More and more companies are using AI-based systems for voice, messaging or text communication with visitors to websites and online shops. Such chatbots are often also used on platforms such as Facebook Messenger, WhatsApp, WeChat, Slack or for answering SMS messages. They are programmed to answer questions and perform tasks.
The acceptance of chatbots by customers increases with the quality of the conversation. Basically, chatbot dialogs save time, and companies can also relieve their customer support. Companies with a presence on Facebook use the chatbot in Facebook Messenger particularly often, as this enables a broad, scalable approach to customers and new prospects.
What can chatbots do and who uses them today? – Our Chatbot FAQ provides answers.
- AI bots
Chatbots as virtual communication robots can optimize processes, support communication with customers and partners and provide new insights into communication partners.
- Google Allo
The messaging app Google Allo has an integrated chatbot, the virtual assistant Google Assistant. For example, he looks for suitable restaurants and helps with the reservation.
Chatbots are also active within Facebook Messenger. There users can subscribe to messages from chatbots, but also block them.
Chatbots are already entering the banking world. Mastercard offers the KAI bot for banks and their customers in order to simplify transactions.
There are a number of ways to develop your own chatbots, such as the Microsoft Bot Framework.
- News bot Novi
Chatbots do not always look as clearly like a robot as in the case of the news bot Novi. The images on chatbots often give the impression of communicating with a person.
- Novi answers questions
The news chatbot Novi can be asked questions about the news. However, the queries are still predetermined. Other chatbots are already working with “free” questions.
- Chatbots learn
A chatbot like Novi can be told whether a topic is interesting or not. In general, chatbots should learn from this and increasingly personalize communication.
A chatbot is an AI program for automated communication. It is a “communication robot” – not a machine made of metal, but software or a service from the cloud. Many readers will already have got to know chatbots, for example if they communicated with Apple’s customer service or asked questions in an online shop.
These users are shown a dialog box on the corresponding websites in which they can enter their questions. The answers then do not come from employees from a call center or from the help desk, but are given by the virtual communication robot, the chatbot.
Chatbots often develop user profiles, as is known from the field of web analytics, to personalize their answers. This is particularly easy in online shops, since a history of purchase habits, time intervals, payment behavior, etc. is available. Good chatbots are capable of learning, they are based on a form of artificial intelligence. Thanks to machine learning (ML), the answers are getting better and better, so that at some point users find it difficult to recognize whether they are talking to a machine or a person. Chatbots thus become virtual online assistants and consultants.
Examples of chatbots are:
Artificial intelligence (AI) is emerging as one of the megatrends for the coming years. Gartner estimates that investing in AI by 2021 will enable new business models valued at $ 2.9 trillion and 6.2 billion hours of saved time.
With modern chatbots, AI is used to continuously determine and evaluate how the speech robot should behave during a customer conversation. The necessary data is retrieved via complex interfaces to the back office applications: Do we know the caller? What are his interests? What is the status of his order? The best possible communication with the customer or prospect is controlled via dialog management.
In addition to chatbots, voicebots are also increasingly being used, which allow people to talk to a software robot. With the help of technologies for speech recognition (speech-to-text), spoken words are converted into text here and then interpreted in a machine-readable format by an AI system. On the basis of a set of rules, what has been said (e.g. “I would like to order a subscription”) is analyzed and an answer is written in text form. With the help of speech synthesis (text-to-speech), the written text is transformed into an audio file.
Chatbots can be used for many communication and organizational tasks. If they are combined with an action order, for example for booking a trip, one speaks of ActionBots. Otherwise, chatbots are used by Nduranc in the medical field to combat Alzheimer’s and dementia. They enable training conversations with sick people and, if the data is collected in the cloud, a medical analysis of language development.
Disney has fans of the animated film Zootopia solve cases together with the cartoon character Judy Hopps – Chatbot technology works in the background. The children’s aid organization Unicef enables people in developing countries to talk about grievances and wishes in the U-Report. For example, pupils in Liberia were asked if teachers had blackmailed them into sexual acts against good grades. The results were so terrifying that Unicef launched a countermeasure program with the Liberian Ministry of Education.
In Corona times, the bot MedWhat should also be interesting, through which patients and doctors can obtain remote diagnoses. It is a constantly learning system that, thanks to sophisticated machine learning algorithms, makes rapid progress in terms of the accuracy of the results. The bot is also constantly fed by the operators with medical research results and scientific work, so that it already has a considerable wealth of medical expertise.
Above all, chatbots are used in marketing. They use the media to convey content to readers and attract new readers via Facebook Messenger. Consumer goods brands let their often animated advertising characters engage in conversation with potential customers. And web shops are trying to address new customers or make existing customers happy.
Chatbots exist within social networks such as Facebook and Twitter and are then referred to as social bots. The popular messenger and chat services such as Facebook Messenger and WhatsApp are of course also among the application fields for chatbots. Here are some examples:
Chatbots take over on Twitter “Quick Replies” and “Welcome Messages”. If you send a direct message to a company’s Twitter account, a chatbot can provide an immediate, first response and a greeting – as an expression of a quick and (apparently) personal customer service.
Political parties use social bots in social networks for the election campaign. There are parties who use social bots to automatically create Facebook posts or tweets and to respond to questions from followers. Critics, however, see social bots as dangerous opinion leaders or opinion robots that should be controlled and their use should be flagged on social networks.
