Create value for billions using chatbots

By 2030, around half of our work will be done by robots, according to Internet trend agency KPCB. And this is becoming increasingly obvious. For instance, with the advent of chatbots: virtual workers who answer your customer queries. Via live chat, messaging or social media.

Well-known Dutch examples are’s Billie,’s Booking Assistant and KLM’s KLM on Messenger.

Two kinds of chatbots

There are 2 kinds of chatbots: ‘normal’ ones and smart chatbots. A ‘normal’ chatbot registers your customer’s query, matches it to a similar question in its database and provides your customer with an appropriate answer. Ask Billie “What is a good book for my mother?” and you will get a standard answer.

Now: Billie doesn’t ask the right follow-up questions

As you can see, this bot is suitable for short, simple conversations. You fill the database with all of the questions you can expect. And there can be many. And you have to take into account issues such as spelling errors, irony, negations, etc.

Smart bots

A chatbot that uses Artificial Intelligence (AI) is a smart chatbot. It can deal with many complicated scenarios. It can interpret a customer query, make connections and understand emotions. It understands the intention of a customer query and gives an answer to the underlaying question. In the case of Billie, a smart bot could reply: “What kind of books does your mother enjoy?”

Future: Billie asks for the right information, so it is able to properly assist you

Working with intentions makes smart chatbots suitable for more scenarios. Moreover, a smart bot can learn. You don’t have to fill it with queries, and you can teach it to come up with its own answers. For instance, IBM’s Watson gets its data from a number of sources, amongst which is Wikipedia. And Replika’s self-learning bot Jasper protects people from social isolation.

Of course, the connection with your systems plays as much a role as your chatbot’s intelligence. But, because this is true for all of your operations, we haven’t included this aspect here.

Conversational copy

Large tech firms like Amazon, IBM and Google are currently developing this form of artificial intelligence. One downside: they are mostly focussed on the English language. Dutch is not yet supported. Here lies a great opportunity for communication experts. For as long as the chatbot isn’t self-learning, online copywriters have the task to create its content. Without this input, after all, a chatbot isn’t able to provide useful information in a ‘human’ way.

More than just customer service

Chatbots are no longer limited to customer service. For instance, chatbot Watson played the quiz show Jeopardy as far back as 2011. In spite of the variety of subjects and indirect questions, Watson was able to decipher questions, give the correct answers and beat its human opponents. And Watson has a successor: CELIA.

How IBM’s smart chatbot Watson wins Jeopardy

Increasingly human

Chatbots continue to improve. It took psychologist and self-proclaimed computer geek Robert Epstein 2 months of emailing to realise that his beloved Ivana wasn’t a woman but a chatbot. Epstein’s specialisation as a psychologist? Love and relationships.

So, it’s no coincidence that chatbots attract a lot of attention. They fit in the trend of personalisation and the rise of the ‘butler economy’: offering custom-made assistance with practical apps, or services such as Uber.

Start small

Do you think a chatbot could be useful for your organisation? Don’t wait until AI is commonplace. Starting now with a less-smart chatbot will provide you with a wealth of information and experience.

What Entopic can do for you
Are you interested in a chatbot and don’t know where to begin? Or who should be on the team? In 1 or 2 workshops, we can help you get started. And if you are already using a chatbot, you can reach out to Entopic for advice on further development and (freelance) conversational copywriters.

Published: 05-2018