Why use Conversation Agents?
What are Conversation Agents, anyway?
Conversational Agents - software to take action on your behalf following a natural conversation - are gaining traction, and with good reason. The approach brings the computer to you (with it learning your language) rather than the more prevalent approach of dragging you to the machine, making you learn its (continuously changing) user interface or “language” for each app.
ChatBots, the distant ancestors of Conversational Agents, are hardly new. The DOCTOR version of Joseph Weizenbaum’s 1966 “Eliza” could have a convincing conversation, playing the part of a psychotherapist, and attracted some controversy.
Many “Conversation Agents” are of course little more than ChatBots in drag, and can be easily caught out. Especially, perhaps, those that “learn” on the internet. But real Conversation Agents exist too, such as the one that the author wrote for a Mars Lander. (Sadly it crashed on landing and before you ask, no, the Conversation Agent was not the pilot.)
As a hard-nosed business man or woman, you will want to the know the business benefits of deploying Conversation Agents. Like any technology, they are not a universal panacea, but in the right circumstances they do offer these benefits over and above the more traditional web, app, or progressive web app:-
Firstly, they can deliver greater user satisfaction, which one would expect to translate into greater customer retention or even, through reputation, acquisition. This is because they:
- Are easy to use, since users already know how to text. Many apps can be confusing to learn.
- Are already familiar to the user, since they simply use the user’s preferred messaging app.
- Automatically adapt to each user’s preferred way of thinking, offering a more flexible experience, since a Conversation Agent does not constrain the user to a specific flow. In contrast a typical app is necessarily optimised to a primary group of users at the expense of other groups.
- Adapt better to each user’s preferred style, especially if they can learn the user’s shortcuts.
- Can offer a more reliable experience: when a third-party or legacy back-end system is temporarily down, a Conversation Agent can know this and suggest other means for the user to achieve their aims.
- Can simply be more fun, implementing (if you wish) different characters.
Secondly, they can be cheaper to run:
- Because of the greater breadth of a Conversation Agent offers, and not being constrained to a particular path, a user is less likely to need to contact an (expensive) call centre.
- As the Conversation Agent has “built-in” access to FAQs, they can provide seamless relevant help.
Conversation Agents can identify new business opportunities. By mining logs with “unsupervised machine learning”, they can tell your business what:
- users would like to find, but are not
- New questions users have which need to be answered
Well designed Conversation Agents (HelloDone’s Cicero springs to mind) cost less to integrate and to maintain. (Of course this is not true if you were to build one from scratch, because, as we shall see, there are many hurdles to jump.) This is because:-
- They are substantially data driven.
- There is no expensive app, web site, or web app to build and maintain forever.
- By dealing with the complexities of a legacy system, a Conversation Agent can obviate the need for an expense re-write of that system.
- A Conversation Agent system can corral the multiple chat-bots many corporates have already unleashed, adding the robustness and capability that, as we shall see later, chat-bots lack.
So that is the upside. Is there a downside? Yes, there are a raft of complex issues to navigate if you choose to build a Conversation Agent, though not if you use a system like HelloDone’s Cicero. What these issues are, and how to resolve them, will be the subject of my second blog.