When anyone can build it, taste is the difference

Last week was about launching. This week I want to talk about the thing you launch.

I keep coming back to one thought. Building a product gets easier every month. A landing page takes an afternoon. A working app is a few days away, if you leave out the app store approval, which takes longer every year. I see it in my own work too. Things that took me weeks a year ago now take a few evenings with AI.

That's good news. It also means a working product doesn't set you apart much anymore. Your competitor can reverse engineer and build the same thing by Sunday.

So what does?

Distribution first, your preferences second

I've written about distribution for months, and I still think it's the moat. If nobody finds your product, nothing else matters much.

But once people do find it, I think the next thing that separates products is taste. How it feels when you actually use it. Is it fast? Does it react when you tap something? Can you get what you came for in two steps, or does it take ten? Do you feel safe putting your card details in?

What Rick Rubin says about taste

Rick Rubin has produced records for Johnny Cash, the Beastie Boys, Adele and the Red Hot Chili Peppers. In a 2023 interview, Anderson Cooper asked about his skills, and his answer surprised a lot of people:

"I have no technical ability. And I know nothing about music."

So what is he paid for? In his own words: "The confidence that I have in my taste and my ability to express what I feel has proven helpful for artists."

And a bit earlier in the same interview: "I know what I like and what I don't like. And I'm decisive about what I like and what I don't like."

I like that a lot, because I think building products is heading the same way. Rubin even made a book with Anthropic in 2025, The Way of Code, about building software with AI. The tools do more and more of the making. What's left for us is deciding what's good, and being decisive about it.

Why AI on its own gives you the average

There's actually a technical reason so many AI-built sites look alike.

Anshu Chimala spent twelve years leading software and design teams at Apple. In a piece for Lenny's Newsletter last month, he explained that whenever a model makes a design decision, it "fills in the tokens it thinks are most likely to please everyone. As a result, the design usually ends up being repetitive and bland."

Researchers at Sakana AI showed the same thing in a much simpler way. They asked models to flip a fair coin. One of them said heads 78% of the time.

So if you let AI make every choice, you get the safe option. And it hands the same safe option to everyone else. You end up with the average, and so does your competitor.

Where taste shows up

When I look at a product, these are the things I notice:

• speed, meaning how long before something useful is on the screen

• animations and haptics, the small signals that tell you something happened without slowing you down

• ease of use: fewer steps, fewer fields, fewer decisions

• security, which nobody notices but so important to all of us

• what's been left out, which is usually the hardest part

Some of it you can even measure. A 2020 Deloitte study called "Milliseconds Make Millions" looked at mobile sites in retail, travel and lead generation. When pages got 0.1 seconds faster, retail sites saw 8.4% more conversions. On lead generation sites, 21.6% more people made it from the first step of a form to the end. It's correlation and it's six years old, so I wouldn't hang anything on the exact numbers.

How I try to avoid the average

Chimala shares a few techniques in his article. I've started using some of them on my own pages, and the difference is easy to see.

1. Push it away from the default. Ask for something specific and a bit bold, or give it a random starting point. The Sakana team found that letting the model first generate a random string, and base its choices on that, brought the coin flip close to 50/50 and made its creative work much more varied.

2. Give it a critic. Let a second AI look at screenshots of the result, without the code, and judge it against a high bar. For a business site I'd also ask the critic about conversion: is there one clear headline, one obvious next step, some proof near the top?

3. Strip it back. Take out everything that doesn't help someone do what they came to do. Chimala leaves this for last. I think it's the step that matters most.

After that, it's your call. I don't think that part gets automated.

Where this meets conversion

Distribution brings people to your page. What they see and feel in the first few seconds decides whether they get in touch, sign up, or close the tab.

That's most of what my conversion engine work is about: speed, the first screen, and how few steps it takes to reach you. If your site works but feels like everyone else's, book half an hour and we'll go through it with your own numbers.

→ Book a call

Not ready for a call? Just reply with the link to your site and one sentence about what bothers you. I'll write back with the first thing I'd change.

What I'm Thinking About This Week

Whether taste is something you learn or something you practise. Rubin says he just knows what he likes. I think most of us know it too, when we use someone else's product. We notice the slow page, the extra field, the button that does nothing. Somehow we stop noticing it on our own.

If you have ten seconds, reply with the name of one product that feels right every time you use it.

P.S. The toolkit, in the order I would use it: Scout → Visibility Audit → GEO/AEO Audit Checklist → How to Get Cited by AI → SEO Strategy Playbook → Title & Meta Audit → Sovereign Idea Workflow. All free.

P.P.S. Know someone shipping fast with AI? Forward this. Fast is the easy part now. They can subscribe at thesovereigntechnologist.com.

That’s all for this week.

See you next Thursday.