We’re living through the camera phone moment of software.
When camera phones first came out, suddenly everyone could take photos and videos.
No DSLR. No studio lighting. No training. Just a phone and something to shoot, upload it on platforms like Instagram and YouTube, while that unlocked a flood of creativity, most of what came out was...well, pretty bad. Blurry shots. Harsh lighting. No composition. No story. Because access doesn’t equal excellence.
Then something interesting happened: some artists stood out. It was not because they had better cameras. Everyone had a similar device. They stood out because they had taste.
They understood framing and composition. They trained their eye, they knew what to leave out. They shot 100 terrible photos/ videos before finding one that worked. They built intuition, one tiny judgment call at a time.
That’s exactly where we are with AI and software right now.
Tools got faster but taste got more valuable.
Tools like ChatGPT, V0, Cursor, Claude, Replit, and more, they made building easier and faster than ever.
Now anyone can create something that looks finished. But the real question isn’t “Can you build it?”
It’s: what should you build?
Typing a prompt is easy. Knowing what to type, when to stop, and whether the result is actually good? That’s the hard part. That’s where taste matters more than ever.
Taste is the difference between shipping noise and shipping something people actually want.
It’s how you make something feel personal, not generic.
And more than anything, it’s how you earn trust and attention in a world full of sameness. That’s the hard part and that’s the craft. That’s still on you.
It used to be that if you could code, you had leverage. Now AI writes code to build the bones of an app. It can draft copy. suggest flows.
Taste is judgment in action.
It’s knowing which features to cut, which copy to keep warm, and which animation to slow down. Taste separates a product from a brand.
And it’s earned by:
Building 10 things that flopped
Rebuilding them from scratch
Watching how users actually behave
Obsessing over small details most people skip
Studying products that age well, not just trends.
Asking "Is this necessary?" more often than "Is this cool?"
Taste was the moat and with the AI age, it's still the moat.
You don’t prompt your way into good design decisions. You earn them through practice, pattern recognition, and deliberate thinking just like building muscle.
In this AI age, we don’t need more interfaces. We need clarity. Relevance. Felt experiences for people.
You can build an MVP in a weekend now. But making something people care about that takes a different muscle. And that muscle is creative judgment.
What it looks like in practice?
One example that I worked on is Truflation, a real-time inflation data platform that aggregates independent sources to offer a transparent on-chain alternative to government-reported metrics.
Most economic dashboards feel like spreadsheets for academics. The early version of Truflation wasn’t much different: dense charts, janky filters, and confusing raw data that only made sense for nerds. It was technically functional but it didn’t guide trust or decisions. The MVP did win a prize from Balaji Srinivasan for his challenge to build a transparent real-time inflation dashboard but to win real users, it needed more than that.
Today, you could’ve rebuilt that version in a weekend with AI, Tailwind, and source data. But that’s the point, a good product isn’t just about building fast. It’s about building with judgment.
We didn’t just redesign the interface. We asked hard questions:
What should users see first?
What’s getting in the way of clarity?
What do people actually want to do with this data?
So I worked with the team and simplified with:
Clear, comparative data narratives instead of throwing numbers everywhere
Gave people just a few filters they needed, not dozens of options for smart defaults over infinite config
A design system that scaled across categories, countries, and tools (widget, calculator, etc.)
And most importantly, we made it feel human.
The result was a product people could use to understand inflation and to compare data to make better decisions. The product didn’t feel like something built by a bot. It felt thoughtful. Personal. Inviting. That came from real craft, testing, and care not from prompting an AI. our decisions helped users feel confident, not confused and that’s what helped us scale Truflation into something people trust.
Read the full case study or view Twitter breakdown thread
Now truflation powers: macro investors, economists, policymakers, former FED members, enterprises, and over 100+ protocols. Recognized by coinbase and its founder as an innovation in financial & economic data transparency. Serving as a foundation for inflation-pegged stablecoins and other DeFi solutions.
Taste = Brand
Taste doesn’t just shape what a product does, it defines how it feels.
And that feeling? That’s your brand.
People often talk about “brand” like it’s a skin you layer on after the fact. A logo. A color palette. A font. But the best products know: Every interaction is the brand. Every pixel is part of the brand and sends a message.
But the best products treat every single interaction as part of the brand.
The tone of the error message.
The way the form autofills.
The delay before a loading spinner appears.
These aren’t just UX decisions. They’re expressions of the product’s values.
That’s brand judgment. That’s product taste. That’s what drives distribution.
This isn’t new. We’ve seen taste-driven products win before AI:
Notion - made writing and organizing feel like crafting. There were already hundreds of note-taking and productivity apps but the notion wasn’t just a notes app, it respected whitespace without shouting loudly features on screen. It brought calm into productivity to organize personal or professional life with its flexible structure and people felt it with early adoption by designers, founders, and indie hackers the right taste-makers.
Headspace - made meditation gentle. Meditation apps existed, but headspace used soft visuals, gentle narration, and playful UX to make calm feel accessible. It didn’t just guide you, it invited you. The taste gave it mass appeal.
