I think you are looking for Sikuli.
This tool will run at the first level to simulate a real user, whether on a webpage or a desktop application. I don't have a first user experience with this tool But in my previous company they have seriously considered it as the main testing tool for one of the projects and saw my colleges working on it.
It seems it provides almost all the functionalists you mentioned.
From my understanding, you can provide the steps in details, or specify the required action to occur on a certain trigger, such as navigate from a web page, click a certain image when it appears on the screen, fill in a form when it loads, like a FB user status when they post etc. You can program it to do a test a user story from the beginning till the end, e.g. user log in functionality, fill in a form, submit a form, upload an image etc. I even heard some people use it to play FB games for them when they are away!
Cited from their site:
Sikuli is an open-source research project originally started at the
User Interface Design Group at MIT.
Sikuli automates anything you see on the screen. It uses image
recognition to identify and control GUI components. It is useful when
there is no easy access to a GUI's internal or source code.
It seems it will search a domain/range of pixles on the screen to find a certain image (which you provided), once found, will trigger a set of actions that you per-defined.
See the following example:
Open cCleaner on a desktop:
Login on a website form: