Chatbots & AI – a match made in heaven?

Chatbots, together with AI (artificial intelligence), hold a lot of promise for making the lives of customer service representatives easier by taking on some of the more mundane questions. But is it really as simple as installing a chatbot plugin?

First, let’s make clear what chatbots are and what artificial intelligence does in the context of chatbots. A chatbot is a simple way to interact with end-users (customers or internal users), usually through typing questions in a messenger-type environment.

Additionally, the interface could also be utilized through voice applications that “translate” the spoken word through natural language processing. Think Google Assistant, Siri, Cortana and the like. In the context of chatbots, AI is established as a set of keywords and rules.

Quality data is key

Depending on the chatbot you use, the user experience can either be truly amazing or it could be terrible. With the key differentiator being the underlying data used.

As Adrian Whitehouse, Director of Business Development at Elastic Path, puts it: “If they (chatbots) don’t have access to the right data presented in the right format, they’re pretty dumb.”

The question is not purely about the amount of data available — the key is having access to the right data in a form that the system understands.

When a new query comes in, the chatbot will analyze the words and break them down, parse them and then figure out which Application Programming Interface (API) to call. The API looks at the keywords and ascertains what information is needed and whether or not it has access to it.

In the case of something simple, like opening hours of a particular retail location, the API retrieves the requested information and sends it back to the chatbot, which then renders the information either as text or voice.

In more complex situations — for example, questions that are tied back to a real person: when will my phone contract end and similar — the API will also look for personally identifiable information about the person asking.

The API knows the user because they have logged in to the website, service or app using Facebook or other means.

This will continue until there are no more questions from the user, or the chatbot doesn’t understand the question asked and will forward it to a customer service representative. Customer service will then have all the information from the previous chatbot interactions and will be better prepared to help the user.

Chatbots and customer experience

Chatbots can be an amazing addition or it could flop. It’s not enough to just have it run because everyone else is using it, it needs to be a part of your whole customer experience and be integrated with all other systems in it.

As Whitehouse puts it: “… if you don’t think through the entire experience, you’ll underuse chatbots and they probably will be an annoyance.”

Chatbots, together with AI, can help companies build better customer experiences when done right (no more waiting in line and listening to elevator music!) and cut costs (less work for customer service representatives). When done incorrectly, however, more problems can be created instead of solved.

This article is based on our podcast episode titled “How chatbots and AI are changing customer experience.”