Developer With No Coding Skills Vibe-Coded a Bedbug Map in 2 Weeks: 300K Users and Wars With Hotels

This story from China shows what happens when vibe coding works too well. The product comes together in two weeks, and the responsibility for it arrives at once, with no adjustment period.

How it started

A 26-year-old man, let's call him Li, moved into a new apartment in the summer. At night, bedbugs started biting him and his girlfriend. Li began reading about them and found out that bedbugs are a widespread problem in China. They are most often reported in connection with hotels. Perhaps apartment residents simply stay quiet, while hotel guests complain more readily. There is almost no response to these complaints.

Li couldn't code, but he was angry. In two weeks, with the help of AI, he built a WeChat mini-app called "Bedbug Map." The mechanics are as simple as they get: you see a bedbug in a hotel, drop a pin, write "bedbugs here," and the pin appears on the map. The mini-app quickly gained 300,000 users and 600+ flagged hotels across the country.

Where it all broke

The problem came down to one detail: a pin could be placed without a photo and without any evidence at all. For a service that labels a specific business "dirty," that is a serious hole.

The consequences came quickly:

• One guest demanded that a hotel refund 3000 yuan. He had two red spots on his body, and the hotel carried a bedbug pin. From that, he concluded he had been bitten there. Pest control specialists brought in to inspect found nothing, and the hotel filed a complaint against the mini-app.
• Another hotel hired a third-party company to inspect all its rooms. It sent the reports and certificates to WeChat and demanded that the pin be removed and its losses compensated.

Li built a product that users liked. But he did not build everything that is usually built around user-generated content: moderation, appeals, verification, a process for resolving disputes. While the app was unknown, that didn't matter. Once it became popular, the missing pieces started hitting it from every side.

What WeChat did

In the end, WeChat shut the mini-app down. The stated reason: an individual cannot operate a service built on user-generated content. This is a standard platform rule that is usually not strictly enforced. The author of the original post suggests it comes down to risk. If hotels sue, there is nothing to collect from a lone developer, and the lawsuit could end up reaching the platform itself. That is his interpretation; no reasons were officially disclosed.

Li did not give up. Now every pin request goes through manual review: the user attaches a photo of a bedbug, and Li looks at it personally. Hotels now have an appeals channel. According to available information, Li has already reviewed more than 700 photos, and almost two-thirds of the pins failed the check. In other words, without the filter the map would have been filled mostly with unreliable data.

The reaction of Chinese state media is also interesting. They assessed the project positively: the homemade map exposed a real gap, and major booking services should add such data themselves. Li does not plan to monetize the project.

What follows from this

This is not a story about vibe coding being dangerous. If anything, the opposite: it sharply lowers the cost of an idea, and ideas can be tested on real people. But it has a side effect. The speed of building now far outpaces the speed at which the author grasps the consequences.

Practical takeaways for anyone building products with user-generated content:

• If users can publicly accuse someone, think through the evidence requirements in advance: a photo, a receipt, a confirmation. A pin without proof turns into a weapon.
• You need appeals from day one. If the subject of a record has no way to dispute it, they will go to court or to the platform.
• Check the rules of the platform you launch on. Restrictions such as a ban on individual operators can go unenforced for years, and then be enforced against you specifically.
• Have a plan for success. Growth of 300,000 users in a short time is no longer an experiment for a solo developer, but an operational load.

What sets this case apart is that the idea turned out to be strong. If large platforms pick it up, they will introduce the verification and moderation that the first version lacked. And the author got something that rarely falls to beginners: the market showed that the problem exists before he had managed to learn to code.