Linear - made issue tracking feel like design. It wasn’t built for PMs drowning in tickets. It was for fast-moving teams who valued speed, clarity, and taste. Every micro-interaction from keyboard shortcuts to dark mode whispered craftsmanship. It spoke to teams who move fast and care how things feel.
Raycast - turned the Mac command bar into a command center. In a sea of clunky launchers, Raycast felt intentional. Fast, extensible, and beautiful by default. You didn’t just use it, you bragged about it. It made your workflow feel elite.
None of these tools introduced entirely new categories. But they redefined what good felt like within them. Because they were built with taste.
And taste comes from judgment. And judgment comes from doing the work.
The bar Is higher now. And it should be.
AI is only going to make it easier to build. This means the bar to what matters will keep getting higher and that bar is made of taste, clarity, and care.
So yes, use the tools. Speed up the grunt work. Prototype faster. Iterate more. That’s the upside of this moment. But don’t let automation become autopilot.
Because the best products don’t just function. They feel right.
Building software still demands critical thinking, ethical thinking, and system thinking.
And that’s something AI still can’t fake.
Topics along this line that’s worth reading
make something heavy and taste is eating silicon valley by Anu Atluru
article by Julie Zhuo on taste part-3.
Paul Graham on taste.
Practice to studying product like Brian Lovin’s app-dissection.
Build and share ideas like Jordan Singer.
When camera phones first came out, suddenly everyone could take photos and videos.
No DSLR. No studio lighting. No training. Just a phone and something to shoot, upload it on platforms like Instagram and YouTube, while that unlocked a flood of creativity, most of what came out was...well, pretty bad. Blurry shots. Harsh lighting. No composition. No story. Because access doesn’t equal excellence.
Then something interesting happened: some artists stood out. It was not because they had better cameras. Everyone had a similar device. They stood out because they had taste.
They understood framing and composition. They trained their eye, they knew what to leave out. They shot 100 terrible photos/ videos before finding one that worked. They built intuition, one tiny judgment call at a time.
That’s exactly where we are with AI and software right now.
Anyone can build. Tools like ChatGPT, V0, Cursor, Claude, Replit, and more. they’ve opened the gates. The question is no longer "Can you build it?" It’s "should you?"
Tools got faster but taste got more valuable.
Typing a prompt is easy. Knowing what to type, when to stop, and whether what came out is actually good? That’s the hard part and that’s the craft.
It used to be that if you could code, you had leverage. Now AI writes code to build the bones of an app. It can draft copy. Suggest flows.
Similar tools. Some people ship noise. Others ship magic. That’s not luck. That’s taste. But knowing what to build? How it should work? Why it matters?
That’s still on you.
Taste is judgment in action. It’s knowing which features to cut, which copy to keep warm, and which animation to slow down. Taste separates a product from a brand.
And it’s earned by:
Building 10 things that flopped
Rebuilding them from scratch
Watching how users actually behave
Obsessing over small details most people skip
Studying products that age well, not just trends.
Asking "Is this necessary?" more often than "Is this cool?"
Taste was the moat and with the AI age, it's still the moat.
You don’t prompt your way into good design decisions. You earn them through practice, pattern recognition, and deliberate thinking.
In this AI age, we don’t need more interfaces. We need clarity. Relevance. Felt experiences.
You can build an MVP in a weekend now but making something people care about that takes a different muscle. And that muscle is creative judgment.
One example that I worked on is Truflation, a real-time inflation data platform that aggregates independent sources to offer a transparent on-chain alternative to government-reported metrics.
Most economic dashboards feel like spreadsheets for academics. The early version of Truflation wasn’t much different: dense charts, janky filters, and confusing raw data that only made sense for nerds. It was technically functional but it didn’t guide trust or decisions. The MVP did win a prize from Balaji Srinivasan for his challenge to build a transparent real-time inflation dashboard but to win real users, it needed more than that.
Today, you could’ve rebuilt that version in a weekend with AI, Tailwind, and source data. But that’s the point, a good product isn’t just about building fast. It’s about building with judgment.
We didn’t just redesign the interface. We asked hard questions:
What should users see first?
What’s getting in the way of clarity?
What do people actually want to do with this data?
So I worked with the team and simplified with:
Clear, comparative data narratives instead of throwing numbers everywhere
Gave people just a few filters they needed, not dozens of options for smart defaults over infinite config
A design system that scaled across categories, countries, and tools (widget, calculator, etc.)
And most importantly, we made it feel human.
The result was a product people could use to understand inflation and to compare data to make better decisions. The product didn’t feel like something built by a bot. It felt thoughtful. Personal. Inviting. That came from real craft, testing, and care not from prompting an AI. our decisions helped users feel confident, not confused and that’s what helped us scale Truflation into something people trust.
Now truflation powers: macro investors, economists, policymakers, former FED members, enterprises, and over 100+ protocols. Recognized by coinbase and its founder as an innovation in financial & economic data transparency. Serving as a foundation for inflation-pegged stablecoins and other DeFi solutions.
