Generative AI enables non-programmers to build functional apps through natural language, a practice termed «vibe coding» (Karpathy, 2025). Platforms such as Lovable.ai make it accessible, yet turning a pedagogical idea into an effective prompt requires structure most teachers lack. This communication reports the first field use of Wizziapp, a conversational seven-stage wizard that guides teachers through pedagogical and technical decisions and turns their answers into a well-formed prompt. It was piloted in an online workshop with in-service and pre-service teachers from Bolivia, following a Design-Based Research rationale. Of 78 participating teachers, just over half completed the wizard and generated a prompt. Post-use surveys indicate favorable reception, low to moderate difficulty, and a large majority who would not modify the prompt before submitting it. Within low-code/no-code literature, results offer exploratory evidence that structured wizards can lower technical barriers for teachers, while raising questions about prompt reinterpretation and transparency.