Artificial intelligence improves customer service
The use of innovative chatbots and artificial intelligence revolutionizes the customer experience while increasing efficiency.
Commerzbank
When customers reach out to their bank, they expect quick professional answers. Customers compare service and response times not only with those of other banks, but with all service providers that they use – from Amazon to Zalando. "Our customers want to be taken care of wherever they are – and at any time. Digital assistants have such a degree of maturity now that they can provide helpful answers to many questions relating to daily banking and can create real added value," explains Marc Ernesti, Managing Director of Commerz Directservice GmbH, Commerzbank's central customer dialog center for private and business customers. The advantages are obvious: co-workers with customer contact can be freed from simple, repetitive customer inquiries and use that time for sales and distribution as well as more complex customer issues.
Use of artificial intelligence
As with most innovative technology concepts, the English term Conversational Artificial Intelligence (AI) has become established in German-speaking countries. What exactly is behind this becomes clearer when you look at how it’s used. "Our chatbot is available 24/7 and responds to questions in a natural language. It can also answer more complex questions via specific dialogs. In addition, we’re constantly working on training the bot's responses to increase its accuracy and to include new topics," says Marc Ernesti. By using neural networks, a computer-simulated "brain" is created that can understand and process the various statements made by users. The development of this technology is complex and requires not only bank-specific knowledge, but also linguistics and logic, coupled with programming expertise. Google's expertise was used here. The first step was to clearly identify the socalled intents, i.e. the relevant intentions or goals of the customers. They’re crucial for success, as possible questions and the corresponding answers are assigned to them, programmed and then extensively tested. The result is intelligent programs, known as chatbots (= chatting bots, communication robots), which can communicate independently with users. An example, people use Siri or Alexa, which are also bots in the broadest sense, to find out the current temperature and weather conditions (intent). The questioned bot provides the answer, regardless of whether the user asks: "What's the weather going to be like?", "How cold is it?" or "Do I need an umbrella?"
Developing a customer chatbot
In just 9 weeks, the team was able to develop and test an initial and already reliable chatbot in an agile project. The Conversational AI Platform from Google called Dialogflow CX was used for this. "Our chatbot Bene is already able to respond appropriately to more than 250 different customer requests and the number keeps rising. It helps with common questions relating to online banking, cards, accounts and payment transactions, for example. It also directs customers directly to service routes and relevant functions in the app if required. If the customer question needs to be answered by an employee, there’s a seamless transfer to a service employee," explains Marc Ernesti. The programmed customer queries cover the full range of standard queries and range from "change address" to "manage online banking access". The chatbot can also actively make product suggestions when appropriate. Just like human service employees, it will also learn and improve with every interaction. We now cover all topics with a comprehensive range of intents, around 30 of which are designated employee intents; these concerns are always addressed to employees. Products are only offered to customers if there is active demand from them or if this demand crystallizes through queries. The bot learns through targeted training through us, not directly through interactions with customers, so we have full control over the learning process.
Chatbot integrated into customer dialog platform
"The launch of Bene required the connection of Google Dialogflow CX to the dialog platform from Genesys - a state-of-the-art platform from Commerz Directservice that is the technological core for orchestrating customer contacts of all kinds. The connection of bot and dialog platform is a milestone for us in efficient, customer-centric and secure customer communication," explains Marc Ernesti. The current architecture runs with a large number of applications on onsite servers and components on Google's cloud platform. The advantages of the cloud also apply here: flexible use of resources, increased system reliability and compliance with data security.
Next stage of development in collaboration with Microsoft
To take the customer experience to the next level, the Bank is planning a banking avatar based on artificial intelligence and recently launched a project for this purpose. The banking avatar is a virtual assistant in the form of a digitized person who interacts with customers in natural language.
Marc Ernesti
Marc Ernesti has been working for Commerzbank for 26 years. After various management functions in sales and a 2-year stay at the headquarters, he became Managing Director of the CDS in summer 2021. His main areas of focus are operative customer business, process management and overall IT management, where the Conversational AI division is located.