๐จ๐๐ถ๐ป๐ด ๐๐ ๐๐ผ ๐๐ถ๐ป๐ฑ ๐๐ฑ๐ด๐ฒ ๐๐ฎ๐๐ฒ๐
You often think a feature is simple. The user clicks a button. The page updates. Everything works. This is the happy path. Real life is messy. Two people click the same button. A user responds late. A shift changes after confirmation. I use AI to find these gaps before I write code.
AI is not my product manager. It is a reviewer. It asks annoying questions. It forces me to slow down.
I ask AI to walk through the flow step by step.
- What happens before the click?
- What happens after?
- Who needs to know?
- What if the action fails?
AI suggests many rules. Some are too complex. Some do not fit. I treat the answers as a list to inspect. I sort them into three groups:
- Handle now.
- Use a simple default.
- Ignore until users complain.
Example: A cover shift feature. It looks easy. Staff says they cannot work. Another person takes the shift. Owner sees it is covered. AI found the gaps:
- What if two people claim it at once?
- What if the original person changes their mind?
- What if the user should not see the address yet?
These gaps change the UI. They change the database. They make the build specific.
My prompt is simple: Walk through the workflow as different users. Find edge cases and failure cases. Do not add new features. Focus on what makes the MVP unreliable.
AI does not replace thinking. It makes thinking mandatory. It turns a vague idea into a real workflow.
Source: https://dev.to/miran969/how-i-use-ai-to-find-edge-cases-before-building-a-feature-39ik Optional learning community: https://t.me/GyaanSetuAi