One year ago this month, the former Tesla AI director posted a casual observation about a new way to write software. He called it vibe coding. Describe what you want. Let the AI build it. Forget the code exists.

“There’s a new kind of coding I call ‘vibe coding,’ where you fully give in to the vibes, embrace exponentials, and forget that the code even exists.”

The tweet went viral. Collins Dictionary named “vibe coding” its 2025 Word of the Year. Lovable alone has generated over 10 million projects, with users creating 100,000 apps per day. App Store releases jumped to 557,000 in 2025, a 24% annual increase. December alone saw a 60% year-over-year surge after three years of flat growth.

How many of those new builders were, six months ago, customers of the apps they just cloned?

I am one of them.

I shipped Good XP, a weight loss app, into a market that grows more crowded every week. I am the problem I am about to describe.

Remember when the internet had civilians?

Last week I wrote: “The audience and the sellers are merging.”

A reader quoted it back to me and said it captured something they had been feeling but could not name.

I think it names the most important shift happening on the internet right now.

There was a time when most people online were just people.

Browsing. Searching. Consuming.

No agenda. No funnel. No strategy.

They were civilians, using the internet the way you use a library. Looking for information, entertainment, connection.

Mixed among them were sellers. Businesses. Creators with something to offer. And those sellers could reach genuine demand because the demand was real. People were searching for solutions to problems they actually had.

The marketplace worked because the ratio was overwhelmingly tilted toward civilians. For every person selling, there were thousands just living online. Sellers could reach genuine demand because demand was everywhere, unaware of itself.

And when building software was the moat?

Production used to be hard.

Making an app required years of learning to code. Building a website required technical skill. Creating professional content required equipment and expertise.

The barrier to production kept most people on the consumer side. That barrier was the moat.

AI did not lower the barrier. It removed it. 84% of developers now use or plan to use AI coding tools. 41% of all code written globally is AI-generated or AI-assisted. The cost to build a basic SaaS MVP dropped from roughly $25,000 to a $20 subscription.

The moat is gone. And with it, the stable division between people who make things and people who buy things.

Your customer is becoming your competitor

The person who searched “best weight loss app” six months ago is now on Lovable building their own. The person who would have paid $29.99/month now thinks, “I could build this myself.”

Every AI tool, every “build your app in a weekend” tutorial, every “$10K MRR” tweet does the same thing. It converts a consumer into a producer.

Demand is being converted into supply, in real time, at scale.

This is not normal competition. Your product does not just compete against other apps in the store. It competes against the customer’s belief that they could build it themselves.

I am proof of this dynamic. I used fitness apps for years. I paid €295 and £250 in coaching. I noticed the space was accessible. I became a producer. I went from searching for “weight loss app” to building one.

That path, from consumer to competitor, is now the default for anyone technical enough to notice the opportunity.

Share

Everything is an ad

The marketing veil used to work.

Influencers recommended products as if sharing personal preferences. Content marketing looked like education. Native ads looked like articles.

Most people did not see through it. Then everyone saw through it.

Your mom knows that Instagram post is sponsored. Your uncle knows the YouTube review is paid. Nobody trusts a recommendation without checking for an affiliate link.

But the response to seeing through the veil was not skepticism. It was participation.

“Oh, they make money doing that? I could do that.”

That is the sentence that rewrote the internet.

Each new participant adds to the supply side while simultaneously becoming more resistant to other people’s supply. They are harder to sell to because they see the mechanics. And they are competing for the same buyers.

Double compression.

Share

Sellers got louder, civilians got quieter

Here is what happened next.

The sellers dominated social platforms. Every feed became a pitch. Promotional content optimized for reach pushed normal people out of the conversation. The platforms rewarded engagement, and sellers are nothing if not engaged.

So the civilians retreated.

Normal people did not stop using the internet. They became more discreet. They moved to private channels. Group chats. DMs. Forwarded emails. The places where nobody is performing.

The data backs this up. According to a RadiumOne study, 84% of all online content sharing now happens through dark social channels: private messaging apps, email, and direct messages. The conversations that matter most are the ones no analytics dashboard can track (or can they, Zuck?)

And the platforms overflowing with sellers? They work perfectly for sellers selling to sellers. That is why everyone has a course or coaching offer now. The sellers found each other. The civilians left the room.

This is the irony. By optimizing for reach and leverage, the sellers attracted more competitors, not more customers.

Leave a comment

The 3 layers that survive when building is free

When production costs collapse, the market does not just get crowded. It fragments into layers based on what is actually hard to replicate.

Layer 1 is commoditized.

Simple tools. Basic apps. Anything a weekend of vibe coding can reproduce. The moat is nonexistent. The race goes to zero.

Good XP lives partially in this layer. A weight loss app is not hard to replicate. I know that.

But here is why I am still building it: weight loss is a market people do not graduate from. Nobody solves weight loss once and leaves. The demand is perpetual. If you are going to compete in a commoditized layer, pick a market where the customers keep coming back, not because your product locks them in, but because the problem never fully goes away.

And then get ready for churn.

Layer 2 is skill-gated.

Products that require genuine taste, design sense, or deep domain expertise. The customer could technically build it. But they would build a worse version. The moat is the quality gap between what AI can generate and what a skilled human can direct AI to create.

But I will be honest about something. AI could eventually learn taste too. The gap between “good enough” and “genuinely crafted” is shrinking. If you are building in this layer, the takeaway is not to coast on current taste. It is to develop it faster than AI catches up.

Study design. Build opinions. Make choices that a model trained on averages would never make.

Layer 3 is trust-gated.

Products where the value is inseparable from the person or community behind them. The customer cannot replicate this because the product IS the relationship.

Making of a Maker lives here. Well Travelled, a community brand I am helping build a software offering for, lives here. The software is secondary to the trust the community already has.

If you are building in this layer, the takeaway is to find or build communities you can serve. Communities organized around a shared problem. Then build the software those communities need. The trust already exists. You are just giving it a tool.

Where the civilians still gather

If the civilians retreated from public platforms, where did they go?

Not maker Twitter. Not Indie Hackers. Not Product Hunt. Those communities are organized around commercial activity. Everyone there is building. Everyone is selling.

A maker community does not have buyers. It has competitors.

The civilians, the people who still want to buy a solution rather than build one, are in two places.

The first is boring, unglamorous search channels.

Google Search. The App Store. ChatGPT. YouTube. Places where someone with a real problem types real words looking for a real solution. Running Apple Search Ads, targeting keywords like “weight loss app,” I am reaching civilians. People who want to lose weight, because this is a private problem people do not advertise.

The second is dark social.

Private messages. Forwarded newsletters. Group chats. The 84% of sharing that happens where no algorithm can see it. Newsletters outperform many social channels in conversion because emails feel intentional. When someone forwards an article, they are vouching for the sender. That kind of trust cannot be manufactured at scale.

Distribution is the only durable advantage when building is free. Not distribution as a concept. Distribution as a practiced skill. The ability to reach people who will pay rather than build, in the places where they still feel safe being buyers.

Who is your civilian?

The internet lost its civilians from the public square.

But they’re still spending money. They’re still looking for solutions. But you won’t find them where you’re used to looking.

No, no, no. We, the makers, the sellers, the hustlers, scared them all away.

— Thiago Ricieri

X.com @makingofamaker
Instagram @thgvr or @makingofamaker
Threads @thgvr or @makingofamaker
LinkedIn @thiagoricieri
Substack Making of a Maker written by @thgvr