The provision of public services is a central objective of public authorities worldwide. Citizens and businesses consume public services either out of necessity (e.g., issuing a passport) or to receive benefits (e.g., welfare grants). To support this interaction, governments usually provide Public Service Description Catalogues (PSDC) that inform citizens/businesses about the available services. However, the provided descriptions are often long and complex, making the public service landscape difficult to navigate. As a result, providing personalized information becomes difficult and time-consuming. Such personalized information may include, for example, tailored information on the required evidence that must be submitted, based on the citizen’s profile. Given this complexity, the use of dialogue-based systems that provide personalized information seems a promising solution. However, the large number of available public services in PSDCs renders the manual ad hoc development of such systems per public service not scalable. To address this issue, this paper proposes an AI-driven approach for the automated production of dialogue-based systems using the structured information available in PSDCs. The proposed approach is demonstrated in the context of the Greek public administration, with a case study based on MITOS, the official national Greek PSDC.