The most popular story in tech right now goes like this: AI tools let one person ship a product in a weekend, so the hard part of being a founder is finally solved. That story is wrong, and a widely shared post on Hacker News lays out exactly why. The author, a co-founder of allkpop, built a platform called ClaudeFolio designed to help people solve their distribution problem. It works perfectly. It also gets about 50 visitors a day and a handful of signups.
That admission is the whole point. The building got easy. Getting found did not.
The bottleneck moved, and most people missed it
For most of software history, the hard part for a solo founder was building the thing. Twenty years of founder advice focused on stacks, time management, and avoiding the technical traps that killed amateur projects. That advice was correct for its era. Once you had working software, you had a decent shot at an audience, because working software was scarce.
That scarcity is gone. According to the Hacker News piece, tools like Claude Code have cut the cost of producing a working platform by roughly two orders of magnitude. The supply of new platforms exploded by a similar factor. Human attention stayed flat. Worse, that attention has grown more allergic to marketing across the same period.
The author’s framing is sharp: “anyone can build a platform now, and almost nobody can get people to find it.” Technical skill used to be the differentiator. Now it’s table stakes. The real differentiator is whether you can reach the people who’d benefit from what you shipped.
Why 18 years ago was a different game
What makes this account credible is the comparison point. The author grew allkpop starting more than 18 years ago, and he’s blunt that it succeeded because of the conditions, not because he was a marketing genius. Three things were different then:
- Thin competition. Covering Korean pop in English meant almost no serious rivals. Anyone searching for that content basically had to land on allkpop eventually. Empty niches like that barely exist now, and the few that appear get filled within weeks.
- A more open web. Social platforms were young and forgiving. You could drop a link in a forum or a Facebook page and people would actually click through. The web was a network of connected sites, not a set of walled gardens that punish you for sending users off-platform.
- High technical barriers as a filter. Running a real site meant managing servers, databases, security, and traffic spikes. That gatekeeping was bad for the people it locked out, but it kept the field thin for everyone who made it through.
AI just removed that third filter entirely. The barrier that used to thin the herd is gone, so the herd is enormous.
Why this matters now
This is the part the “look what I built in a weekend” wave keeps skipping. The celebration of how easy building has become has crowded out the harder conversation about what happens after you ship. Most solo founders, the author included, are radically underprepared for that part. You can ship something genuinely useful and watch it sit in a digital desert with nobody walking past.
What stands out here is the honesty. The person writing about the distribution problem built a tool to solve it and got caught by it anyway. That’s more useful than any growth-hack thread.
Practical takeaways
If you’re building with AI tools, treat distribution as the actual product work, not an afterthought:
- Budget your effort backwards. If building now takes 20% of the work, plan for distribution to take the other 80%. Start it before launch, not after.
- Pick the niche for its emptiness, not just your interest. A crowded category means you’re fighting hundreds of equally-cheap-to-build rivals for fixed attention.
- Build an audience before the product. In a walled-garden web, a following you already own beats a cold launch every time.
- Stop measuring progress by what you shipped. Measure it by who showed up.
The weekend-build era is real, and it’s genuinely exciting. But the scoreboard changed. Shipping is no longer the win. Being found is. Full details are at the original source on Hacker News.