Levelling Up: AI-Powered Common Service Desk Chatbot

Generative AI offers a vast array of opportunities and drives innovation across multiple fields in #DTIT with rapid advancements like Chat GPT2 the horizon for new possibilities seems boundless. Naturally, it was time for our team, Workplace Experience, and our AI experts to evolve the Common Service Desk (CSD) chatbot, taking it to the next level, with CSD chatbot Next Level (NL).

What is CSD Chatbot?

CSD chatbot is DT IT’s digital assistant, designed to assist employees with IT related questions. Currently available to employees in Germany, it will gradually expand to the rest of our countries where we have presence, starting from next year. The chatbot provides 24/7 support, offering direct information with no waiting time, addressing questions related to the IT workplace, while it supports both German and English language.

Why do we need a chatbot?

A chatbot makes perfect sense because digitalization and automation are giving us smart, sustainable ways to reduce call volumes where it matters. Take incident handling, for example—bots can now handle basic troubleshooting, freeing up agents to focus on more complex problems. This not only saves time but also boosts efficiency by streamlining processes and making the most of the resources available. Automation helps simplify things, so teams can work smarter, not harder.

Moreover, by improving the user experience (UX), we believe that the chatbot can serve as a valuable instrument in resolving users’ issues faster and better, contributing to a more satisfying interaction overall.

The technology behind

The backbone of the new CSD chatbot lies on the platform #RASA CALM, a conversational AI platform to build and deliver next-level AI assistants. This platform is where the chatbot is built, trained, and fine-tuned. It utilizes LLMs such as ChatGPT model helping to orchestrate chat with users, even most of the dialog steps are predefined by chat content editors. With an enabled customization we could integrate into chatbot our own RAG (Retrieval Augmented Generation) solution using an in-house developed search engine called NIR (Neural Information Retrieval). This is being used to serve answers to How-Do-I type of questions with an internal knowledge base in the background.

CSD NL is more than an ordinary QA chatbot

CSD Next Level is more than just a typical QA chatbot—it leverages user-specific information (like contact details, device inventory details, etc.) to address issues more effectively. It can open incident tickets on behalf of the user, redirect chat conversations to a live agent when necessary, and even function as a QA bot, thanks to our RAG solution. This advanced functionality makes it a versatile tool that not only assists with troubleshooting but also enhances the overall user experience by streamlining processes. Another great benefit is that users can initiate certain requests through the chatbot on their private devices, even without Intranet access, as long as they have Internet connection. One such possibility is creating tickets in cases where Intranet access is no longer possible.

What’s Next?

The CSD Chatbot is continuously learning, and we expect many improvements in the coming months. Our statistics show a steady increase in usage, with more employees relying on the chatbot each week. In fact, the highest number of chat sessions in a single day has surpassed 900, serving 791 customers—an impressive milestone! Although it is not yet perfect, we are committed to its continuous development. The chatbot will evolve over time and become even more capable of meeting our users’ needs.

Our goal is to position the CSD Chatbot as the primary tool for employee assistance. With human agents in the background, the chatbot will be able to handle simpler issues through self-service, allowing users to resolve their problems more efficiently.

I would like to thank the team behind this amazing achievement: Slavomir Lihan, Roland Csibi, Marijana Vukoja, Roland Sandor , Heiko Dotzauer, Martin Lukan, Wladislaw Labizin, Branislav Popadic, Stanislava Zabalueva, Vivek Chandra, Kundan Jha, Chetna Pardeshi, Tünde K?teles , Ralf Bücker, Róbert Zsolt Nagy, Mónika Kerese, Christian Becker, Sandor Mitru, Eszter Csilla Schneller, Rita Bárdos, éva Vida, ?Zsuzsanna Hoffmann, ?Roland Goyák, Anette Dorothea Kálmán, Mária Jolán Kele ?and of course our sponsors and members of the DT IT IT Leadership, Omair Ahmed Khan and ?rs Cseresnyés.