we think that's easier to integrate as there is nothing to install, nothing to maintain. We are also doing some pre/post processing which allow orientation detection for example. We also think that's might be useful for mobile as they are limited in resources. Also, on a PDF, we have an hybric way to perform both to not use slow OCR methods on easy to extract text but that's still quite a young product. New advanced features are coming.
The benefit is stamplin.com gets insight on what people are viewing and reading. They get to see what the user sees. They can compile a database and use or sell that information to be used for marketing purposes.
Also, it's an "API" (looks more like a url poiting to a CGI program to me, but whatever). API's are "cool" and "fun", while running local programs that you have control over is old and boring and not the future of computing.
our API is quite new and we understand it doesn't give an outstanding value for everybody as it target easy of use for the moment but next release is going to add more advanced things.
About us using your data, our privacy policy will clarify that.