How to gain knowledge in field service with the help of GenAI

Every day, companies let the important expert knowledge of their employees slip through their fingers. At a time when there is a massive shortage of skilled workers and baby boomers are retiring, this is a precious commodity. Frequently asked questions and helpful answers are already stored in chat systems, but are not processed and therefore not shared effectively within the company.

The first working day of service employee "Marcel" serves as a fictitious illustration:

Marcel starts his first day at work in the support team of a medium-sized industrial company. After reviewing his induction plan, he is ready for his first customer calls. He starts his computer and opens the new software he was introduced to this morning. The software plays a crucial role in handling the customer inquiries that Marcel and the team have to deal with.

This becomes clear when a few hours later, after Marcel has narrowed down the problem, all that remains is a detailed question about one of the customer's older machines. He initially tries to find the knowledge himself using a built-in search, but without success. He remembers his colleagues' recommendation: "If you don't know what to do, you have to ask Kerstin."

Kerstin is a helpful older colleague who has been with the company for so long and almost always has an answer ready. The service manager had raved about her that morning and talked about the gap she would soon leave in the company. Marcel forwards the customer inquiry to her without further ado. After just a few seconds, Kerstin gives a suitable answer. Marcel is impressed and hopes that Kerstin will never leave.

The service manager has faced up to reality and quickly converted the software into a digital assistant so that the knowledge from the chats is not lost and Kerstin's solution is preserved.

How does this work?

Solution knowledge in chats is automatically structured using language models and indexed in the knowledge database using artificial intelligence so that it can be easily found later. Kerstin concentrates on writing down her active knowledge and saving her solutions to difficult customer questions in the system and sharing them with everyone.

This fictitious example describes everyday service life for many medium-sized and large companies around the world. Generative AI (GenAI), especially in the form of Large Language Models (LLM), offers a real opportunity to automatically generate easily readable and efficiently usable knowledge from unstructured chats. Companies can develop specialized models that take into account the company's branding and style guidelines in order to generate consistent and brand-appropriate knowledge material.

Other artificial intelligence methods can be used so that the generated knowledge can later be applied by all employees. For example, NLP algorithms (Natural Language Processing) can be used to provide the knowledge with the correct metadata. This ensures that the information can be stored and reused in a meaningful way.

AI-supported search algorithms can also be used to provide the generated knowledge in a context-dependent manner and thus optimize access to relevant knowledge. Automatic machine translation is also useful in order to make knowledge directly usable for all users despite language barriers.

It is important to note that the use of GenAI is also associated with challenges. These include quality assurance of the generated knowledge, compliance with data protection regulations and ethical considerations when dealing with generated content. Companies should therefore take a responsible and careful approach to the implementation and use of GenAI solutions. In reality, specially trained models are required here - this is known as fine-tuning - in order to meet the high standards of such a system. General GenAI solutions such as ChatGPT or Google Bard are not able to do this.

Conclusion

Overall, GenAI offers promising opportunities to automate and optimize the knowledge generation process in companies to a large extent. By using LLM and other AI methods, companies can capture and structure their internal knowledge more effectively and make it accessible to all employees. This increases productivity, improves efficiency and better prepares the company for the challenges of our time. At the same time, it needs to be combined with other AI technologies, such as knowledge graphs, to ensure that knowledge is correct and not just eloquent.

Knowledge graphs work the way we humans think. They bring all information together in one place, link it and map it. They bring together all product information in one place - the "single source of truth" - link it and map it. In this way, they form the logical framework and source of knowledge, e.g. for customer service. Technicians and end customers can deduce affected parts, components and devices from error codes. This enables service technicians to solve problems more quickly, with the appropriate repairs or replacement with the correct spare part.

The combination of GenAI and knowledge graphs opens up new opportunities for knowledge management and field service support.