Don’t be fooled, Bots are NOT your friends.

In the time since F8 and Facebook’s announcement of its bot platform on Messenger, much has been written about how NLP Bots will completely change the way users interact with technology, for the foreseeable future. Some of this prognostication has been for affect, but there are serious and experienced players who have offered their $0.02 around what the future looks like — and they see user behavior evolutions on a grand scale.

There’s also been no shortage of haters. Everyday there’s a new member of the hater’s ball.

While it’s tempting to be filled with excitement and a sense of possibility around our ability to simulate human dialog + intelligent agent functions — don’t be fooled, Bots are NOT your friends.

The majority of this first wave of Bots are not betas that need further iterations, their premise is misguided:

It’s Too Early.

Conversational bots can be really creepy. I mean seriously, how often have you sat around wanting to have a conversation, in real-time, with a robot? It sounds odd to a significant segment of ordinary web users. There are specific cases where you might say: “Siri, play John Legend” or “Read my emails to me”, but constructing full sentences, using idioms and innuendos, with non-human agents consistently (e.g. daily, weekly, etc) can not only feel quirky to some, but can feel like a lonely way to interact with the world. Will that change over time? Possibly, but we’re years out and many conventions from that. 
Facebook’s main use case, conversational customer service, seems to miss the point. The dominant use case I experience when contacting customer-service is to complain about something I don’t like. A piece of their thesis is correct — users don’t want to wait on hold for someone to decide they’re important enough to talk to. Good. But, at that very moment of frustration, do I want to talk to a robot who may not understand my level of frustration or a robot who can’t empathize with my concerns? Potential to make me feel more like a number and less like a human the company values. We’ve got a long way to go to inject empathy into bots. And, of course the question then becomes how much empathy, when coming from a machine, is creepy?

Who actually wants to be this guy?

It’s Too Much Work.

Browsing is easy. Search is easy when I know what I’m looking for. Conversations around mundane tasks are cumbersome. Let’s play this out — you’ve all seen it by now:

Wait, what? I just used 120 characters and 30 secs of my life to read then respond to this robot, and after all that, it gets the date for Memorial Day incorrect?
The point of the above conversation with random travel bot is not to point-out mistakes of a beta iteration. As a UX designer and Engineer, the goal is to always reduce the user’s workload by at least half. Then, once you do that, reduce it by half again. So, from a user perspective, typing that was a lot of work (especially on mobile) just to add the origin and destination of a flight.

I don’t want to type or say more. The novelty wears off quickly.

As a user, I’d ideally like to just grunt at my phone and have it know what I want, and do it. I want bots to be the mature iteration of the app Yo
I want it to just alert me of things. Not have dinner conversation with me. If I’m having a conversation with a friend over text about going to Miami for Memorial Day — buzz me and say: ‘don’t know if you’re ready to pull the trigger but here are a few Memorial Day deals that might be worth a look”. Enable me to do less work and simply make the important decisions. Conversation is added fluff I don’t need. We’re several conventions from optimizing this.

The Margin-of-Error is Razor Thin.

List of products I give second chances to when they misunderstand what I type:

Bots like everything else in this world are selling something. They’re selling the idea that giving them your time is worth it because, in the end, you’ll save time & work. When they add to my work, I never go back. It’s like willingly talking to a customer service agent who can only speak half your language. Ain’t nobody got time fo dat.

Case in Point:

I’m NOT blasting any developers or their products. Quite the opposite. I’ve seen some of the smartest people I know working on these problems. Rather, I’m saying, we seem to have been swept off our feet by the prospect of conversational bots, when in fact, our collective talent could raise the tide on products that more immediately meet user needs.

It Misses the Point.

This brings me to my last point — just make life easier. I’m not lonely. I’m not cruising the web for bot friends. I don’t need a buddy to converse with.

What I really need, what’s really a painkiller for me are tools that reduce work in an elegant way. There are many that do this effectively on Slack. One of my favorite is Geekbot. We also built a team password bot called KindaLikeBrian. It fetches specific logins when any team member needs it.

So what do these bots look like?

Many of the good examples I’ve seen are paired with hardware — IoT technologies. I ordered a Particle to install on my Vespa so when it moves (under certain conditions), I get a notification. I’ll know quickly if someone decides to make off with my bike while I’m inside (Maybe just a NYC thing). 
But, clearly that’s not it. And, I certainly don’t have a monopoly on good Bot ideas.

There are a few things I’ve noticed in the great experiences. They’re (1) limited in their scope: simple = better (2) based on alerts/notifications: it’s frustrating to need to pull data whenever events happen. Pull data in the background, complete tasks and only involve me on important decisions (3) solving a specific use case: this should go without saying, but it should fix a current problem (4) great at organizing data around multiple users: not super easy on the Messenger platform yet but Slack’s platform is awesome for this (5) rewarding me for using: it’s just nice to hear that what I’ve done or whats been done on my behalf is or is on its way to being successful. If that’s the case, let me know.

I’m a sucker for new platforms and technologies. That’s why I’ve been so active in Matt’s facebook group, as well as building. I’ll continue to build bots and I know many of you will, too. It’s ironic, but my hope is we continue to think about specific cases that don’t require much conversational input from users, as a bridge to the days where these cases make more sense for the average person.

We don’t need buddies to chat with, we just need stuff that works.


We’ve tried with some success to do this ourselves. We’ve got a long way to go.

We started Postreach with the idea that we could automate a solution for Empty Page Syndrome — when a business’s/user’s social media pages go weeks without updates. By automating the content discovery and publishing of relevant trends, we could help businesses stay in-touch with their audience around topics their audience cares about with little to no work. We’re working so our bots meet the standards mentioned above.

I hope you enjoyed reading my post. If you click the little green 💚 below this article, it will turn red and warm and really appreciate your acknowledgement! Give it a try :]

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.