As part of my AI Trend Report, which will be published at the end of February 2024, I looked at a large number of Generative AI projects and analyzed them in terms of the opportunities and possibilities of automation and collaboration between humans and machines.
In this article, I present a successful practical example from the Unite Group, together with its partner BSI.
Initial situation and challenges before the introduction of AI at Unite
The company has an enormous variety of products, which makes it difficult to provide a service-oriented and efficient service, as not every product and the supplier behind it can be known to employees without prior research. In addition, the customer structure is extremely heterogeneous, with small commercial customers, public sector clients and large corporations as stakeholder groups on the platform. Every month, Unite receives up to 80,000 inbound contacts, mainly by e-mail, while telephone inquiries are on the decline. Due to the international coverage, inquiries have to be processed in several languages, which is why 10 foreign languages are spoken in customer service at the Leipzig location. In addition, many new order requesters with individual requirements have to be taken into account in the shortest possible time. Individual opinions and individual needs must also be recorded and processed. In view of Unite’s growth ambitions, the solutions must also be scalable. These diverse and complex challenges form the starting point for the introduction of the new automated solution for customer relationship management.
About Unite Network SE
Unite Network SE is a group of companies based in Leipzig/Germany, which was founded in 2000 and is active in twelve European countries. Mercateo, the B2B trading platform belonging to the Unite Group, gives commercial customers access to an integrated marketplace product range and the suppliers in the Unite network. In Germany alone, this range comprises more than 66 million items, which are listed in hundreds of supplier catalogs. As a platform operator, Mercateo always remains supplier-neutral and does not have its own warehousing or prioritization of articles through advertising budgets. Mercateo also offers an e-procurement portal with a wide range of interaction and transaction options for buyers, suppliers, manufacturers and service providers. In 2022, Unite generated sales of 447.7 million euros and employed more than 700 people.
The solution with Generative AI
Unite has opted for automation in customer service with the Customer Suite from BSI Business Systems Integration AG, which addresses the company’s challenges and at the same time offers a high level of performance and scalability in day-to-day operations that is appreciated by employees and customers alike.
Artificial intelligence never takes over the service process completely, but provides advice in the form of an AI companion. The recommendations for action made by the AI are finally checked by the employees before the responses generated by the AI are sent. People at Unite therefore take on a “gatekeeper function”, which allows them to delegate individual process responsibility to artificial intelligence.
In practice, artificial intelligence at Unite ensures high efficiency and empathy in customer communication. The LLM not only automatically recognizes the content and tone of a conversation, but also creates response suggestions in real time that are complete, empathetic, spelling and legal compliant and individual. And this in one of 10 languages spoken at Unite. It is also possible to train the artificial intelligence to use different tonalities for different areas of application and have these output automatically depending on the situation.
The following screenshots show the process from the receipt of a new customer inquiry to the response to this inquiry. At the beginning, Generative AI summarizes the customer request for the consultant. The AI then suggests a possible “next best action”. This best action can be further customized by the consultant. The advisor can give further commands to the AI in the form of prompts. In the end, it is still up to the individual to decide which answer to send to the customer.





As many people as necessary, as much automation as possible.”
(Thomas Zoller, COO at Unite)
The advantages of using Generative AI
The use of artificial intelligence in customer service offers Unite a number of advantages. The high efficiency and speed of the automated response to customer inquiries is particularly evident for standard questions that have already been trained in advance using the Unite language model. Automation also ensures consistent service quality, as human errors in standard processes are minimized as far as possible. The efficiency that comes with automation allows employees to concentrate on value-adding and specialized tasks. This in turn leads to greater employee satisfaction, as they can focus on more fulfilling tasks instead of routine activities. Automation also offers the opportunity for professional development through specialization and further training. These advantages make the use of the new solution in Unite’s customer service extremely attractive.
More quality, thanks to the right balance between employees and AI
Unite takes a balanced approach, focusing on the balance between people and machines in customer service. The company aims to automate customer service as far as possible without compromising quality. More complex tasks will continue to be handled by Unite experts in person with the customer. Despite automation, customer service employees can respond individually to emotional needs and offer empathetic advice, particularly in the area of complaints management. Self-service options for customers are only offered if they offer real added value, such as tangible time savings. This balanced approach ensures optimal customer care and customer satisfaction at Unite.
With open source language models for more security and data protection
In addition to satisfied customers and employees, data protection is also a top priority for Unite. Unite ensures that all customer data remains in its own system environment and is not passed on for training purposes. This measure ensures the protection of data and compliance with data protection regulations. By implementing context limitation techniques, Unite minimizes potential risks such as hate speech or hallucinations. This ensures the safe and responsible use of artificial intelligence. Unite can also ensure a high level of consistency in output to provide a reliable customer experience. Unite has therefore decided against the use of ChatGPT and uses open source language models. The use of open source LLMs complies with the requirements of the EU AI Act. This approach demonstrates Unite’s commitment to the responsible and data protection-compliant use of artificial intelligence in customer service.
More cases and trends?
Further practical examples of Generative AI, trends and current figures can be found in the Generative AI Trend Report.
The Swiss AI tool SwissGPT from AlpineAI was used for this blog post. SwissGPT has created appealing and coherent paragraphs in continuous text from individual keywords or colloquially worded paragraphs.
