Bots in 2016: Hype or Not?


(Amelia Salyers) #1

Hi crew,
I work at Mixpanel, and we just jumped on the bots bandwagon with a story about Kik and the “Bot Gold Rush” (clickbait-y, I know): https://mixpanel.com/blog/2016/03/09/we-re-in-a-bot-gold-rush-kik-tells-you-how-to-strike-it-rich. In addition to Kik’s position as an early adopter, it looks like Facebook is going to announce a bot platform for Messenger in April.

I’m curious, though–do y’all believe that bots in messaging platforms are the future of product? Or is this hype? We actually quote Andrew in the story saying that he thinks that messaging platforms are the next wave for distribution for new products, but is this just wishful thinking?


(Adam Michalski) #2

Hey Amelia!

I recently led up Bus Dev & Operations @GoButler - one of the first of the new generation of consumer “bot” companies that are now becoming mainstream. While I strongly believe in the value of a conversational UI, it does have its limits.

So what do I mean by limits? The reality is that a conversational UI isn’t always best for everything - which I believe is why we’re seeing new players to the market focusing heavily on specific verticals (Lola Travel for Travel, Operator for E-Commerce, etc). The trend we’re seeing play out is that where historically there was an “expert” involved that you would have chatted with via phone and/or in-person, we’re now seeing that discussion moving toward messaging because of it’s convenience and, also, the benefits to the company from a pure business perspective (e.g. agents answering queries can handle >1 conversation at a time vs call center where they could historically only handle 1 convo; see: https://blog.intercom.io/messaging-apps-just-getting-started/).

It’s also worth thinking about the different segments where bots will be effective. My bets are on the following:

  • Enterprise (handling customer support via SMS)
  • Interactive Notifications (Bank Statements, Flight Notifications, etc)
  • “Expert” (Travel, E-Commerce, more coming…)

Where will these live? More than likely within existing messaging platforms as - for the vast majority of use cases - the value-add won’t be enough for the average consumer to download an entire additional application. Distribution = scale and very few of these companies are going to have the resources out of the gate to both pioneer a new business model/interface paradigm while also trying to focus on distribution and making the case for the end-user to add another application to their daily stack. [my guess is that we’ll soon be talking about Facebook Messaging Bot Platform + Slack App Directory in many ways as the “new App Store”].

So, to sum it up, yes Bots are coming. Yes, they’re going to have a big impact. Yes, they’re going to disrupt much of what we know today…

But in terms of what/where, it’s still too early to tell. I, for one though, am excited and will be watching closely. :slightly_smiling:

Cheers,

-Adam


(Amelia Salyers) #3

I like your segment breakdown, Alex. Interactive notifications seem like a no-brainer – they’re basically a shift for transactional communication from email/apps to messaging apps, which makes a lot of sense. The “Expert” segment, though, seems like it has the most potential – in the Kik article, they talk about four areas for bots on messaging apps along the axes of engagement and usage. The area with the most potential is also the most under-developed: high engagement and high usage. I wonder if the “Expert” segment would fit in that area, though the Expert segment also seems like the most difficult to pull off.

In any event, yes, time will tell, and it seems like it will tell sooner rather than later.

Best,
~ Amelia


(Matthieu Varagnat) #4

Thank you @Amelia for this great question! I had read your article before and found it very interesting. There are much less articles on Kip as a platform, versus Slack and Messenger (which is unfair, of course). I’m myself very interested in new platforms such as SMS/newsletter based apps and messaging platforms.

To your question, I think “conversational interfaces” (bots are a subset of CI, as well as pure human back-end and hybrid human/AI) will be big. They give the ability to get info / interact in a fluid fashion, integrated with the apps on which we already spend 60-80% of our time.

However as @ajm5338 pointed out, not everything is suitable for CI. For example, I wrote this on Slack apps specifically : https://medium.com/@MVaragnat/what-makes-a-good-slack-app-23edf2ea7088.

App designers will have to think hard on what strength/weakness the new medium offers. For example, quickly browsing through pictures/news titles/links is not easy on CI. On the other hand, you can express in natural langage very deep queries that would take many clicks/taps on a traditional interface.

I think these CI apps will be cross-platforms, such as slack+hipchat for Entreprise and Messenger+Kik+Telegram for Consumer. There is no sense of having a separate app for them, it would be even contradictory with their advantage.


(Meydad Marzan) #5

Hi @Amelia,

I admit I wasn’t an advocate for bot or conversational UI apps, till after I did this micro-research that helped realize when bots are good, and when they are just a gimmick

I’ll add that all this is in the era where AI is a bit premature for these apps, and some of them might evolve to provide great value for users. This is where things stand TODAY and from a UX point of view.

Hope you find it interesting.

Meydad


(Amelia Salyers) #6

Yeah, I think you’re likely right about the incumbent messaging apps having the advantage here. Anecdotally, I’ve been using the Quartz app as a time-waster once or twice a week, and it’s nice, but I’d rather have a better interface in Telegram (which I use a lot) or another messaging app.


(Rob) #7

I think there’s lots of potential in bots, whether it be just for fun or as first responder to customers needing help or prospects that could do with a nudge in the right direction etc.

I’ve been experimenting with some simple bots recently, currently Twitter focused but could just as easily be applied to other platforms. For example, go on to Twitter and post hi @womplify or for something a little seasonal try hi @womplify and mention st patricks day. The ‘hi @womplify’ part is what activates the bot and whatever comes after will alter the response, if any… if you’re really bored, you can also ask what time is it.

Those are trivial autoresponders and just for fun, but for something a little more useful, try posting something like can we talk? or I’d like to book a demo or I need help etc. I plan to add much more stuff there to match more types of requests, there’s not much defined yet. There’s also a real human behind the @womplify handle of course, it’s not only a bot.

These are still very simple things, not smart AI bots, and anyone can build them in a few minutes using Womplify. It could easily be expanded though (and will be) to integrate with some AI service, or WolframAlpha or any other thing that can be hooked up through an API on the back end. Bots + Integrations can be really cool.

Note that in these examples, I class them as intentional bots, in that the human is intentionally sending a message to the bot (whether or not they know its a bot). Another form is of course bots that simply look for relevant posts anywhere, not directed at the bot or business, and then responds… high potential for being classed as a spam bot in such a case but if used wisely, can be very effective.

The way things are going though, it wont be long before social networks and messaging platforms are entirely populated by bots talking to each other… we may already be there!


(Matthieu Varagnat) #8

I agree with you @Meydad. I think there are two paths forward.

Either be clear that the app is “robotic”, ie it requires a specific syntax, much like a command line. If it’s well done, it can be quite straightforward even for non-technical users.

Or, start with humans, and only when the value proposition is firmly established, introduce NLP and AI to automatize part of the process.

Starting directly with a scripted, pseudo NLP bot is, in my experience, always leading to big frustration when you think you’re talking to a really, really stupid human.