Facebook Messenger and its AI chatbots - should the enterprise care?

Profile picture for user jreed By Jon Reed April 25, 2016
Summary:
Facebook Messenger's AI chatbots were rolled out with hyperbolic fanfare - and mixed reviews. As messaging becomes a platform unto itself, should the enterprise be paying attention?

chat-bot-robot
Ordinarily I greet Facebook events with a big, indifferent yawn. But the Facebook F8 developer show earned my eyeballs (full video replays here). I wasn't the only enterprise scribe paying attention. The big question is: why bother? F8's centerpiece announcements pertained to Facebook Messenger - not exactly the app enterprise IT managers are eager to incorporate.

But Facebook Messenger is no longer just a place to stalk annoy your old schoolmates. With the debut of several new bot-powered Messenger services, the commercial viability of AI (artificial intelligence) services is a pressing debate.

To gain enterprise clarity, these questions need an answer:

  • Why is Facebook doubling-down on Messenger?
  • How is Messenger changing from a chatty timesuck to chatbot-enabled services?
  • Should enterprises investigate Facebook Messenger services?
  • What are Facebook's risks commercializing Messenger? Will there be an enterprise equivalent, e.g. Slack?
  • Are AI-enabled chatbots sophisticated enough to warrant enterprise investment?

Why messaging environments are a threat/opportunity for Facebook

Doubling down on Facebook Messenger addresses three trends Facebook must be obsessing about:

  1. Personal sharing, the biggest differentiator of the News Feed, is declining. Messenger addresses platform "stickiness."
  2. Messaging environments may provide an answer to the vexing question of "too many apps."
  3. Messenger also provides an ideal environment for AI chatbot services.

The proliferation of mobile apps isn't a viable way for smart phone users to consume services - or so Facebook contends. During the "Introduction to Bots on Messenger" announcement (replay here), Facebook's Jessica Lee criticized the limitations of standalone mobile apps:

  • Procuring services through Messenger is "so much better than email", including ease of use, better images, and an interactive bot helping consumers make product choices.
  • The mobile app shopping experience (outside of Messenger) is tedious, involving slow loading links and difficulty comparing products, etc.

Facebook is not the only chat service on the case. In China, for example, WeChat is creating a complete environment for in-messenger service procurement.

More than one billion messages are exchanged between businesses and consumers per month on Facebook Messenger. As Facebook's Frank Montefeller gloated, "It's instant, keeps context, and establishes a permanent, canonical thread that is so valuable." Dominant use cases include customer service, e-commerce, and embedded services, such as Uber and Lyft.

The rise of the AI chatbots

The big Messenger news at F8 was the launch of the Messenger platform and the debut of the aforementiona chatbots. The Messenger platform is supported by send/receive APIs which allow companies to interact with Facebook users either via a bot, a real-live human, or some combination of the two (the combo is where Facebook claims the "magical experiences" are to be had).

Three different bot-enabled Messenger use cases were demoed at F8:  a weather bot (Poncho the weather cat, who Facebook modestly says is "hilarious"), CNN (a Messenger-specific, personalized news briefing), and "your personal shopping assistant", aka Spring. Facebook's Messenger team admitted the bots are rudimentary compared to what they could someday be, a truth quickly exploited by Buzzfeed in 6 Things That Definitely Sucked About Facebook’s New Shopping Bot.

In his TechCrunch Enterprise column, Ron Miller raised issues: Facebook sends a loud message about Messenger bots, but will we hear it? Most are overstated, except this:

The thing is, if we started using Messenger for everything, how will we get the messages from our friends and family we actually want to see (presumably)? And then there’s the whole business of ads or what Facebook calls “sponsored messages,” which will mean you have even more extraneous stuff to wade through just to get those meaningful conversations.

Miller wonders whether Facebook announced these limited bots prematurely, disillusioning early adopters. That's not a deal-breaker, and these chatbots will improve. The kicker question is: how much can they improve? In Chat bots, conversation and AI as an interface, Benedict Evans warns your AI objective must start narrow. Example: autonomous highway driving is a much easier AI nut to crack than complex urban driving. Both sides must accept the narrowness of the AI relationship:

Hence, the challenge in plugging an AI into a 'conversational' chat bot interface is that you don't have HAL 9000 but are in some sense pretending to the user that you do. You speak to it, it speaks back, and it uses natural language (either in voice or text), but it isn't general AI at all. So, how likely is it that the users can ask something that breaks it? How badly? How narrow is the domain of questions that they might ask, and how can you define the users' expectations such that they understand the domain? Perhaps more importantly, can you get people to accept the domain?

If users accept the chatbot's narrow purpose, Evans sees promise:

All of this means that for now, it seems that a bot or conversational UI might work best for something very specific - where the user knows what they can ask, and where those are the only things that they will ask. However, when it does work, it becomes very interesting indeed, particularly now because it happens to align pretty well with the second preoccupation - getting around the app-installation problem.

When Ron Miller downgrades the prospects of Messenger services because "I can just check my weather app," he's missing something Facebook has figured out: users might download a weather app, but they are sick of downloading apps for every function under the sun. That's what Evans means by the "app-installation problem."

Enterprises aren't waiting for AI chatbots to get started on the Messenger platform. Salesforce and Dropbox have already launched Messenger services:

Powered by Salesforce Lightning, the platform will enable customers to deliver personalized engagement at scale with CRM data. The company claims that each message can be linked directly to a customer’s history in the Salesforce CRM platform, enabling brands to deliver personalized messages to customers.

Salesforce believes this integration will provide another way for companies to compete on the basis of customer experience. As Salesforce's Paul Smith sees it, "Brands will be able to create deeper, more personal 1-to-1 customer journeys within chat."

My take

Is Facebook correct in its gamble that messaging environments will supplant apps? I believe yes, but it would be foolish to assume an unsuccessful mobile app would work as a messaging service.

Enterprises can hedge their bets by building apps that are largely applicable across several platforms (iOS, Android, Facebook Messenger). My understanding is that to apply to be on Facebook Messenger, your app must already be on another mobile apps store at any rate.

Right now, an AI-powered chatbot is a great way to pour money down a sinkhole without the guarantee of a result. Given the mixed-to-negative reviews of Facebook's first three chatbots, it makes sense to approach such bots with a minimalist, sandbox mentality.

It would be ironic if Facebook wound up with an winning e-commerce environment, but lost the glue of friend-to-friend messaging. If its social graph is diminished, Facebook is vulnerable to being Amazoned by, well, Amazon.

Will messaging services become part of the enterprise IT mix? Most collaboration tools have missed the mark. Slack has pole position on changing that, but it's far from a done deal. Influitive CEO Mark Organ sees messaging services making an enterprise crossover:

If you’re on Slack, there’s a product where you can say, “I need to find a hotel room in downtown Boston at least 4 stars, 90% approval rate” or whatever. You may have a human on the other end, but that human is often augmented with AI. This is called ‘human interloop processing,’ and it’s very powerful. It’s what Facebook is doing with their Messenger product as well. I think this combination of chat and AI is huge. I think it represents the future. I think it’s what the Internet looks like in a few years. I don’t know what that does to society or whatever, but I think it’s coming.

I could live without Facebook's zeal that Messenger is changing the world as we know it. This is about convenience and ease of service consumption, not an end to human suffering. But that doesn't mean enterprises shouldn't be paying attention.

End note: this is the first in a short series on Facebook's enterprise relevance. I'll also be hitting on News Feed's algorithmic secrets and the implications for enterprise publishers. I also recommend you check out my colleague Den Howlett's take on Facebook Live - the video killer app for the enterprise.