built an 800K LOC Uber competitor in 7 months. Claude wrote the code. Codex reviewed it. I was the entire engineering team.
An individual claims to have built an Uber competitor in Romania within 7 months, generating 800K lines of code with Claude as the primary development engine and OpenAI Codex as the code reviewer. This single person acted as the entire engineering team, architect, product manager, and QA. The platform has passed a governmental audit, has 50 branded cars, and 153 registered passengers, though it is not yet operational. The creator acknowledges concerns about AI-generated code but highlights the achievement of reaching this stage with minimal human input.
时间与来源
时间显示为 UTC
显示时区：UTC
本地时区尚不可用，暂时显示 UTC。
发布当时偏移：UTC+02026年10月10日 11:44 UTC
收录当时偏移：UTC+02026年10月10日 17:00 UTC
- 发布
- 2026年10月10日 11:44
- 收录
- 2026年10月10日 17:00
- 来源类型
- 开发者社区
- 档位
- 社区
- 信源状态
- 正常
档位是按信源手工设定的编辑判断，不是逐条打分。
7 months ago, I decided to build a ridesharing platform to compete with Uber and Bolt in Romania.
And I became the architect, product manager, QA department, and the guy responsible when everything broke.
Passenger and driver apps on iOS and Android. Backend infrastructure. Ride dispatch. Real-time tracking. Payments. Fleet management. Admin systems.
The platform has completed its governmental audit, and we're waiting for authorization to begin operating.
I'm not going to pretend an AI-assisted codebase is magically immune to technical debt, security vulnerabilities, or architectural mistakes.
Today, it has one that has made it through a government audit, with actual vehicles on the streets and people registering before the service is even operational.
But the fact that we've reached this point raises one I find even more interesting:
Are we witnessing the beginning of genuinely one-person software companies, or are AI coding tools just allowing individuals to accumulate technical debt at unprecedented speed?