There are three distinct "bots" for OpenCog: CogBot, CogitaBot and the embodiment bot.
The first iteration of CogBot is an interface to OpenSim for OpenCog, implemented by by Kino High Coursey (mentored by Andre Luiz de Senna) for GSoCProjects2008. TODO: What was learned here? What was the result? Where can the code be found?
Cogita-bot is a simple question-answering IRC chatbot. It serves two purposes: to demo the OpenCog natural language processing pipeline, and to provide a foundation for question answering and concept formation research within OpenCog. When it is working (and not being worked-on) it can be found on the #opencog channel on the freenode.net IRC network. A more detailed description of what it does, and how it does it, can be found in the source file opencog/nlp/chatbot/README. The README includes a sample dialog of the chatbot answering questions, and detailed instructions on how to install and run it. The README also describes (a little bit) what it's limitations are, and how to overcome some of it's limitations. Contact Linas Vepstas for more details.
The embodiment work is proceeding apace, and the pet dogs can now understand rudimentary English, and to respond in English as well. The primary goal of the embodiment work is to demonstrate (unsupervised) learning in a 3D virtual environment, and to perform research on learning. Work is ongoing; several papers published in academic journals provide more detail. Multiple videos on you-tube demo the pets in action, interacting with humans. See Category:Embodiment for a listing of embodiment-related wiki pages, including those pertaining to language processing.