Read the full case study or view Twitter breakdown thread
Read the full case study or view Twitter breakdown thread
The first version looked like most MVP dashboards: raw data, confusing charts, and janky filters. Useful for crypto-native users but confusing for everyone else.
You could’ve rebuilt that UI in a weekend using AI, Tailwind, and data. But that’s the point, taste isn’t about moving pixels and random data. It’s about knowing what not to ship.
The MVP did win a prize from Balaji Srinivasan during his challenge to build a transparent inflation dashboard but to win real users, it needed more than that.
When I came in, the goal was clear: make inflation data feel real-time, global, and personal, not just for researchers, but for protocols, macro investors, individuals, and policymakers.
That’s not prompting. That’s product taste, earned through iteration, collaboration, and saying “no” to most things. Taste is how you move from possible to meaningful.
Every pixel is part of the brand and sends a message.
People often talk about “brand” like it’s a skin you layer on after the fact. A logo. A color palette. A font.
But the best products treat every single interaction as part of the brand.
The tone of the error message.
The way the form autofills.
The delay before a loading spinner appears.
These aren’t just UX decisions. They’re expressions of the product’s values.
A good product works.
A great product feels right.
It feels inevitable. Consistent. Considered.
And in a world where users are flooded with choices, feeling right is what makes them stick.
That’s brand judgment. That’s product taste. That’s what drives distribution.
This isn’t new. We’ve seen taste-driven products win before AI:
Notion - made writing and organizing feel like crafting. There were already hundreds of note-taking and productivity apps but the notion wasn’t just a notes app, it respected whitespace without shouting loudly features on screen. It brought calm into productivity to organize personal or professional life with its flexible structure and people felt it with early adoption by designers, founders, and indie hackers the right taste-makers.
Headspace - made meditation gentle. Meditation apps existed, but headspace used soft visuals, gentle narration, and playful UX to make calm feel accessible. It didn’t just guide you, it invited you. The taste gave it mass appeal.
Linear - made issue tracking feel like design. It wasn’t built for PMs drowning in tickets. It was for fast-moving teams who valued speed, clarity, and taste. Every micro-interaction from keyboard shortcuts to dark mode whispered craftsmanship. It spoke to teams who move fast and care how things feel.
Raycast - turned the Mac command bar into a command center. In a sea of clunky launchers, Raycast felt intentional. Fast, extensible, and beautiful by default. You didn’t just use it, you bragged about it. It made your workflow feel elite.
None of these tools introduced entirely new categories. But they redefined what good felt like within them. Because they were built with taste.
And taste comes from judgment. And judgment comes from doing the work.
AI is only going to make it easier to build. This means the bar to what matters will keep getting higher and that bar is made of taste, clarity, and care.
So yes, use the tools. Speed up the grunt work. Prototype faster. Iterate more. That’s the upside of this moment. But don’t let automation become autopilot.
Building software still demands critical thinking, ethical thinking, and system thinking.
Topics along this line that’s worth reading
make something heavy and taste is eating silicon valley by Anu Atluru
article by Julie Zhuo on taste part-3.
Paul Graham on taste.
Practice to studying product like Brian Lovin’s app-dissection.
Build and share ideas like Jordan Singer.
When camera phones first came out, suddenly everyone could take photos and videos.
No DSLR. No studio lighting. No training. Just a phone and something to shoot, upload it on platforms like Instagram and YouTube, while that unlocked a flood of creativity, most of what came out was...well, pretty bad. Blurry shots. Harsh lighting. No composition. No story. Because access doesn’t equal excellence.
Then something interesting happened: some artists stood out. It was not because they had better cameras. Everyone had a similar device. They stood out because they had taste.
They understood framing and composition. They trained their eye, they knew what to leave out. They shot 100 terrible photos/ videos before finding one that worked. They built intuition, one tiny judgment call at a time.
That’s exactly where we are with AI and software right now.
Anyone can build. Tools like ChatGPT, V0, Cursor, Claude, Replit, and more. they’ve opened the gates. The question is no longer "Can you build it?" It’s "should you?"
Tools got faster but taste got more valuable.
Typing a prompt is easy. Knowing what to type, when to stop, and whether what came out is actually good? That’s the hard part and that’s the craft.
It used to be that if you could code, you had leverage. Now AI writes code to build the bones of an app. It can draft copy. Suggest flows.
Similar tools. Some people ship noise. Others ship magic. That’s not luck. That’s taste. But knowing what to build? How it should work? Why it matters?
That’s still on you.
Taste is judgment in action. It’s knowing which features to cut, which copy to keep warm, and which animation to slow down. Taste separates a product from a brand.
And it’s earned by:
Building 10 things that flopped
Rebuilding them from scratch
Watching how users actually behave
Obsessing over small details most people skip
Studying products that age well, not just trends.
Asking "Is this necessary?" more often than "Is this cool?"
Taste was the moat and with the AI age, it's still the moat.
You don’t prompt your way into good design decisions. You earn them through practice, pattern recognition, and deliberate thinking.
In this AI age, we don’t need more interfaces. We need clarity. Relevance. Felt experiences.
You can build an MVP in a weekend now but making something people care about that takes a different muscle. And that muscle is creative judgment.
