Let's go to work: How bots took over business

From HAL to Slackbot

Robot workers photo via shutterstock

Once cloud was accepted as something with various meanings, none of which our customers understood, the IT industry searched for the next big buzzword. It came up with not one but three terms often used interchangeably by people who don't know any better – bots, artificial intelligence and machine learning.

This is great news for software developers, who can write some code and slap an AI label on it – right?

Maybe we're not giving users enough credit here. You would hope they understood a chatbot returning an answer match from an FAQ set is different to facial recognition or sentiment analysis. An interesting thing about the rise of the bots is the pre-conditioning of users that had to happen first.

We're not a 1968 audience watching HAL in 2001: A Space Odyssey singing Daisy (based on an IBM 704 computer singing the same song). Instead, we've grown up alongside technology to the point where instant messaging, search algorithms and targeted advertising are a normal part of our day. We're so used to communicating in text that it's not a big, scary leap to think a computer is answering us back instead of another human. In our instant world, we expect information when it's convenient and we don't care who serves it – as long as it's accurate.

Inside our industry, we've been putting the machines to work for a while now. IT pros pieced together batch files to build machines back in DOS and Windows 3.1 days. Imaging and automated application deployment used to be the bastion of the enterprise and is now common (and affordable) in SMB environments. We understood automation, variables and APIs. Then we unleashed them on the world. The world was ready.

Let's blame Twitter

All a baby bot needs is a platform of users and an API. Thanks, Twitter!

Unfortunately, Twitter created a bot problem for itself. It uses MAUs (monthly active users) to prove to the world, shareholders and advertisers that it's a valuable platform. But it can't accurately split out which of those 328 million MAUs are bots, and therefore useless to advertisers as an audience. A dusty report from 2014 showed the number of MAUs that used the API (a good indication that they are non-human) had doubled compared to MAUs that didn't use the API to tweet. It's not likely that the problem got miraculously better and the current guess stands at 48 million non-humans.

But hey, some twitterbots are harmless fun, some are useful and some are used to influence presidential campaigns. Helpful bots tweet about earthquakes (@nz_quake is particularly active) or school closures (@OHClosings). Twitter even expanded its emergency capabilities, enabling you to sign up with your favourite emergency agencies and get special app notifications or SMS messages during a major incident.

Sadly, useful interactive bots are hard to find. @Ultrahal is a conversational bot trying to learn our natural language, but he failed to reply correctly to my Twitter account because it has an underscore in the name. Brands seemed to have abandoned Twitter bots, favouring Facebook Messenger or Slack instead, but a few like Pizza Hut and Airbnb played with message bots to take orders or answer questions.

Slack caught the bot bug too

If you're new to Slack, your first interaction will be with a bot. Slackbot is a less annoying version of Clippy. It gets you started with the basics, then doesn't interrupt you unless you message it. It can get easily confused – if you ask Slackbot: "How do I ask you a question?" the response you receive is: "I'm sorry, I don't understand! Sometimes I have an easier time with a few simple keywords. Or you can head to our wonderful Help Center for more assistance!" OK, that one was a little mean, sorry slackbot.

The Slack team admins can easily configure Slackbot to monitor and reply to certain keywords and phrases. One Slackbot we know sees "NSA" and replies: "Nothing to see here, move along", but you could always add something more useful.

The real fun though comes from Slack's bot directory, which features more than 600 of the automated little things. Nikabot asks team members what they've been working on each day, creating graphs to give you an overview by person or project. Not sure that it will kill time recording, but it's a start. Captain Feedback asks people for feedback on your skills or performance, when yo'’re too scared to ask them yourself. And standup meeting bots help you to run standup meetings – we kid you not. Workbot connects cloud apps to Slack, pulling information in and letting you action tasks in those apps without leaving Slack.

Brands like Uber, Lyft and Marriott Rewards make an appearance with their Slack app integration, capitalising on your need to do a thing without clicking on a different browser tab.

Slack grew because of technical communities, so it's natural that things like monitoring alerts and JIRA integrations would be a win. As acceptance by the business community grew, productivity bots and integrations soared. They're not all created equal but if you don't like one, there's likely another seven you can try instead that do a similar thing. With an overzealous admin, they are crowded and annoying, especially if there's no structure to which Slack channels they appear in.

What seemed like a good idea at the time can wear thin if you get that one team member who uses all the bots, all the time. So, like most technology decisions, it's not about the bots, it's what you do with them that counts. Have fun trying to remove one once it's in, though. There will always be a portion of the population that loved that bot and can't believe you killed it. Oh, and create a Testing channel where people can go bot mad without annoying others, until they get the hang of the syntax.

Haven't forgotten you, Facebook

With no 140-character limit, the platform where people live their lives is getting a lot more attention from brands. Recently, Facebook made it even easier for those brands by adding a row of bot suggestions inside Facebook Messenger's search pane. If you're on a smaller iOS-based phone, it's partially hidden by the keyboard, which you need to minimise.

Once again, the bots vary from sending out automated messages (a subscription service minus the hefty SMS charges) to letting you order stuff (because browser tabs are so hard). Actually, from a mobile device perspective, read "because using another app is so hard". But then some Messenger bots start to get useful. Trim connects to your bank accounts and reminds you what subscriptions you are paying for, letting you cancel them with a quick reply (US only right now). Duo Lingo lets you converse in French, German or Spanish, because the next best thing to talking to a native speaker is talking to a native-speaking bot. And Burberry's assistant will help you find that perfect scarf, poncho or cape. Of course Uber came to the party, but it added some more smarts – letting you enter a pickup address and send a message to your date, so they can click the link to book their ride, and letting you message your colleagues your location when you're running late for work (en route in your Uber).

The future or another fad?

Tim Ferriss, author of The 4-Hour Work Week, once said "automate the routine, humanise the exception". He was talking about procedures for handling customer complaints, but the world seems to be applying it to everything. We're comfortable enough ordering an Uber via an app without talking to a taxi call centre operator. We're not so thrilled at having to deal with a bot when we have a complaint, especially if we know something shouldn't have happened but did. Bots don't do exceptions well.

They do handle laziness well, though. If you want to live in a single interface as much as possible, bots are for you! Or maybe they are a good way of presenting information in one stream of consciousness, with your other communications, thereby improving your productivity.

It's still early days for brands to give their seal of approval. The big names know that a bot is a sign you are innovative and that's alluring to your millennial customers. Back in September 2016, cloud-based accounting software firm Xero demonstrated its Facebook Messenger bot to chat with your bank balance and accounts receivable. Nine months later and the bot is yet to deliver, with live functionality still limited to finding a local Xero adviser.

The world just hasn't fallen in love with bots yet. The useful ones seem US-centric with no grammar bot to change "on-premise" to "on-premises" and I struggle to find a bot that is truly indispensable.

Maybe this bot thing isn't as easy as people think. Or maybe I just haven't found the right bot yet. ®

We'll be covering machine learning, AI and analytics – and ethics – at MCubed London in October. Full details, including early bird tickets, right here.


Biting the hand that feeds IT © 1998–2017