Demand for software outruns the supply of engineers, so citizen developers build too, on low-code and no-code platforms. This paper reviews the literature on putting AI, especially LLMs and LLM-based agents, inside those platforms: what it changes, what it costs and where the research is thin. It closes with gaps worth studying: task-technology fit, evaluation metrics and proactive assistance.
01The gap
Low-code and no-code platforms let people with little programming background assemble applications from visual parts 1. Professional engineers, meanwhile, use AI for code generation and recommendations across the development lifecycle. The two groups have different skills and different tools, and the question is what happens when the same AI sits in front of both.
02How AI gets in
The review lists five AI capabilities arriving in low-code platforms: natural-language interfaces, intelligent search for components and functions, code generation, automation of development tasks such as testing, and friendlier AI-assisted interfaces 6. New generative-AI features keep being announced 3. 1
03What the literature reports
Reported benefits are efficiency gains in time, effort and cost, and better collaboration between citizen developers and engineers 2. One user study found no significant difference in correctness or time between programmers and citizen developers on simple applications 2. Reported challenges fall into three groups: user skills (novices struggle when their goal is unclear), technical limits of the platforms (fragmentation, security), and organisational factors such as role conflicts and governance 4.
04Open questions
The review names six research directions. Three stand out. Task-technology fit: how AI features match different development tasks. Performance metrics: citizen-developer success has no standard measure. Proactive assistance: AI that offers help in the user's workflow instead of waiting for a prompt, which links to LLM-based agents for software engineering 4. The others are better support for novices, clearer digital-skill definitions, and AI in platform design.
References
- 1. M. O. Ajimati, N. Carroll, M. Maher. Adoption of low-code and no-code development: a systematic literature review and future research agenda. Journal of Systems and Software 222, 2025.
- 2. T. Guthardt, J. Kosiol, O. Hohlfeld. Low-code vs. the developer: an empirical study on the developer experience and efficiency of a no-code platform. MODELS Companion '24, 2024.
- 3. O. Bruhin, E. Dickhaut, E. Elshan, M. Li. The rise of generative AI in low code development platforms: an analysis and future directions. 2024.
- 4. H. Jin et al. From LLMs to LLM-based agents for software engineering: a survey of current, challenges and future. 2025.
- 5. B. Yetiştiren, I. Özsoy, M. Ayerdem, E. Tüzün. Evaluating the code quality of AI-assisted code generation tools. 2023.
- 6. G. Kacheru, N. Arthan, R. Bajjuru. Artificial intelligence (AI) for low-code and no-code development: making non-developers developers in 2024. Formosa Journal of Multidisciplinary Research 4(1), 2025.