One example that I worked on is Truflation, a real-time inflation data platform that aggregates independent sources to offer a transparent on-chain alternative to government-reported metrics.
Most economic dashboards feel like spreadsheets for academics. The early version of Truflation wasn’t much different: dense charts, janky filters, and confusing raw data that only made sense for nerds. It was technically functional but it didn’t guide trust or decisions. The MVP did win a prize from Balaji Srinivasan for his challenge to build a transparent real-time inflation dashboard but to win real users, it needed more than that.
Today, you could’ve rebuilt that version in a weekend with AI, Tailwind, and source data. But that’s the point, a good product isn’t just about building fast. It’s about building with judgment.
We didn’t just redesign the interface. We asked hard questions:
What should users see first?
What’s getting in the way of clarity?
What do people actually want to do with this data?
So I worked with the team and simplified with:
Clear, comparative data narratives instead of throwing numbers everywhere
Gave people just a few filters they needed, not dozens of options for smart defaults over infinite config
A design system that scaled across categories, countries, and tools (widget, calculator, etc.)
And most importantly, we made it feel human.
The result was a product people could use to understand inflation and to compare data to make better decisions. The product didn’t feel like something built by a bot. It felt thoughtful. Personal. Inviting. That came from real craft, testing, and care not from prompting an AI. our decisions helped users feel confident, not confused and that’s what helped us scale Truflation into something people trust.
Now truflation powers: macro investors, economists, policymakers, former FED members, enterprises, and over 100+ protocols. Recognized by coinbase and its founder as an innovation in financial & economic data transparency. Serving as a foundation for inflation-pegged stablecoins and other DeFi solutions.
Read the full case study or view Twitter breakdown thread
Read the full case study or view Twitter breakdown thread
The first version looked like most MVP dashboards: raw data, confusing charts, and janky filters. Useful for crypto-native users but confusing for everyone else.
You could’ve rebuilt that UI in a weekend using AI, Tailwind, and data. But that’s the point, taste isn’t about moving pixels and random data. It’s about knowing what not to ship.
The MVP did win a prize from Balaji Srinivasan during his challenge to build a transparent inflation dashboard but to win real users, it needed more than that.
When I came in, the goal was clear: make inflation data feel real-time, global, and personal, not just for researchers, but for protocols, macro investors, individuals, and policymakers.
That’s not prompting. That’s product taste, earned through iteration, collaboration, and saying “no” to most things. Taste is how you move from possible to meaningful.
Every pixel is part of the brand and sends a message.
People often talk about “brand” like it’s a skin you layer on after the fact. A logo. A color palette. A font.
But the best products treat every single interaction as part of the brand.
The tone of the error message.
The way the form autofills.
The delay before a loading spinner appears.
These aren’t just UX decisions. They’re expressions of the product’s values.
A good product works.
A great product feels right.
It feels inevitable. Consistent. Considered.
And in a world where users are flooded with choices, feeling right is what makes them stick.
That’s brand judgment. That’s product taste. That’s what drives distribution.
This isn’t new. We’ve seen taste-driven products win before AI:
Notion - made writing and organizing feel like crafting. There were already hundreds of note-taking and productivity apps but the notion wasn’t just a notes app, it respected whitespace without shouting loudly features on screen. It brought calm into productivity to organize personal or professional life with its flexible structure and people felt it with early adoption by designers, founders, and indie hackers the right taste-makers.
Headspace - made meditation gentle. Meditation apps existed, but headspace used soft visuals, gentle narration, and playful UX to make calm feel accessible. It didn’t just guide you, it invited you. The taste gave it mass appeal.
Linear - made issue tracking feel like design. It wasn’t built for PMs drowning in tickets. It was for fast-moving teams who valued speed, clarity, and taste. Every micro-interaction from keyboard shortcuts to dark mode whispered craftsmanship. It spoke to teams who move fast and care how things feel.
Raycast - turned the Mac command bar into a command center. In a sea of clunky launchers, Raycast felt intentional. Fast, extensible, and beautiful by default. You didn’t just use it, you bragged about it. It made your workflow feel elite.
None of these tools introduced entirely new categories. But they redefined what good felt like within them. Because they were built with taste.
And taste comes from judgment. And judgment comes from doing the work.
AI is only going to make it easier to build. This means the bar to what matters will keep getting higher and that bar is made of taste, clarity, and care.
So yes, use the tools. Speed up the grunt work. Prototype faster. Iterate more. That’s the upside of this moment. But don’t let automation become autopilot.
Building software still demands critical thinking, ethical thinking, and system thinking.
Topics along this line that’s worth reading
make something heavy and taste is eating silicon valley by Anu Atluru
article by Julie Zhuo on taste part-3.
Paul Graham on taste.
Practice to studying product like Brian Lovin’s app-dissection.
Build and share ideas like Jordan Singer.
When camera phones first came out, suddenly everyone could take photos and videos.
No DSLR. No studio lighting. No training. Just a phone and something to shoot, upload it on platforms like Instagram and YouTube, while that unlocked a flood of creativity, most of what came out was...well, pretty bad. Blurry shots. Harsh lighting. No composition. No story. Because access doesn’t equal excellence.
