Comment by babyshake

One thing I am now wondering is why bother with the Selenium script at all? Why not have the AI model describe the same things it would do in a Selenium script in detailed natural language, you could always store that in a DB or file with more efficiency than storing the video, and you could just feed the natural language description to a model for automation? And the major benefit is it is much easier for humans to review and modify if needed.