AI in Customer Support: Enhancing Efficiency, Collaboration, and Service Quality
By introducing AI into our support processes, we are raising the bar on both speed and quality of service. Our automated system handles routine inquiries instantly, so our experts are available to help you with more specific needs. At the same time, AI fosters seamless international collaboration by bridging language and cultural barriers, making expertise and solutions globally accessible.
Our Methodical Approach
The introduction of our AI solution began with a thorough evaluation of various public providers. To test the effectiveness of the potential AI models, we conducted ten product specific tests. The two providers that delivered the highest rates of correct and helpful responses were Google Gemini and Azure OpenAI.
The technical integration was carried out via the REST API of our TOPdesk customer support ticketing system. A decisive success factor during implementation was prompt engineering. The quality of the answers depends heavily on the precision of the prompts given to the AI. By carefully formulating these instructions, we ensured that the AI delivered not only fast but also high-quality and relevant responses.
Tool Calling: Expanded Knowledge for Better Answers
One of the most important features integrated into the AI solution is the so-called tool calling. This function allows the Large Language Model (LLM) to use external tools, such as Google Search. Through this connection, the AI can access current and expanded data sources. This improves response accuracy for time-sensitive or complex queries and ensures that answers are not only based on static training data, but also take into account dynamic, up-to-date information.
RAG and Internal Knowledge Databases
Further development of the AI solution is already planned. A central element of the next phase is the integration of Retrieval Augmented Generation (RAG). RAG will connect the AI models from Google Gemini and Azure OpenAI with another key data source: Wibu-Systems’ internal documents. These include manuals, guides, and other exclusive resources. This strategy ensures that the AI can access the company’s consolidated internal knowledge. By combining the broad knowledge of LLMs with specialized in-house expertise, the quality of answers will be improved further. To ensure optimal performance, parameters such as chunking (splitting documents into smaller sections) and document indexing will be carefully fine-tuned.
Field Report: First Deployment of AI in Support
Our first AI-powered support deployment exceeded expectations and delivered faster, smarter assistance for our customers from day one. Particularly striking was that the costs for implementation and operation were significantly lower than expected. This greatly lowered the barrier to entry and showed us that AI solutions can deliver great value even with a manageable budget.
Another success factor was that public providers such as Google Vertex AI or Microsoft Azure OpenAI Service came with an impressive baseline understanding of our product information. This allowed us to achieve results very quickly without time-consuming, extensive training.
Our Areas of Application in Everyday Work
- Ticket Summarization: AI generates precise and understandable summaries, speeding up processing.
- Translations for International Subsidiaries: Language barriers are almost completely overcome through fast and high-quality translations.
- Automatic Creation of Knowledge Base Articles: Recurring solutions are systematically documented and made available to the entire team.
- Automatic Incident Qualification: Standard cases can be quickly pre-filtered by AI, enabling staff to focus on more complex issues.
AI is unlocking powerful new opportunities to serve our customers better, from faster more efficient support to smarter collaboration and real-time knowledge sharing. However, there are also limits: especially with complex issues involving multiple interconnected aspects, AI still reaches boundaries. Human expertise remains indispensable to evaluate context correctly and develop holistic solutions.
Ongoing Steps for Introducing AI in Customer Support
Automatic Responses for Customer Inquiries
End user requests will initially be answered automatically by AI. This will allow us to cover standard questions and recurring topics faster, freeing up valuable support time for more complex cases.
Optional for Customers
The first AI-generated responses will be available to all customers. If you no longer wish to use this feature, simply let us know, and we will deactivate it for you.
Special Option for Support Contract Customers
Customers with support contracts will by default not receive automatically generated answers, ensuring they continue to enjoy the familiar, personalized service. However, if desired, we can also activate AI-based first responses for this customer group, giving them the added benefit of immediate and rapid reaction times – even outside of business hours.
Future: Automated Assistance in the Ticket Process
Our next step is to have AI actively support ticket processing to provide you with faster and more accurate assistance. It will automatically request additional supplementary information such as logs or CmDust files before a support agent takes over the ticket. This will significantly reduce processing times, as the necessary preparatory work will already have been completed in advance.
The introduction of AI in our support organization marks a decisive step toward a new service culture. Customers benefit from faster, round-the-clock responses and higher-quality solutions, while our support staff gain the freedom to focus on complex issues instead of routine tasks. At the same time, AI connects our global team beyond language and cultural boundaries, making knowledge and expertise universally accessible and enabling truly borderless collaboration.
KEYnote 50 - Edition Fall/Winter 2025