Then something interesting happened: some artists stood out. It was not because they had better cameras. Everyone had a similar device. They stood out because they had taste.
They understood framing and composition. They trained their eye, they knew what to leave out. They shot 100 terrible photos/ videos before finding one that worked. They built intuition, one tiny judgment call at a time.
That’s exactly where we are with AI and software right now.
Anyone can build. Tools like ChatGPT, V0, Cursor, Claude, Replit, and more. they’ve opened the gates. The question is no longer "Can you build it?" It’s "should you?"
Tools got faster but taste got more valuable.
Typing a prompt is easy. Knowing what to type, when to stop, and whether what came out is actually good? That’s the hard part and that’s the craft.
It used to be that if you could code, you had leverage. Now AI writes code to build the bones of an app. It can draft copy. Suggest flows.
Similar tools. Some people ship noise. Others ship magic. That’s not luck. That’s taste. But knowing what to build? How it should work? Why it matters?
That’s still on you.
Taste is judgment in action. It’s knowing which features to cut, which copy to keep warm, and which animation to slow down. Taste separates a product from a brand.
And it’s earned by:
Building 10 things that flopped
Rebuilding them from scratch
Watching how users actually behave
Obsessing over small details most people skip
Studying products that age well, not just trends.
Asking "Is this necessary?" more often than "Is this cool?"
Taste was the moat and with the AI age, it's still the moat.
You don’t prompt your way into good design decisions. You earn them through practice, pattern recognition, and deliberate thinking.
In this AI age, we don’t need more interfaces. We need clarity. Relevance. Felt experiences.
You can build an MVP in a weekend now but making something people care about that takes a different muscle. And that muscle is creative judgment.
One example that I worked on is Truflation, a real-time inflation data platform that aggregates independent sources to offer a transparent on-chain alternative to government-reported metrics.
Most economic dashboards feel like spreadsheets for academics. The early version of Truflation wasn’t much different: dense charts, janky filters, and confusing raw data that only made sense for nerds. It was technically functional but it didn’t guide trust or decisions. The MVP did win a prize from Balaji Srinivasan for his challenge to build a transparent real-time inflation dashboard but to win real users, it needed more than that.
Today, you could’ve rebuilt that version in a weekend with AI, Tailwind, and source data. But that’s the point, a good product isn’t just about building fast. It’s about building with judgment.
We didn’t just redesign the interface. We asked hard questions:
What should users see first?
What’s getting in the way of clarity?
What do people actually want to do with this data?
So I worked with the team and simplified with:
Clear, comparative data narratives instead of throwing numbers everywhere
Gave people just a few filters they needed, not dozens of options for smart defaults over infinite config
A design system that scaled across categories, countries, and tools (widget, calculator, etc.)
And most importantly, we made it feel human.
The result was a product people could use to understand inflation and to compare data to make better decisions. The product didn’t feel like something built by a bot. It felt thoughtful. Personal. Inviting. That came from real craft, testing, and care not from prompting an AI. our decisions helped users feel confident, not confused and that’s what helped us scale Truflation into something people trust.
Now truflation powers: macro investors, economists, policymakers, former FED members, enterprises, and over 100+ protocols. Recognized by coinbase and its founder as an innovation in financial & economic data transparency. Serving as a foundation for inflation-pegged stablecoins and other DeFi solutions.
Read the full case study or view Twitter breakdown thread
Read the full case study or view Twitter breakdown thread
The first version looked like most MVP dashboards: raw data, confusing charts, and janky filters. Useful for crypto-native users but confusing for everyone else.
You could’ve rebuilt that UI in a weekend using AI, Tailwind, and data. But that’s the point, taste isn’t about moving pixels and random data. It’s about knowing what not to ship.
The MVP did win a prize from Balaji Srinivasan during his challenge to build a transparent inflation dashboard but to win real users, it needed more than that.
When I came in, the goal was clear: make inflation data feel real-time, global, and personal, not just for researchers, but for protocols, macro investors, individuals, and policymakers.
That’s not prompting. That’s product taste, earned through iteration, collaboration, and saying “no” to most things. Taste is how you move from possible to meaningful.
Every pixel is part of the brand and sends a message.
People often talk about “brand” like it’s a skin you layer on after the fact. A logo. A color palette. A font.
But the best products treat every single interaction as part of the brand.
The tone of the error message.
The way the form autofills.
The delay before a loading spinner appears.
These aren’t just UX decisions. They’re expressions of the product’s values.
A good product works.
A great product feels right.
It feels inevitable. Consistent. Considered.
And in a world where users are flooded with choices, feeling right is what makes them stick.
That’s brand judgment. That’s product taste. That’s what drives distribution.
This isn’t new. We’ve seen taste-driven products win before AI:
Notion - made writing and organizing feel like crafting. There were already hundreds of note-taking and productivity apps but the notion wasn’t just a notes app, it respected whitespace without shouting loudly features on screen. It brought calm into productivity to organize personal or professional life with its flexible structure and people felt it with early adoption by designers, founders, and indie hackers the right taste-makers.
