Low-code platforms are gaining momentum, allowing the creation of complex applications directly from the browser. This allows their use by individuals with a wide range of backgrounds, but they pose some problems. On the one hand, newcomers may want to grasp the platform capabilities or pinpoint where can they access some functionality, which may prove difficult without guidance. On the other hand, their navigability can be complex: their functionality can be distributed across many webpages and user interfaces, each managing different concepts. Additionally, users may deem more convenient addressing some tasks using natural language instead of navigating visual interfaces. For these reasons, we introduce LowcoBot, a model-driven solution to generate LLM-based chatbots out of low-code platform design models. We demonstrate its capabilities by generating a chatbot for Dandelion+, a low-code platform built within an industrial context, and showing the range of tasks it supports out-of-the-box.