The COVID-19 pandemic led to a rapid shift in customer behaviour, especially in the area of online sales and service requests. The DIY chain toom faced the challenge of how to effectively deal with these unforeseen changes in customer behaviour. Its focus here was on improving service through technical innovation. Finding a way to satisfy the increasing demands of customers while also creating a scalable solution was the goal.

As part of an innovation project, toom worked with adesso to develop solution concepts for digital services that were really innovative. Once this was complete, the next step was to build and expand on them to create digital prototypes. This is how the service companion came into being. In-depth testing ensured that the solution met the highest standards in terms of functionality and user experience.

The integrated conversational AI solution delivered by adesso includes several systems to support customers and employees. The central component of the solution is the ‘Bit’ chatbot, which was seamlessly integrated into the backend systems and is available on the website or as a voice bot over the phone. This innovative technology enables standard requests to be processed and generally closed at any time, day or night.

The concept of ‘human hand-over’ comes into play when the chatbot reaches its limits or if the request requires additional attention by a real person. In such cases, the requests are pre-classified and transferred accordingly to a live chat with the service call centre or community advice desk. This way, customers receive personal support from qualified specialists, while the chatbot continues to automatically process standard requests.

Routeing and semi-automated responses to requests over other channels are also possible with the conversational AI platform. This relates, for example, to the handling of Trusted Shops ratings. Interactions with customers are smooth and efficient in part because the platform is seamlessly integrated into a variety of communication channels.

The project was carried out as an agile contract for services in two phases:

• The strategy for and scope of the chatbot were defined at a series of concept workshops. The project was carried out together with a team of experienced method coaches, digital designers and chatbot professionals from adesso.
• In phase two, toom developed the relevant system content. The team from adesso provided coaching to employees working for the central customer service team. This put them in a position where they were able to train fellow staff members and configure the AI solution on their own, living up to toom’s motto: ‘Respect for those who do it themself’.

adesso was given the task of shaping the personality of the ‘Bit’ chatbot. The chatbot was integrated into toom’s website and connected to the various backend systems, including the product and market information system, the CRM system, the contact centre software and the advantage card system with single sign-on.

A chatbot widget was created and adapted to the website. It includes a link to Google Maps, incorporates the community live chat function and boasts integrated forms. In addition to that, live chat was implemented with the aid of the contact centre software.

To ensure efficient chat management, large language models were used in addition to rule-based chats.

In addition to the website, voice/phone bots are available as other channels of communication, with additional options in the works. Thanks to the adesso cognitive communication centre, automated request routeing is also available for asynchronous communication. NLP (natural language processing) methods and large language models (LLMs), including GPT-3, are used to classify and generate suggested responses.

The whole thing can also be run as an SaaS solution, including ongoing upgrades to incorporate additional use cases and the integration of further channels and backend systems. toom also receives mentoring for training the chatbot and chat design as required.

The new service companion makes it possible to respond to standard requests day or night in self-service. Customers can access key information whenever they need it. Frequently asked questions are answered quickly, while more complicated queries are expedited by escalating pre-qualified cases. Customers receive personal support and recommended solutions from employees in the live chat.

Shop employees can also access backend systems using voice bots that assist them in their day-to-day work. This leads to more efficient communication, speeds up the exchange of information and optimises work processes, the end result being higher employee satisfaction.

The (partial) automation of asynchronous communication processes ensures that requests can be processed more quickly. In addition, there is also a constant line of communication with customers.