Headspace - made meditation gentle. Meditation apps existed, but headspace used soft visuals, gentle narration, and playful UX to make calm feel accessible. It didn’t just guide you, it invited you. The taste gave it mass appeal.
Linear - made issue tracking feel like design. It wasn’t built for PMs drowning in tickets. It was for fast-moving teams who valued speed, clarity, and taste. Every micro-interaction from keyboard shortcuts to dark mode whispered craftsmanship. It spoke to teams who move fast and care how things feel.
Raycast - turned the Mac command bar into a command center. In a sea of clunky launchers, Raycast felt intentional. Fast, extensible, and beautiful by default. You didn’t just use it, you bragged about it. It made your workflow feel elite.
None of these tools introduced entirely new categories. But they redefined what good felt like within them. Because they were built with taste.
And taste comes from judgment. And judgment comes from doing the work.
AI is only going to make it easier to build. This means the bar to what matters will keep getting higher and that bar is made of taste, clarity, and care.
So yes, use the tools. Speed up the grunt work. Prototype faster. Iterate more. That’s the upside of this moment. But don’t let automation become autopilot.
Building software still demands critical thinking, ethical thinking, and system thinking.
Topics along this line that’s worth reading
make something heavy and taste is eating silicon valley by Anu Atluru
article by Julie Zhuo on taste part-3.
Paul Graham on taste.
Practice to studying product like Brian Lovin’s app-dissection.
Build and share ideas like Jordan Singer.
When camera phones first came out, suddenly everyone could take photos and videos.
No DSLR. No studio lighting. No training. Just a phone and something to shoot, upload it on platforms like Instagram and YouTube, while that unlocked a flood of creativity, most of what came out was...well, pretty bad. Blurry shots. Harsh lighting. No composition. No story. Because access doesn’t equal excellence.
Then something interesting happened: some artists stood out. It was not because they had better cameras. Everyone had a similar device. They stood out because they had taste.
They understood framing and composition. They trained their eye, they knew what to leave out. They shot 100 terrible photos/ videos before finding one that worked. They built intuition, one tiny judgment call at a time.
That’s exactly where we are with AI and software right now.
Anyone can build. Tools like ChatGPT, V0, Cursor, Claude, Replit, and more. they’ve opened the gates. The question is no longer "Can you build it?" It’s "should you?"
Tools got faster but taste got more valuable.
Typing a prompt is easy. Knowing what to type, when to stop, and whether what came out is actually good? That’s the hard part and that’s the craft.
It used to be that if you could code, you had leverage. Now AI writes code to build the bones of an app. It can draft copy. Suggest flows.
Similar tools. Some people ship noise. Others ship magic. That’s not luck. That’s taste. But knowing what to build? How it should work? Why it matters?
That’s still on you.
Taste is judgment in action. It’s knowing which features to cut, which copy to keep warm, and which animation to slow down. Taste separates a product from a brand.
And it’s earned by:
Building 10 things that flopped
Rebuilding them from scratch
Watching how users actually behave
Obsessing over small details most people skip
Studying products that age well, not just trends.
Asking "Is this necessary?" more often than "Is this cool?"
Taste was the moat and with the AI age, it's still the moat.
You don’t prompt your way into good design decisions. You earn them through practice, pattern recognition, and deliberate thinking.
In this AI age, we don’t need more interfaces. We need clarity. Relevance. Felt experiences.
You can build an MVP in a weekend now but making something people care about that takes a different muscle. And that muscle is creative judgment.
One example that I worked on is Truflation, a real-time inflation data platform that aggregates independent sources to offer a transparent on-chain alternative to government-reported metrics.
Most economic dashboards feel like spreadsheets for academics. The early version of Truflation wasn’t much different: dense charts, janky filters, and confusing raw data that only made sense for nerds. It was technically functional but it didn’t guide trust or decisions. The MVP did win a prize from Balaji Srinivasan for his challenge to build a transparent real-time inflation dashboard but to win real users, it needed more than that.
Today, you could’ve rebuilt that version in a weekend with AI, Tailwind, and source data. But that’s the point, a good product isn’t just about building fast. It’s about building with judgment.
We didn’t just redesign the interface. We asked hard questions:
What should users see first?
What’s getting in the way of clarity?
What do people actually want to do with this data?
So I worked with the team and simplified with:
Clear, comparative data narratives instead of throwing numbers everywhere
Gave people just a few filters they needed, not dozens of options for smart defaults over infinite config
A design system that scaled across categories, countries, and tools (widget, calculator, etc.)
And most importantly, we made it feel human.
The result was a product people could use to understand inflation and to compare data to make better decisions. The product didn’t feel like something built by a bot. It felt thoughtful. Personal. Inviting. That came from real craft, testing, and care not from prompting an AI. our decisions helped users feel confident, not confused and that’s what helped us scale Truflation into something people trust.
Now truflation powers: macro investors, economists, policymakers, former FED members, enterprises, and over 100+ protocols. Recognized by coinbase and its founder as an innovation in financial & economic data transparency. Serving as a foundation for inflation-pegged stablecoins and other DeFi solutions.
