to give you a lot more conviction in what you need to ship for them

Ravi: And you won’t have to fall back on, well, we’ll AB test it and see.

Shamanth: Yeah. Yeah. And you need the taste to figure out what to actually test and they be test.

Shamanth: Versus just putting things together at random, right? Yeah. Yeah.

Shamanth: Excellent, Ravi. I think that’s been incredibly insightful.

Shamanth: Also because I think we are in an age where

Shamanth: Token maxing is a thing. You know, if you go on any of the socials,

Shamanth: You’ll see a lot of people post screenshots of how many tokens they’ve used.

Shamanth: And I think this is a very good and useful reminder for everybody, myself included,

Shamanth: About

Shamanth: Important as it is to

Shamanth: Use AI,

Shamanth: It’s just not important to take a step back and understand

Shamanth: Why and how and what the hell we’re trying to do with it.

Shamanth: And then I think that’s a very, very useful reminder.

Shamanth: This is perhaps a good place for us to start to wrap up.

Shamanth: Thank you so much for your time. I know you have a very, very busy schedule, so I really appreciate you taking time to talk to me today.

Ravi: Yeah, thanks for having me.

Shamanth: Yeah, as we wrap, could you tell folks more about you and

Shamanth: Everything you do and how they can find out more about you?

Ravi: Yeah, absolutely. So about five years ago, I left full-time product work.

Ravi: I now spend about half my time as a fractional CPO, helping companies navigate this moment,

Ravi: Figure out their product strategy, build their product team. So if you’re interested in that sort of help, please reach out.

Ravi: I also write on Substack. My Substack is called Top 1% Builder.

Ravi: So relaunching that tomorrow with this focus on you know, not just product management, but also, you know, what are the skills that we need for everyone in an organization that is working on product development.

Ravi: What are the product thinking skills that you need to build the best possible products?

Ravi: I think that there’s been this idea that product management is dead, and I would say actually it’s more important than ever before.

Ravi: And you know, product management is just not optional when it comes to delivering great products for your customers.

Ravi: So look for me on Substack and LinkedIn, and reach out if I can be helpful.

Shamanth: Wonderful. And we will link to your Substack and I highly recommend your email emails.

Shamanth: Certainly very, very insightful and inspired a lot of the thinking that led to this recording.

Shamanth: Highly recommend that we will link to your Substack, we will link to your LinkedIn, link to your website.

Shamanth: And by the time this is out, your new Substack will have your

Shamanth: New Substack will have launched. Did you say you’re rebranding the Substack? Did I understand that?

Shamanth: You’re

Ravi: Yeah, rebranding from Ravi on product to top 1% builder.

Ravi: And really focusing, there’s a lot of substacks on like how to use AI.

Ravi: What I’m focusing on is how to build with AI, how to build great products with it.

Shamanth: Excellent. We will also link to your webinar which I hope will be online.

Shamanth: That’s tomorrow. It will be in the past by the time this episode is released.

Shamanth: I would also highly recommend that people check that out. I’m registered for that.

Shamanth: I would highly recommend that. I will link to that assuming it is available for post webinar

Shamanth: Download and review.

Ravi: Amazing. Yeah, I’ll definitely share the link.

Shamanth: Wonderful.

Shamanth: Wonderful, Ravi. This is a great place for us to wrap up. Thank you again for your time.

Ravi: Yeah, thanks for having me.