“Tell me more about your relationship with your father.” What sounds like the scene from a therapy session is from one of the first dialogues ever conducted with a chatbot. ELIZA, the grandmother of modern technology that is currently so trendy, simulated conversational situations from psychotherapy as early as 1966 in order to examine the communication between man and machine.
While ELIZA in the 60s was still on the edge of what was technically possible, today chatbots are state of the art. Companies increasingly see them as an attractive means of simply contacting their customers. Kai Matzutt from Tallence AG knows all about fully automated chatbots. We talked to him about whether and when it is worthwhile for companies to invest in a chatbot.
“A chatbot is a software that automatically interferes in discussions, chats, or conversations and contributes concrete statements”, explains Kai. In order to keep people in front of a keyboard oblivious to the fact that they are talking to a program, developers always aim to give the bot as natural a voice as possible. To meet this criterion, a chatbot must do more than just recognise what the general subject of the conversation is. The software must also be able to anticipate and answer possible questions on this topic. To achieve this, chatbot systems access extensive databases from which they obtain their information, similarly to a full-text search engine. The more complex the database, the more natural and entertaining the dialogue with the bot becomes.
However, according to Kai’s assessment, most chat bots are still more or less far from being able to make a really meaningful contribution to the discussion.
Nevertheless, web giants Facebook, Google, Amazon, Apple and co. are striving to achieve exactly that. For quite some time now they have been working on developing personal assistants who can answer any question in the best possible way and resolve smaller tasks or even bribe with their own character. For example, the Internet is full of funny “eastereggs” and witty sayings that Apple’s Siri has been spouting out, an example of which is her cheekily parried marriage proposal with “Let’s just stay friends”. But for smaller companies and smaller budgets, such gimmicks are inefficient.
The trend is for companies to strongly focus on their core competencies. “In a project that Tallence AG recently completed the focus of the bot is clearly on providing specific answers about the product and general information about the company,” reports Kai Matzutt. In this case study, the dialogue between brand and customer takes place automatically via a chat. Tallence AG uses Facebook Messenger for this purpose.
This platform offers us various advantages: in order to be able to offer convincing Natural Language Processing (NLP), Facebook has bought the company “wit.ai”, a platform that offers free speech processing via an interface. This allows projects to be quickly and efficiently set up without the development team having to spend time on basics and the frontend. Facebook Messenger also offers numerous, quick entry points and a large number of users who have been familiar with the chat client for a long time. Messenger is thus becoming another communication channel for companies, where information is played out in an even more personal way. In addition, the chatbot function can relieve employees from customer service tasks, as the technology replaces established dialogue channels such as hotlines to a certain extent.
However, regardless of the platform, companies must define a clear objective that they want to pursue by using a chatbot. “To develop a chatbot without a concrete goal, just because the topic is “in” right now, is a waste of money”, Kai thinks. However, the investment is worthwhile precisely when the use of a bot supports service employees or when customers experience a similar added value. In order to keep the profitability of the project high, brands and service providers should always check whether a complete in-house development is necessary, or whether the development team will fall back on one of the many platforms.
As our example shows, players such as Facebook now provide sophisticated tools that allow chat bots to be developed in a way that makes them efficient and targeted.