Mobile apps like Quartz and Resi as well as services like Novi are examples of news services that use chatbots to send users messages automatically and in a personalized manner and to be able to respond interactively to user queries.
For Facebook users, there is also the social bot JOBmehappy, which wants to make users aware of suitable jobs.
GYMI is a fitness bot. In contrast to normal fitness apps, he can conduct dialogues with the users. GYMI answers questions from users via Facebook Messenger.
HDFC Bank uses a chatbot on Facebook Messenger to offer customers advice and certain financial transactions. At the Royal Bank of Scotland, the chatbot Luvo is supposed to answer simple customer inquiries. Mastercard introduced the Bot Kai for banks and their customers.
The chatbot Edward, which is available in several Radisson Blu hotels of the Edwardian Group, allows guests to send a text message, request hotel services such as additional towels, obtain information about restaurants or make complaints.
More and more companies from a wide range of industries are also using chatbots in their contact centers. There they can receive and answer service inquiries in coordinated cooperation with human colleagues or independently.
“If you want X, press Y” – the Neanderthal among the voice bot has not yet died out, but there are much more modern communication robots that have evolved over several evolutionary stages to their current level of performance. In the meantime, Sprachbots are complex software applications that combine several components to enable a natural, goal-oriented conversation. The chain of technology almost always begins with the connection to a contact center, which controls the calls as well as the employees and voice bot to answer the calls. Contact centers are not only able to manage phone calls, but also other contact channels.
A basic distinction is made between horizontal and vertical automation when using voice bots in telephone customer contact. In vertical automation, the focus is on business processes that are completely mapped via the Sprachbot to enable customers to use Voice Self-Service. Frequently used processes with a high degree of standardization are ideal for this, such as information about account balance, order status or the request for insurance certificates.
The Sprachbot carries out various checks for plausibility and validity, which may trigger special treatment or trigger other processes, for example sending documents as in the case of the insurance certificate. Such voice-provided self-services provide a round-the-clock service for customers, take the call load off their human colleagues and free up time for more demanding tasks such as advice-intensive and value-adding phone calls.
Horizontal automation is suitable for sections of customer communication that are often used in different processes. There are three classics across all industries: concern determination, identification and legitimation. These modules can be linked as required, for example the voice messenger can first query the request in order to then identify and legitimize the caller depending on the case or to directly convey it to an employee. Querying the customer number and the associated request, such as the “complaint” case, would be an automated horizontal preselection.
It is important that the determined and verified information is passed on to the human colleague. Ideally, the right processing mask opens with the right customer data for the employee. In the case described above, the right customer file with the appropriate forms for complaints, but also, for example, stored interview guidelines.
A simple calculation example shows how the average conversation time of the agents can be reduced by horizontal automation. With 5,000 phone calls a day and 45 seconds of conversation for identification and legitimation in human-to-human conversation and a success rate of 80 percent of automatic identification and legitimation, a voice messenger takes 5,000 x 0.8 x 45 seconds = 180,000 seconds = 50 hours from its human colleagues Talk time per day. This time can be used to complete more complex tasks.
Horizontal and vertical automation can of course also be linked: via the horizontal automation, the voice messenger determines callers and concerns and, in a complicated case, transfers them to a human colleague, as in the case of the complaint described. In the case of simple tasks, he takes care of the case himself, for example if the customer only requests a new invoice.
Most companies can benefit from chatbots, depending on the type of business and customer communication to different degrees.
The examples listed show that chatbots can be used in a variety of ways. Chatbots are undoubtedly helpful when it comes to a quick first response to customer inquiries, to “show presence” in support. Organizational processes such as bookings and purchase advice for online shopping with chatbots can also be mapped well if the processes are standardized.
However, there are areas in which chatbots will not make any sense in the foreseeable future or only if they become even more intelligent: Wherever dealing with special problems and individual questions is concerned, chatbots are inferior to human advisors because special experience, “gut feeling” and real transfer performance.
Specialized agencies offer their support as well as the operators of messaging services such as Google, Facebook and Microsoft. Facebook, for example, enables developers to create chatbots that communicate with customers and prospects within Facebook Messenger.
Microsoft’s Cortana Intelligence Suite uses technologies from the areas of big data, machine learning, analytics and largely automatic computer programs (“bots”). Developers and companies can use the bot framework and the Azure bot service to program self-learning applications and bots that communicate with their users in a personalized manner.
Other providers or solutions for your own chatbots can be found here, for example:
Chatbots should learn from user input in order to be able to provide more appropriate and personalized answers. Depending on the chatbot, user data is collected and evaluated, which the user did not enter, but which reflect his behavior on a website and his interests. If the user is not informed about this profiling and does not agree to it (informed consent), this is critical for data protection. Before deciding to use or use a chatbot, the respective data protection declaration must be checked.
Chatbots will cost jobs in certain industries and occupations. According to an Oracle study, chatbots can take over a large part of customer service sooner or later: 80 percent of the major brands will therefore introduce chatbots as customer advisors.
In the area of development and artificial intelligence, however, new jobs will be created if the trend towards the increased use of chatbots is confirmed. However, a lot can also be automated in the area of development.
Ipswitch’s report “Intelligent Systems in Action: The Rise of the Machines Has Already Begun” states that 76 percent of the IT experts surveyed expect intelligent solutions such as the bots to simplify IT processes, but 32 percent fear that at some point to be replaced by intelligent systems.