Turn-based context awareness
Turn-based context awareness is a conversational system’s ability to carry state across the turns of a conversation. A later utterance is understood in light of earlier ones, and nothing already answered is asked again.
Also called: multi-turn context, conversational state, dialogue state tracking
In a conversation, most sentences only make sense because of what came before. "The second one", "that address", "same as last time", "actually, make it Thursday", each refers back. A system with turn-based context awareness resolves those references against the conversation so far and updates its understanding as the caller adds or corrects information.
This is the concrete difference between a conversational agent and a phone tree. An IVR is stateless: each menu is a fresh question, and the caller repeats their account number at every step. A single-turn question-and-answer bot is similar; it answers what it was just asked with no memory of the last exchange. A context-aware agent remembers that the caller already gave their name, that they said morning appointments do not work, and that "it" means the boiler they mentioned two turns ago.
It is also what stops the most irritating failure in automated calls: being asked for something you have already said. On a test call, giving a piece of information early and then referring to it obliquely later is the quickest way to check whether an agent has it.