Starting in April, Facebook has introduced chatbots to its widely used messenger service. Chatbots are programs, usually created by a company or brand, designed to interact with customers in a conversation-like interface. Their widespread use began with Slack’s slackbot in late 2015, and has quickly been followed by offerings from Kik, Microsoft, and, of course, Facebook.
So how does Facebook’s new chatbot system work? A conversation with a chatbot will begin with some kind of welcome screen or initial message from the bot. This message lets users know more about the bot and invites them to use it. For example, Mica, the Hipster Cat Bot says “Purr purr…Nice to meet you J Tell me your city or location and I’ll tell you a nice location close by”. The user can then click “Get started” to start interacting with the bot.
Facebook then offers a couple of different ways to interact with the bot. Recently, it’s added buttons, allowing users to respond to the bot by tapping one of a selection of pre-written answers. This method ensures that users don’t become frustrated by trying to guess at the right syntax to use when interacting with each bot. It can also display a ribbon of responses to a query, which provide options for the user. In the case of Mica, on learning your location, he presents a ribbon of possible places to visit with a picture, along with two options beneath each location: “Show Details” and “Navigate me there”. Both of these options then open up other pages to provide further details. At the end of the ribbon, Mica offers an option to show more results.
A bot can also send gifs, text based messages, pictures, and videos. This array of possible responses allows you to tailor your bot to respond in the most effective way to different queries. While providing an array of options is great for bots that return news stories or restaurant recommendations, it isn’t the most effective choice for bots that offer a weather forecast. These may opt for text, gifs, pictures, or all three.
Finally, Facebook is offering access to its wit.ai Bot Engine. This is a natural language processing platform, designed to make interacting with bots more natural. It takes user input and breaks it down into data categories like “intent”, which describes what the user is trying to get the app to do, and then relevant other categories. For example, if asked to translate something, the intent is translation, and the engine will also return the text to be translated and the target language. This system helps developers to create an easier to interact with bot.
Because Messenger Bots are still in their beta phase, all bot submissions are carefully reviewed before being allowed out into the marketplace. This process helps to ensure the best possible experience, both for users and for developers. Although the offerings are still limited, there is a lot of potential for chatbots in Facebook Messenger as a platform for users to access information without downloading yet another app.