Read the full case study or view Twitter breakdown thread
Read the full case study or view Twitter breakdown thread
The first version looked like most MVP dashboards: raw data, confusing charts, and janky filters. Useful for crypto-native users but confusing for everyone else.
You could’ve rebuilt that UI in a weekend using AI, Tailwind, and data. But that’s the point, taste isn’t about moving pixels and random data. It’s about knowing what not to ship.
The MVP did win a prize from Balaji Srinivasan during his challenge to build a transparent inflation dashboard but to win real users, it needed more than that.
When I came in, the goal was clear: make inflation data feel real-time, global, and personal, not just for researchers, but for protocols, macro investors, individuals, and policymakers.
That’s not prompting. That’s product taste, earned through iteration, collaboration, and saying “no” to most things. Taste is how you move from possible to meaningful.
Every pixel is part of the brand and sends a message.
People often talk about “brand” like it’s a skin you layer on after the fact. A logo. A color palette. A font.
But the best products treat every single interaction as part of the brand.
The tone of the error message.
The way the form autofills.
The delay before a loading spinner appears.
These aren’t just UX decisions. They’re expressions of the product’s values.
A good product works.
A great product feels right.
It feels inevitable. Consistent. Considered.
And in a world where users are flooded with choices, feeling right is what makes them stick.
That’s brand judgment. That’s product taste. That’s what drives distribution.
This isn’t new. We’ve seen taste-driven products win before AI:
Notion - made writing and organizing feel like crafting. There were already hundreds of note-taking and productivity apps but the notion wasn’t just a notes app, it respected whitespace without shouting loudly features on screen. It brought calm into productivity to organize personal or professional life with its flexible structure and people felt it with early adoption by designers, founders, and indie hackers the right taste-makers.
Headspace - made meditation gentle. Meditation apps existed, but headspace used soft visuals, gentle narration, and playful UX to make calm feel accessible. It didn’t just guide you, it invited you. The taste gave it mass appeal.
Linear - made issue tracking feel like design. It wasn’t built for PMs drowning in tickets. It was for fast-moving teams who valued speed, clarity, and taste. Every micro-interaction from keyboard shortcuts to dark mode whispered craftsmanship. It spoke to teams who move fast and care how things feel.
Raycast - turned the Mac command bar into a command center. In a sea of clunky launchers, Raycast felt intentional. Fast, extensible, and beautiful by default. You didn’t just use it, you bragged about it. It made your workflow feel elite.
None of these tools introduced entirely new categories. But they redefined what good felt like within them. Because they were built with taste.
And taste comes from judgment. And judgment comes from doing the work.
AI is only going to make it easier to build. This means the bar to what matters will keep getting higher and that bar is made of taste, clarity, and care.
So yes, use the tools. Speed up the grunt work. Prototype faster. Iterate more. That’s the upside of this moment. But don’t let automation become autopilot.
Building software still demands critical thinking, ethical thinking, and system thinking.
Topics along this line that’s worth reading
make something heavy and taste is eating silicon valley by Anu Atluru
article by Julie Zhuo on taste part-3.
Paul Graham on taste.
Practice to studying product like Brian Lovin’s app-dissection.
Build and share ideas like Jordan Singer.
When camera phones first came out, suddenly everyone could take photos and videos.
No DSLR. No studio lighting. No training. Just a phone and something to shoot, upload it on platforms like Instagram and YouTube, while that unlocked a flood of creativity, most of what came out was...well, pretty bad. Blurry shots. Harsh lighting. No composition. No story. Because access doesn’t equal excellence.
Then something interesting happened: some artists stood out. It was not because they had better cameras. Everyone had a similar device. They stood out because they had taste.
They understood framing and composition. They trained their eye, they knew what to leave out. They shot 100 terrible photos/ videos before finding one that worked. They built intuition, one tiny judgment call at a time.
That’s exactly where we are with AI and software right now.
Anyone can build. Tools like ChatGPT, V0, Cursor, Claude, Replit, and more. they’ve opened the gates. The question is no longer "Can you build it?" It’s "should you?"
Tools got faster but taste got more valuable.
Typing a prompt is easy. Knowing what to type, when to stop, and whether what came out is actually good? That’s the hard part and that’s the craft.
It used to be that if you could code, you had leverage. Now AI writes code to build the bones of an app. It can draft copy. Suggest flows.
Similar tools. Some people ship noise. Others ship magic. That’s not luck. That’s taste. But knowing what to build? How it should work? Why it matters?
That’s still on you.
Taste is judgment in action. It’s knowing which features to cut, which copy to keep warm, and which animation to slow down. Taste separates a product from a brand.
And it’s earned by:
Building 10 things that flopped
Rebuilding them from scratch
Watching how users actually behave
Obsessing over small details most people skip
Studying products that age well, not just trends.
Asking "Is this necessary?" more often than "Is this cool?"
Taste was the moat and with the AI age, it's still the moat.
