Image rather than source code

[pachico]

This is a discussion rather than a bug or feature request. Sorry in advance if this adds too much noise.

I'm wondering if you did any tests uploading screen captures for GPT to process text and choose elements to click on.

[unconv]

Hi! Yes, I in fact have a separate project that does that and I made a video building it.