Facebook’s bots are fairly simple on the surface.
If you’re among the upper crust of the technorati in 2016 and you haven’t released an intelligent bot framework, you’re behind the times.
Facebook announced the Facebook Messenger Platform into a beta period at its F8 developer conference in San Francisco. The Facebook Messenger Platform is intended to give the 50 million businesses that use Facebook the opportunity to integrate smart bots into their communication and commerce experiences through Messenger.
“I’ve never met anyone that likes calling a business,” said Facebook CEO Mark Zuckerberg during the F8 keynote. “We think that there’s gotta be a better way to do this … We think you should message a business in the same way you’d message a friend.”
Facebook Messenger was the fastest growing app in the U.S. in 2015 and has 900 million users worldwide. People on Facebook Messenger and WhatsApp send a cumulative 60 billion messages a day. In comparison, the height of SMS texting was 20 billion messages a day.
The bot framework for Facebook Messenger is really intended to spur conversations and commerce (conversational commerce, perhaps) in an intuitive and personal manner. Bots can be programmed to understand natural language queries and respond with the appropriate actions. Facebook believes that a hybrid between automated bots and human interaction will be the best use of the Messenger bots in the immediate future.
What’s In A Facebook Messenger Bot?
The Facebook Messenger Platform has three primary components:
- The new Send/Receive API which integrates a company’s services and pushes information through the bot on Messenger.
- Generic message templates that set up most of the cursory user experience for the developer.
- Welcome screens and calls to action that help a person communicate with the bot or make a purchase.
Facebook is creating several different ways for people to discover and begin a conversation with Messenger bots. Just as Facebook has placed the “Like” button all over the Internet, a similar Messenger “chat with us” type of button will be available for websites and apps. Companies can integrate the Messenger bot button on to their Facebook Pages as well. A section within Messenger will allow people to browse and search for available bots.
The three initial partner bots that Facebook is showing off on Messenger are from ecommerce platform Spring, a funny weather cat named Poncho and news from CNN.
How do the bots work in background? Essentially the bots function as a connector between the server (a company’s background processes) and the client (Facebook Messenger) as a bit of a bridge through the Send/Receive API. For instance, if a company is using Zendesk, Twilio, Shopify or something like Braintree on the backend to process ecommerce, the bot will integrate with those systems just like the user interface of any app.
Bot discovery in Messenger is centered on four aspects:
- Messenger Codes: an elegant type of QR code that links directly to your Messenger username.
- Messenger Links and usernames: A link that can be placed anywhere—a email signature or a website—that opens a link straight to your personalized Messenger username.
- Search from Messenger.
- Plugins for Facebook pages and websites that link directly to Messenger.
“Now, to order from 1-800-Flowers, you never have to call 1-800-Flowers again,” Zuckerberg said.
Machine Learning And Natural Language
For more conversational bots, scripts can be written and natural language processors can be added to respond to most types of queries. Facebook employs a natural language engine it calls Wit.ai that it started using with the its M personal assistant in Messenger in 2015.
“This enables you to create conversational bots that can automatically chat with users,” wrote Facebook engineer Seth Rosenberg in a post announcing the Messenger Platform. “The wit.ai Bot Engine effectively turns natural language into structured data as a simple way to manage context and drive conversations based on your business or app’s goals.”
As neural networks are still fairly new science in 2016, Facebook’s natural language engine will only be able to get developers most of the way towards conversational interactions with bots in Messenger. Spring, one of Facebook’s bot launch partners, also implements its concierge service for a hybrid artificial intelligence/human experience. Because sometimes a human understands the context of a question much better than a bot currently can.