What the Hell are ChatBots AND why should they be part of my Strategy?

ChatBots have been around for years in one form or another. However, the technology behind them has been developing at a remarkable rate, making them now 10 times faster to develop than a similar application performing the same function. This changes the whole value equation for decision makers looking for solutions of how to automate interacting with customers, employees and suppliers. At the same time, the proliferation of messaging apps that people use to communicate with each other has become the defacto standard method of communication people use to communicate.

 

This technology will have a massive impact on the way we do business. Consider this statistic from Gartner, that artificial intelligence will account for 85% of customer relationships by 2020. This is a fundamental shift in the way humans interact with technology. It is as big as the jump from command line operations to the GUI windows used in MAC and Windows OS in the 80’s. Even the best intuitive UI/UX now still depends on the user navigating down through a series of steps (manually) to find what they’re after. For example Men’s Fashion -> designer -> t-shirts -> short sleeve, filter on white and and size Medium. This could be replaced by a conversation that you would typically ask a shop assistant – “Show me your men’s white v-neck t-shirts in Medium”. You probably didn’t even need the ‘men’s’ or the ‘size’ as the chatbot may know these details before you even start the conversation.   

 

In April 2016, Facebook opened up to developers to build ChatBots for its Messenger service. Think of the implications of this. Marketers and businesses will be able to interact with customers and prospects. For some businesses, ChatBots could replace humans in customer service, while for others, it would become a powerful marketing tool. The slow replacement of customer service operators is likely to happen in stages. To begin with, ChatBots will be used in conjunction with humans operators. They will establish the conversation and solve many similar common requests from customers like “When is the next flight to Adelaide?”, “How many demerit points do I have left?” or “How many annual leave days do I have?”. If the requests becomes too difficult, the bot will hand-off to human operators with a complete history of what stage they were up to. ChatBots are even sophisticated enough to detect frustration in customers’ sentiment and hand off to a human operator early.

 

In any instance, the time a human operator needs to be on-task will drop meaning that the number of customer service operators needed will slowly decline. Once chatbots are in place it’s only a matter of adding further logic to their AI thus slowly increasing their utility and at the same time reducing the need for the human element.

 

Developers are now working in human like personality attributes to make interaction for users even enjoyable. Adding a bit of humor to increase engagement maybe the new version of UX we see currently in terms of the GUI.

 

There are a number of bot frameworks and platforms on the market today. Frameworks and platforms are sometimes mistakenly lumped together, but there are in fact different from each other.

 

Frameworks use a set of predefined functions and classes that developers use for faster development. Examples are Microsoft Bot framework, Wit.ai and Api.ai. They are used by developers to develop bots from scratch using programing language like C#. Platforms are online ecosystems where ChatBots can be deployed and interact with users. Some examples are Motion.ai and Chatfuel. Platforms can be created by nontechnical users and developed without coding. These can be developed quite rapidly but lose some flexibility.

 

At their core, ChatBots require natural language processing. This is where most of the magic happens and without it, developers would need to handle every possible variant of what someone could say. Additionally, the frameworks have some persistence that gives context for what a user is saying. A user may ask, “How much time have I consumed in project Alpha?” The chatbot would reply with the hours spent. Then user could ask “How many hours were scheduled?” Here the chatbot would already know that we’re already talking about project Alpha and reply accordingly.

 

We are currently running several ChatBot projects using the Microsoft Bot Framework into SAP, allowing employees to perform a number of HR functions and vendors to verify a number of procurement queries. I expect that the quality and scope of interaction will grow quite rapidly in conjunction with the underlying framework that we’re using. At the rate we’re going, it’s hard enough for me predict where we’ll be in a year, let alone further out.

 

The iconic 65-year-old Turing Test (Can a machine convince a human it was speaking to another human more than 30% of the time within five minutes?) was passed for the very first time by computer program Eugene Goostman during the Turing Test 2014, held at the renowned Royal Society in London. Reading about it made me simultaneously both scared and excited. My hope is that this type of technology will favourably augment our already complicated lives, which is great…..right?

by

Rod Taubman

No Comments

Sorry, the comment form is closed at this time.