Hi @muslimboutique
You need to expect that OCR might give you unreliable results in some cases.
What I do in flows where OCR is needed to grab text that is not spyable:
1. Test different OCR engines, in my case tesseract works best for my language
2. Test with different image scaling in ocr action, i usually use value of 10 with tesseract.
3. Expect that you will get cases where it fails and build logic to handle those cases
And at the end - what are you trying to automate?
Do you really need to use OCR instead of UI or web automation actions that allow you to work with text directly via code/ui/web elements?