The AI journey of DKB Service GmbH - PART 2: GenAI as the key to success in customer service

DKB Service GmbH's AI support mainly draws on existing knowledge modules and is specifically linked to them. The prerequisite for this is the internal knowledge platform with all the information that the real agents currently use in their day-to-day work. The knowledge of AI support is enriched by FAQs, website data and existing text modules. A few AI-specific modules complete the functional AI solution.

Regardless of how our customers contact us, they receive consistent answers. The AI also recognises when it can no longer help customers and decides which further solutions or contact options it can offer. In this way, the AI not only pools information and makes it available to users, but also guides them to the next step. Whether this is a telephone contact, the use of a self-service or a link to a specific form is decided independently by the AI support team on a case-by-case basis.

There may be moments in the chat with AI support when the conversation can no longer be continued in a meaningful way. For example, users ask the same question several times, a next step needs to be taken to solve the problem or the user's mood changes. At these points in the conversation, the user and AI are unable to reach a solution and find themselves in a kind of "continuous loop". In order not to overstretch the situation, AI Support will suggest a concrete and expedient 'exit' at sensible points.

AI Support can either direct the user to a self-service or to a form link, offer the telephone number to our Customer Service or support the customer with a written enquiry via the contact form. In this case, AI Support creates a summary of the request and customers can check this summary, adapt it if necessary and then send it directly as an enquiry.

DKB Service GmbH, together with its parent company DKB AG and its sister company DKB Code Factory, set up a team of experts right at the start of work on the AI Support solution. Agile and iterative collaboration, which incorporated customer opinions, wishes and evaluations throughout the entire development process, is still the recipe for success today. Our customers' initial ideas have already been incorporated into the development of the MVP (Minimum Viable Product). The initial focus here was on requirements such as accessibility, simplicity and solution-orientation. To this day, we continue to refine the scope and services as well as the linguistic and rhetorical skills of AI support with the support of our users and experts.

A human-centred use of GenAI requires focus and time. Time to understand, time to develop and time to adapt.

The journey of using AI at DKB Service GmbH is not yet complete. Making the solution fully available to all customers and non-customers continues and requires further expansion stages and technical load tests.

Availability in web banking, on the website and also "To-Go" in the banking app will be successively scaled and new features added. In addition, further investment is needed in intelligent data management and a scalable IT infrastructure. Since 2020, DKB has invested an additional 400 million euros in digitalisation and has created important prerequisites for the use of GenAI. These include the use of cloud infrastructure. We are also working on our own AI governance and the integration of an AI strategy into our business strategy. In doing so, we are laying important foundations for the future use and further development of AI-supported services in the Bank in the context of the requirements of the EU AI Act. This journey has only just begun for us and still harbours enormous potential.

Read also PART 1