You don’t prompt your way into good design decisions. You earn them through practice, pattern recognition, and deliberate thinking.
In this AI age, we don’t need more interfaces. We need clarity. Relevance. Felt experiences.
You can build an MVP in a weekend now but making something people care about that takes a different muscle. And that muscle is creative judgment.
One example that I worked on is Truflation, a real-time inflation data platform that aggregates independent sources to offer a transparent on-chain alternative to government-reported metrics.
Most economic dashboards feel like spreadsheets for academics. The early version of Truflation wasn’t much different: dense charts, janky filters, and confusing raw data that only made sense for nerds. It was technically functional but it didn’t guide trust or decisions. The MVP did win a prize from Balaji Srinivasan for his challenge to build a transparent real-time inflation dashboard but to win real users, it needed more than that.
Today, you could’ve rebuilt that version in a weekend with AI, Tailwind, and source data. But that’s the point, a good product isn’t just about building fast. It’s about building with judgment.
We didn’t just redesign the interface. We asked hard questions:
What should users see first?
What’s getting in the way of clarity?
What do people actually want to do with this data?
So I worked with the team and simplified with:
Clear, comparative data narratives instead of throwing numbers everywhere
Gave people just a few filters they needed, not dozens of options for smart defaults over infinite config
A design system that scaled across categories, countries, and tools (widget, calculator, etc.)
And most importantly, we made it feel human.
The result was a product people could use to understand inflation and to compare data to make better decisions. The product didn’t feel like something built by a bot. It felt thoughtful. Personal. Inviting. That came from real craft, testing, and care not from prompting an AI. our decisions helped users feel confident, not confused and that’s what helped us scale Truflation into something people trust.
Now truflation powers: macro investors, economists, policymakers, former FED members, enterprises, and over 100+ protocols. Recognized by coinbase and its founder as an innovation in financial & economic data transparency. Serving as a foundation for inflation-pegged stablecoins and other DeFi solutions.
Read the full case study or view Twitter breakdown thread
Read the full case study or view Twitter breakdown thread
The first version looked like most MVP dashboards: raw data, confusing charts, and janky filters. Useful for crypto-native users but confusing for everyone else.
You could’ve rebuilt that UI in a weekend using AI, Tailwind, and data. But that’s the point, taste isn’t about moving pixels and random data. It’s about knowing what not to ship.
The MVP did win a prize from Balaji Srinivasan during his challenge to build a transparent inflation dashboard but to win real users, it needed more than that.
When I came in, the goal was clear: make inflation data feel real-time, global, and personal, not just for researchers, but for protocols, macro investors, individuals, and policymakers.
That’s not prompting. That’s product taste, earned through iteration, collaboration, and saying “no” to most things. Taste is how you move from possible to meaningful.
Every pixel is part of the brand and sends a message.
People often talk about “brand” like it’s a skin you layer on after the fact. A logo. A color palette. A font.
But the best products treat every single interaction as part of the brand.
The tone of the error message.
The way the form autofills.
The delay before a loading spinner appears.
These aren’t just UX decisions. They’re expressions of the product’s values.
A good product works.
A great product feels right.
It feels inevitable. Consistent. Considered.
And in a world where users are flooded with choices, feeling right is what makes them stick.
That’s brand judgment. That’s product taste. That’s what drives distribution.
This isn’t new. We’ve seen taste-driven products win before AI:
Notion - made writing and organizing feel like crafting. There were already hundreds of note-taking and productivity apps but the notion wasn’t just a notes app, it respected whitespace without shouting loudly features on screen. It brought calm into productivity to organize personal or professional life with its flexible structure and people felt it with early adoption by designers, founders, and indie hackers the right taste-makers.
Headspace - made meditation gentle. Meditation apps existed, but headspace used soft visuals, gentle narration, and playful UX to make calm feel accessible. It didn’t just guide you, it invited you. The taste gave it mass appeal.
Linear - made issue tracking feel like design. It wasn’t built for PMs drowning in tickets. It was for fast-moving teams who valued speed, clarity, and taste. Every micro-interaction from keyboard shortcuts to dark mode whispered craftsmanship. It spoke to teams who move fast and care how things feel.
Raycast - turned the Mac command bar into a command center. In a sea of clunky launchers, Raycast felt intentional. Fast, extensible, and beautiful by default. You didn’t just use it, you bragged about it. It made your workflow feel elite.
None of these tools introduced entirely new categories. But they redefined what good felt like within them. Because they were built with taste.
And taste comes from judgment. And judgment comes from doing the work.
AI is only going to make it easier to build. This means the bar to what matters will keep getting higher and that bar is made of taste, clarity, and care.
So yes, use the tools. Speed up the grunt work. Prototype faster. Iterate more. That’s the upside of this moment. But don’t let automation become autopilot.
Building software still demands critical thinking, ethical thinking, and system thinking.
Topics along this line that’s worth reading
make something heavy and taste is eating silicon valley by Anu Atluru
article by Julie Zhuo on taste part-3.
Paul Graham on taste.
Practice to studying product like Brian Lovin’s app-dissection.
Build and share ideas like Jordan Singer.