Works a bit like Bethesda's, but maybe a little cleaner interface.

Responses: holds on to a piece of text, associates sound files to read it, and sound files can be tagged to speakers (to support generic dialogue on random NPCs.)

Conversations have a list of speakers involved in the dialogue. Then each node which contains text is assigned to a speaker (the GUI only shows those listed in the entire dialogue's speaker list.)

Speaker groups: a way to have a list of valid targets for the conversation node. You can point at the group instead of a person. For players, or generic civilian dialogue.

Conversations show a noodle graph of nodes to visualize the flow of the conversation.

Nodes can have "tagged text" nodes. These are tested in order and the first valid one is chosen. Conditions allow a node to be excluded, such as giving race-specific dialogue.

Tagged text nodes may have multiple output texts; a random one is selected.

Node types

Greeting: roots of a dialogue, start here.

Question: shows the list of questions to the speaker which owns those questions.

Answer: supposed to be a response to a question, though they don't give much of an example.