What’s Behind the Bot Revolution?
In the last couple of months, a flurry of big names in messaging – including Microsoft, Facebook, Kik and Line - announced they were opening up API access to their messaging platforms and creating “bot stores”. These high profile launches and the hundreds of millions of dollars being invested in the technologies around them is a sign that they believe bots will be the next big thing. But what are bots and why are tech companies suddenly getting excited about them?
What are bots?
In the simplest sense a bot is a program that automates tasks for you, like showing you the weather each morning or collating updates from a number of people into a report. More complex “chatbots” use Natural Language Processing (NLP) which allows them to understand human language and accomplish tasks through two-way conversation.
Bots in various forms, including chatbots, have been around for a long time. I remember about 10 years ago having a very charming and realistic conversation with a program called ALICEbot (Artificial Linguistic Internet Computer Entity) which was originally created in 1995. In actual fact they go much further back than that, as one of the first recognised chatbots - MIT’s ELIZA – was developed in 1966.
While early chatbots were useful academic exercises, they didn’t really accomplish much apart from being entertaining, but following recent advances in Artificial Intelligence (AI) and NLP, bots have now evolved to do relatively complex tasks including customer service, website tours and even basic training and education functions.
They have also had recent commercial successes fronting online stores and taking payments in China’s top messaging service, WeChat, and this has proved that bots can be monetized. More evidence of their usefulness can be seen in business team chat service, Slack, where they can improve productivity by handling anything from automating expenses submissions to pulling data from CRM systems into the chat window.
The Convergence of Bots and Messaging Apps
In many ways it’s natural that chat tools and bots should come together in the same ecosystem because the way we communicate with a bot has the same two-way, interactive quality we have in conversations with our friends and co-workers.
What the big messaging platforms are hoping is that instead of consumers going to many different apps to search, buy something or be entertained, with bots you’ll do that from within your favourite chat program. As a result, messaging providers like Facebook, Slack, Microsoft’s Skype and Kik have been opening their platforms to make it easier for third parties to develop and publish bots. Similarly, AI companies have been doing the same, which means bot-builders can harness intelligent conversational abilities through simple integration rather than through lengthy development.
The result for potential third party tech companies is that it’s now easier, cheaper and quicker to make very sophisticated bots, and by putting a bot live on a messaging platform it’s instantly accessible to the millions of users who already use that chat tool every day – unlike with an app which has to be downloaded separately by each user.
So what can we expect the next generation of bots to do?
My favourite vision of how the bot landscape might shape up was presented at the Microsoft Build 2016 conference, where Skype’s Corporate VP, Gurdeep Pall, showed how their virtual digital assistant, Cortana, might work alongside bots to help consumers do more.
Pall identified Cortana as being an “Agent” rather than a bot. He wants you to think of it like a real-life personal assistant that learns about you and what you like, and helps out wherever you go. A “bot” however is a much simpler program that is only given the minimum information required to accomplish a very specific task.
One shortfall that currently exists with Agents like Cortana, Google’s Now or Apple’s Siri, is that when you can ask them find a nearby pizza restaurant, they can show you a map, give directions and even reviews, but they can’t order the pizza for you. However, in the future that Pall demonstrated, if the restaurant has written a simple bot that can handle the specifics required to take and process an order, Cortana can help you choose the restaurant and add the right bot into the Skype conversation to complete your transaction and get your meal delivered.
With this example it’s easy to see how it increases the utility of agents like Cortana and the potential for third party developers to capitalise on this and provide a more engaging customer experience. While it may be some time before we get to see this level of integration working in the real world, if you want to understand what bots are doing now, my view is that the easiest way is to try some out for yourselves.
Fantastic Bots and Where to Find Them
A good starting point for locating bots is botlist.co, which tells you what’s available on different messaging platforms and mobile operating systems.
You can also browse in the mobile versions of most messaging apps, or in platform specific bot stores.
Some platforms like Facebook and Kik have developed their own quick codes, (similar to QR codes) that you can scan to find particular apps.
I tried out a few bots on some of the most popular platforms and have summarised my experience here.
Microsoft’s Bot Store isn’t available yet, but there are some basic demos from the Build 2016 conference you can Skype chat at bots.botframework.com. My favourite was Murphy, which attempts to answer “What if” questions in mash-up images. For example, typing in “What if David Bowie was on the moon?” produced an image of the moon with David Bowie’s face superimposed on the earth. There’s obviously a few kinks in the algorithms as when I typed “What if Donald Trump had a camel’s hump?”, Murphy went strangely silent.
For business productivity use cases I’d suggest you take a look at the bots available on Slack. There’s a variety of fairly mature and useful tools there - particularly for technical teams as many of these bots were created by developers trying to make their working lives simpler. Note that in order to use get the bots working your Slack administrator will need to add them into your account.
Although bot integration is still in its early days on Facebook, I found the selection on Messenger to be rather disappointing. Some of the more interesting ones (1-800 Flowers, Uber and shopping service “Assist”) are only available in the US, and the ones I could try didn’t seem to be that sophisticated.
I wouldn’t expect to be able to chat with Hewlett Packard’s Print Bot about the weather, but I might reasonably ask it to respond to a question “how much ink have I got left?” and have it come back with something other than its stock response of a welcome message and a list of menu choices. Similarly, text interactivity with the CNN bot is very restricted. On a day when one of the top headlines was the death of 74 people after a fuel tanker crashed into a bus in Afghanistan, the CNN bot was flummoxed by enquiries about “fuel tanker Afghanistan” and “fuel tankers”, however “fuel tanker” obviously hit the right pattern match and brought up the story I was looking for.
As Kik is primarily aimed at a teen audience, a lot of the bots there focus on entertainment. Riffsy GIF Keyboard lets you find animated GIFs relevant to whatever text you type so you can forward them to your friends, while FOD provides you with a stream of amusing videos from www.funnyordie.com. The bot I enjoyed playing around with the most was from H&M, which starts off by asking a few questions like your age and sex, then to work out your clothes style shows you some pictures of outfits for you to indicate which you like. Once it’s learned your taste you can then browse suggestions of outfits that it thinks you might like and buy them through H&M’s (US) website.
Building your own bot
If you want to try making your own bot, you can do this either by using the development and integration frameworks of the major messaging platforms or by using third party companies such as pandorabots.com.
Although the success of bot storefronts in WeChat has proven there is a credible case for making money from the ecommerce side, my view is for other bot applications the evidence is less clear. Messaging companies have pursued a “build it and they will come” approach by creating the platform, and so far there have been relatively few companies that seem to have hit the money making formula.
It’s now down to enterprising tech companies to work out their bot strategy and capitalise on the hundreds of millions of daily users that use chat programs every day. With the relative scarcity of engaging, commercially viable bots out there this certainly could be an opportunity for a well-planned early-mover to take the lead.
Steve Hodson is Founder and Principal Consultant at Runlevel 7, a management consultancy that provides business and technical advisory services to startups and growing businesses.