Conversational Agent
A conversational agent, or better , a conversational system, is a computer system intended to converse with a human in a natural langue.
Let's see what wikipedia states for conversational agent:
A dialogue system, or conversational agent (CA), is a computer system intended to converse with a human. Dialogue systems employed one or more of text, speech, graphics, haptics, gestures, and other modes for communication on both the input and output channel.[1]
The wikipedia definition is pretty good but a bit updated, because even the voice interaction is not mentioned!
In common practice, conversational agent in an umbrella term generally used as synonym of chatbot, even if a chatbot is a conversational agent that communicate just via texts, on chat-based platforms (as instant-messaging mobile apps).
Especially in academic parlance, the term conversational agent was synonym of chatbot, and the reason emphasis on the "agent" was maybe because conversational agents was considered an special application of intelligent systems[2] / Multi-agent systems[3] research.
The term "agent" is misleading, because assimilating any conversational system as an agent that mediate between an end-user (e.g. a consumer of a business platform) and a business operator. That's in general not true. Instead this is commonly referred as a special case of conversational system and the human operator intervention is called "human-in-the-loop". There are many cases where the conversational system is not an agent, by example in pure command-and-control conversational systems that just execute automation commands to action (electro-mechanical) actuators, by example in home-automation or industrial realms.
Nevertheless the concept of "conversational agent", even if ambiguous, is currently an active area of research of virtual assistants, in the sense of a real agentive/automation technology that assists humans to solve real-life problems, in real-time in a reactive and proactive way (co called mixed-initiative). Voice assistants are a special case of virtual assistants[4], where the human-computer interaction is mainly done via voice (voicefirst). Smartspeakers are HW device terminals that enable humans to interact with a (voice) assistant.