A LOW-CODE APPLICATION DEVELOPMENT PLATFORM INTEGRATING AI-ASSISTED DEVELOPMENT
Keywords:
Low-code platform, AI-assisted development, Unified metadata model, Sandbox, Skills, MESAbstract
In the era of large language models, AI-assisted programming has substantially reduced the coding cost of application software, yet difficulties in communicating with clients and inefficiencies in deployment and maintenance persist. Rework and even project failure in real-world projects often stem from discrepancies between the client and the vendor in their understanding of the business, rather than from technology itself. To address this, this paper designs and implements a low-code application development platform that integrates AI-assisted development. Targeting three technical difficulties of current low-code platforms (front-end and back-end fields and rules are hard to configure dynamically, the security of online scripts is hard to constrain, and the reliability of AI-generated output is hard to guarantee), we propose three solutions: (1) a unified metadata model that automatically links front-end and back-end configurations; (2) a sandbox mechanism that restricts data sources, execution permissions and runtime; and (3) purpose-built Skills that constrain the steps and output format of AI generation. The platform is built on an improved RuoYi-Vue-Plus framework with an enhanced permission system, integrates the magic-api script engine, a visual design engine and sandbox technology, and builds its AI-assisted development capability on ByteDance's TRAE. Application in a manufacturing client's MES project shows that the development time of a conventional module fell from one week to 1–2 days, an efficiency gain of about 78%. The platform supports rapid on-site deployment and timely response to changing client requirements, effectively improving software delivery efficiency and success rate.References
[1] Chen M, Tworek J, Jun H, et al. Evaluating large language models trained on code. arXiv:2107.03374, 2021.
[2] Peng S, Kalliamvakou E, Cihon P, et al. The impact of AI on developer productivity: Evidence from GitHub Copilot. arXiv:2302.06590, 2023.
[3] Richardson C, Rymer J R. New development platforms emerge for customer-facing applications. Cambridge, MA: Forrester Research, 2014.
[4] Bock A C, Frank U. Low-code platform. Business & Information Systems Engineering, 2021, 63(6): 733-740.
[5] Di Ruscio D, Kolovos D, de Lara J, et al. Low-code development and model-driven engineering: Two sides of the same coin?. Software and Systems Modeling, 2022, 21(2): 437-446.
[6] Pearce H, Ahmad B, Tan B, et al. Asleep at the keyboard? Assessing the security of GitHub Copilot's code contributions//Proceedings of the 2022 IEEE Symposium on Security and Privacy (S&P). IEEE, 2022: 754-768.
[7] Sahay A, Indamutsa A, Di Ruscio D, et al. Supporting the understanding and comparison of low-code development platforms//Proceedings of the 46th Euromicro Conference on Software Engineering and Advanced Applications (SEAA). IEEE, 2020: 171-178.
[8] Hou X, Zhao Y, Liu Y, et al. Large language models for software engineering: A systematic literature review. ACM Transactions on Software Engineering and Methodology, 2024, 33(8): 220.
[9] Wang L, Ma C, Feng X, et al. A survey on large language model based autonomous agents. Frontiers of Computer Science, 2024, 18(6): 186345.
[10] Wahbe R, Lucco S, Anderson T E, et al. Efficient software-based fault isolation//Proceedings of the 14th ACM Symposium on Operating Systems Principles (SOSP). ACM, 1993: 203-216.