I am surprised by this point of view. It seems to me that extending the web-browser UI to display complex non-web-page documents is a bad idea, that it makes the user experience more complicated and mysterious.
Most of the time, the user-experience problem with PDFs on the web isn't that a separate app has to be started to read the PDF, but that the information in question is in a PDF at all, when a simple HTML page is generally safer and more useful.
Most of the time, the user-experience problem with PDFs on the web isn't that a separate app has to be started to read the PDF, but that the information in question is in a PDF at all, when a simple HTML page is generally safer and more useful.