Cognitive services like machine learning and chatbots can make humans and machines work better together by complementing each other. It’s about understanding that humans and machines are better at different things. Humans are very good at handling unpredicted situations with their common sense, and machines are very good at making a lot of calculations and digging through vast amounts of data. So working together makes a lot of sense and can help create more value.
Chatbots are especially helpful in automating clearly scoped interactions that can be available anytime and for many people at the same time. Regarding the scope, it’s important not to try to make them able to answer any question, but instead be clear with what they understand. They can have excellent memory, and remember everything that you told them, like your personal info, your preferences, and your identity. If a chatbot knows your identity (and its confirmed that it’s really you), it should have access to everything about you, like your previous transactions, support tickets, and any upcoming reservations. With an appropriate sensitivity and respect for privacy, it could even be proactive with recommendations based on what it knows about you and other people. This require omnichannel services with deep integrations into backend systems and external services.
However, when chatbots are no longer able to help, they should make sure the user get in contact with a human for further assistance. The human should be provided with the current context (the user’s previous conversation with the chatbot), and the chatbot could stay around listening to the conversation to learn how it was handled by the human. This way the cooperation between human and machine can improve over time.
One interesting aspect of chatbots is the new ways that the interaction has to be designed. It’s not so much about graphics design or even screens, but more about how to use language to create an excellent user experience, and requires humanistic and linguistic skills. Things like “tone of voice”, attitude, and use of certain words is very important, and becomes an essential part of branding. The design of the interaction need to take into account that the users feel they are actually having a conversation directly with the brand.
Another interesting aspect is that once the chatbot interaction is designed, it works the same way through multiple channels, whether it’s an instant messaging client, a web page, or an app. It can even be the same using voice, e.g. over the phone or via an intelligent speaker like Amazon Echo (Alexa), Google Home, and Apple HomePod. Actually, the interaction can be made available on anything that can handle text or voice, like a kiosk, a physical robot, or any other connected device.
Just like building a web or an app is no longer a one-off project of implementation and then maintenance, a chatbot is a new channel that require a dedicated team over a longer period of time. I recommend a cross-functional autonomous team working together in an agile way I call beautiful delivery. You can see what I mean by that in this introduction video:
You also find a more detailed video on www.youtube.com/watch?v=0BeoBf_ddds.
Chris Forsberg is Sogeti's Global Chief Architect. Read more SogetiLabs blogs here.