Low-Code/No-Code (LCNC) platforms aim to democratize software development, but non-expert users often face challenges with usability, learnability, and maintainability. This study investigated if an AI-powered chatbot could enhance LCNC platform accessibility. Following an iterative design process with multiple prototypes, we conducted a between-subjects user study (n = 26) comparing a standard LCNC interface with one augmented by a domain-knowledgeable AI chatbot. Results showed significant improvements with the chatbot: System Usability Scale (SUS) scores were higher (p = .03), task completion was faster (p<.001), and participants needed less external help (p<.001) while making fewer errors (p<.001). Qualitative data indicated strong user trust in the chatbot, an appreciation for its concise explanations, and better feature discovery. While most users preferred a hybrid approach, some experienced users eventually favoured the visual builder alone. These findings demonstrate that integrating AI chatbots can substantially improve LCNC accessibility, offering key design implications for their optimization.