The events team at a global pharmaceutical company coordinates room bookings and events across multiple systems (a ticketing system, a document platform, and a calendar system) and locations. At the main Swiss site alone, more than 2,000 requests were processed in 2025. Many concern simple, recurring questions such as room availability, generating repetitive email exchanges and manual effort, despite a lean internal team supported by external delivery partners.

Semi-structured interviews with the booking and planning teams and management, combined with a full year of ticket data, were used to identify recurring request types and requirements. Technical feasibility was assessed through API documentation and a self-built, production-deployed integration between two of the existing systems. Findings were consolidated into a prioritised requirements catalogue and a target architecture recommendation.

Nearly half of all support tickets concern room availability alone, and the three most common request types together account for roughly two-thirds of total volume, a strong indicator that a chatbot could meaningfully reduce first-line workload. Interviews confirmed that a chatbot should inform, collect and pre-qualify requests, while final booking decisions and complex support-model matching remain with human staff. No fundamental technical barrier was found to integrating a chatbot with the three existing systems; the main prerequisite is improving the data quality of the ticketing system's intake form. The thesis recommends a combined architecture that reuses all three systems rather than replacing them, and proposes a time-boxed pilot at the main site as the first implementation step, followed by a gradual, validated rollout to other locations once cost and governance details are formally confirmed.