What this case covers
The author used Codex and Claude to build a Xiaohongshu AI workstation. Within just over ten days of launch, it received over ¥10,000 in top-ups. The author reviews requirement validation, product development, supplier stability, and the monetization process.
Questions it answers
- Do users really need a single-point tool or a complete process?
- How can you validate demand by working backward from the product?
- How do Claude and Codex divide the work?
- How do you handle API instability after launch?
- Why do tools need to be combined with services?
Our observations
- Product value comes from reducing cross-tool shuttling and repeated communication.
- AI speeds up execution, but demand judgment and acceptance remain human responsibilities.
- Stable delivery after launch matters more than the number of features in the demo stage.
From the author
What truly determines whether this can succeed is still demand judgment, process design, and stable delivery.