As part of the Build Developer Conference 2016, the Microsoft CEO Nadella surprised with the statement: “Bots are the new apps”. A year later, Gartner Group reaffirmed the trend with the announcement: “By 2021, more than 50% of enterprises will spend more per annum on bot and chatbot creation than traditional mobile app development.”
Today, bots and chatbots can actively support the call flow process in a contact center on each level or even work independently.
There are many possible applications of bots within a contact center (list is not exhaustive):
Bots are not yet able to replace an agent. They lack the ability to answer all the questions, as well as empathy, in order to react appropriately in every situation.
In the article “Man and Machine in Customer Service of the Future”, Dr. Hansjörg Leichsenring sums it up perfectly: “Increased automation through artificial intelligence is on the rise. Man, however, remains a central element in direct customer contact».
With bots, customer data can be digitized and recorded centrally. This is the basis for being able to conduct topic and trend monitoring. It also gathers customer experiences, which is useful for training bots to increase their responsiveness and the quality of their responses.
The goal is to relieve contact center staff of simple routine tasks and give the agents time for real customer service. Many requests that are made to a customer service can be averted by a pre-selection by the agent. Previous preselections such as language versions are today standard and make consultation more effective.
A preselection (for example, according to language variants or topics) is not new and is made possible by IVR systems. However, these systems often leave a negative impression and don’t usually reduce the caller’s waiting time. An artificial assistant, which acts as a natural communication partner, not only allows the assignment of the caller to certain topics and thus to expert knowledge, but also partly solves inquiries independently. Additionally, the customer is accepted directly by the artificial assistant without having to wait and the annoying queuing is kept to a minimum.
Even if a chatbot is a continuously learning system, the first version must already be able to answer the central inquiries, otherwise a negative customer experience is generated. A customer who is not helped will not use the communication channel again and as a result the bot lacks conversation data which prevents it from self-learning and improving. Existing support inquiries, which were received via chat or web chat, are suitable to teach a system. Getting started with the new technology is made easier by the fact that chatbots can adopt the nature and language of the conversation. But also content of phone calls and emails are suitable for improving the ability of the bots.
Dealing with unknown topics and questions must be clarified before using chatbots. This also applies to questions that cannot be answered by the bot, or when it is determined by the customer’s reaction that he is dissatisfied. It helps that the systems available today are capable of learning. They can analyze customer behavior (for example, by classifying or recognizing words, but also complete sentences) and decide how to react to negative statements.
Equally as important is the analysis of support requests, which are common. Depending on the industry, there are recurring terms, places, people or organizations. Being able to extract these is very helpful for the system and the further processing of a query. If bots allow this knowledge to flow into the answer, it not only impresses the customer, but also offers real added value.
Bots must be able to conduct a natural conversation with the customer and classify requests independently. During interaction with the customer, the bot must be able to gather new information and decide whether the request has already been answered or whether the customer needs to be redirected to an agent. Today, the systems available on the market are capable of taking decisions of this nature.
The dream of speech recognition is over half a century old. But machines have only able to recognize simple commands for a long time. The capacities needed for processing were expensive. Cloud solutions lower the price of data storage. There are enough processing capacities, better algorithms and unlimited computing power.
This has an impact on the learning ability of the machines. They can be trained on real data. As a result, bots are able to fulfill their task in the best possible way and learn as they go along. The question asked in the title “Man and machine – ideal symbiosis” can be answered with YES. Today, using bots is the right way to go forward if the machines can relieve agents of tasks or support conversations with customers with additional information. However, this is only possible if the complexity of the requests is not too high and the systems have been equipped with the necessary knowledge through training beforehand. The acceptance largely depends on how people see themselves supported by the machines and with what quality and speed inquiries are processed.
With the Microsoft Bot Framework, Luware is able to develop its own bots for the Luware applications based on Microsoft Skype for Business and Microsoft Teams. The framework allows Luware bots to be built which listen and understand people, but also answer questions and provide information. First bots are currently being tested and further developed in pilot projects.
As an innovation leader in the areas of contact center and collaboration, driven to pick up new trends and develop and implement these for its customer.