How to build a Chatbot
Getting a chatbot to interact with end-users is harder than it looks.
Today’s customers are comfortable using text messaging apps and voicebots to talk to brands. In China, chatbots on WeChat already let users pay invoices, order taxis or even reserve karaoke stations. Every day, millions of people now say hello, goodbye and even thank you to what are essentially machines.
For chatbots to be visible for this new class of user, it is important to combine a range of new technologies so that brands can maintain a high-quality conversation with their users, with the right answers for the right questions and the right context.
Many different building blocks are required to manage and understand a conversation, based on artificial intelligence algorithms and machine learning. Thanks to the evolution of technologies such as Natural Language Processing (NLP) and Natural Language Understanding (NLU), chatbots on instant messaging platforms such as Facebook Messenger, Amazon Echo Alexa and Google Home are now able to parse text and voice successfully and in some cases even to analyze a user’s sentiment.
Chatbots rely on these technologies to detect an end user’s intentions and to extract what we call ‘named entities’, such as a city where they want to travel or a product that they want to order. With dialogue management and Natural Language Generation (NLG), the chatbot can then provide appropriate responses in either voice or text format and engage in a flowing conversation with the end user.
Using these technologies, enterprises can create not only valuable conversations but a continuous improvement loop, in which chatbots learn from a user’s questions and responses and improve their understanding all the time. If organizations track their end users, review interactions and follow up any unanswered questions, they’ll be able to increase the power and flexibility of their chatbots.
You can pool chatbot technologies from a range of sources, including DialogFlow from Google, Watson from IBM and LUIS from Microsoft Azure. As well the big tech players, we also like to use OWI, a French start-up that allows customers to store their data on their own data centers.
But while there are a large number of off-the-shelf solutions available, these are not by themselves enough to create a top notch chatbot.
For a chatbot to deliver real value to an organization, it has to be integrated with the other IT systems of an enterprise, such as CRM (Customer Relationship Management) software and other legacy systems.
By connecting the chatbot to the CRM software or contact center, it can extract more information about the customer and develop the conversation in interesting directions. For example, by integrating a chatbot with a mobile wallet we can enable payments and deliver other smart services over the platform.
But ultimately, the main challenge is not technical. It is a customer experience challenge. It is about getting the customer experience right and providing a continuous user experience at all times, whether on a mobile app or on a website or connected device.
So, for a chatbot to succeed, it must rely on some quite sophisticated technologies. But at the same time, it must be simple to use. Sometimes in life, it can be hard to be simple.
Chatbots and AI: 2018 theme of our international student competition “Atos IT Challenge”
Artificial Intelligence is a game changing technology but some are also saying it could be a huge threat – with calls for regulation to protect humanity against AI running out of control.
We believe it is an important area for students to explore as one thing is certain: if you master the chatbot, you master the world!
More about the Atos IT Challenge competition