Turning Low-Code Development Platforms into True No-Code with LLMs

The relevance of low-code / no-code development has grown substantially in research and practice over the years to allow non-technical users to create applications and, therefore, democratise software development. One problem in this domain still persists: many platforms remain low-code as the underlying modeling layer still requires professionals to write/design a model, often using DSLs. With the rise of generative AI and \glspl{LLM} and their capabilities, new possibilities emerge on how LCDPs can be improved.

This paper shows how the capabilities of LLMs can be leveraged to turn DSL-based low-code platforms into true no-code. We analyzed how textual modeling can be replaced by generating the required model using LLMs. We conducted a user experiment to compare textual modeling with the use of LLM for that task. Our results show that task completion time could be significantly improved, and the majority of users prefer using the LLM-aided modeling. Based on these findings, we discuss the integration of these techniques into an existing low-code platform to transform it into true no-code.