Large language models in software architecture

a case study of an intelligent tutoring system

Authors

  • Francisco Afonso Matos Pereira Júnior Universidade Federal Rural do Semi-Árido, UFERSA, Pau dos Ferros, RN, Brasil https://orcid.org/0009-0007-2531-2296
  • João Batista de Souza Neto Instituto Federal de Educação, Ciência e Tecnologia do Rio Grande do Norte, IFRN, Natal, RN, Brasil. https://orcid.org/0000-0002-8142-2525
  • Reudismam Rolim de Sousa Universidade Federal Rural do Semi-Árido, UFERSA, Pau dos Ferros, RN, Brasil https://orcid.org/0000-0001-9728-0130

DOI:

https://doi.org/10.47385/cadunifoa.v21.n56.6489

Keywords:

large-scale language models, software architecture, prompt engineering

Abstract

Software architecture is a crucial activity in systems development, defining the structure and qualities that directly impact project success. In this context, Large-Scale Language Models (LLMs) emerge with significant potential to assist and optimize various architectural tasks. This work investigated the potential and limitations of LLMs in supporting software architecture activities. The methodology includes a literature review of software architecture fundamentals, LLMs in software engineering, and prompt engineering, culminating in a guideline for using LLMs in software architecture tasks. Practical experiments were conducted, simulating tasks such as architecture proposal and evaluation, solution comparison, and generation of architecture documentation, focusing on the application of prompt engineering strategies. The results revealed the ability of LLMs to understand complex requirements and generate structured proposals, as well as standardized artifacts. However, they also highlighted limitations such as generalization into specific details, restricted context interpretations, and syntax errors, reinforcing the need for human supervision and refined prompts. The study concludes that collaboration between humans and LLM, in which AI acts as an intelligent co-agent and the human architect retains critical judgment, is the most promising approach to leverage the power of these technologies, despite inherent limitations.

Downloads

Download data is not yet available.

Author Biographies

Francisco Afonso Matos Pereira Júnior, Universidade Federal Rural do Semi-Árido, UFERSA, Pau dos Ferros, RN, Brasil

Graduante em Engenharia de Software pela Universidade Federal Rural do Semi-Árido, UFERSA, RN, Brasil.

João Batista de Souza Neto, Instituto Federal de Educação, Ciência e Tecnologia do Rio Grande do Norte, IFRN, Natal, RN, Brasil.

Professor no Instituto Federal de Educação, Ciência e Tecnologia do Rio Grande do Norte, IFRN, Natal, RN, Brasil.

Reudismam Rolim de Sousa, Universidade Federal Rural do Semi-Árido, UFERSA, Pau dos Ferros, RN, Brasil

Docente na Universidade Federal Rural do Semi-Árido, UFERSA, Pau dos Ferros, RN, Brasil. Doutor em Ciência da Computação pela Universidade Federal de Campina Grande.

References

AHMAD, A.; WASEEM, M.; LIANG, P.; FAHMIDEH, M.; AKTAR, M.; MIKKONEN, T. Towards Human-Bot Collaborative Software Architecting with ChatGPT. In: INTERNATIONAL CONFERENCE ON EVALUATION AND ASSESSMENT IN SOFTWARE ENGINEERING, 27., 2023, Anais [...] New York, 2023. p. 279–285. Disponível em: https://dl.acm.org/doi/10.1145/3593434.3593468. Acesso em: 2 ago. 2026. DOI: https://doi.org/10.1145/3593434.3593468

BROWN, S. Software architecture for developers: Volume 2 - visualise, document and explore your software architecture. [S. l.]: Leanpub, 2016.

BROWN, S. The C4 model for visualising software architecture. [S.l.]: Leanpub, 2022.

CAELEN, Olivier; BLETE, Marie-Alice. Developing apps with gpt-4 and chatgpt: build intelligent chatbots, content generators, and more. Sebastopol: O’Reilly Media, 2023.

DHAR, Rudra; VAIDHYANATHAN, Karthik; VARMA, Vasudeva. Can LLMs Generate Architectural Design Decisions? An Exploratory Empirical Study. In: IEEE INTERNATIONAL CONFERENCE ON SOFTWARE ARCHITECTURE, 21., 2024. Anais[...] Piscataway, NJ, 2024. p. 79–89. Disponível em: https://ieeexplore.ieee.org/document/10592785. Acesso em: 2 ago. 2026. DOI: https://doi.org/10.1109/ICSA59870.2024.00016

EISENREICH, Tobias; SPETH, Sandro; WAGNER, Stefan. From Requirements to Architecture: An AI-Based Journey to Semi-Automatically Generate Software Architectures. In: INTERNATIONAL WORKSHOP ON DESIGNING SOFTWARE, 1., 2024, Lisbon, Portugal. Anais [...] New York, 2024. p. 52–55. Disponível em: https://dl.acm.org/doi/10.1145/3643660.3643942. Acesso em: 2 ago. 2026. DOI: https://doi.org/10.1145/3643660.3643942

GANESH, S.; SAHLQVIST, R. Exploring Patterns in LLM Integration: a study on architectural considerations and design patterns in LLM dependent applications. 2024. Dissertação (Mestrado em Ciência da Computação) – Department of Computer Science and Engineering, Chalmers University of Technology, Gothenburg, Sweden, 2024. Disponível em: https://gupea.ub.gu.se/items/fa09b066-f5ff-4107-b6a6-40b394d2e26b. Acesso em: 2 ago. 2026.

HARRISON, N. B.; AVGERIOU, P.; ZDUN, U. Using Patterns to Capture Architectural Decisions. IEEE Software, Washington, DC, v. 24, n. 4, p. 38–45, jul./ago. 2007. Disponível em: https://dl.acm.org/doi/abs/10.1109/MS.2007.124. Acesso em: 2 ago. 2026. DOI: https://doi.org/10.1109/MS.2007.124

JAHIĆ, Jasmin; SAMI, Ashkan. State of Practice: LLMs in Software Engineering and Software Architecture. In: IEEE INTERNATIONAL CONFERENCE ON SOFTWARE ARCHITECTURE COMPANION, 2024. Anais [...] Piscataway, 2024. p. 311–318. Disponível em: https://ieeexplore.ieee.org/document/10628428. Acesso em: 2 ago. 2026. DOI: https://doi.org/10.1109/ICSA-C63560.2024.00059

OZKAYA, I. Application of Large Language Models to Software Engineering Tasks: Opportunities, Risks, and Implications. IEEE Software, Washington, v. 40, n. 3, p. 4–8, 2023. Disponível em: https://ieeexplore.ieee.org/document/10093056. Acesso em: 2 ago. 2026. DOI: https://doi.org/10.1109/MS.2023.3248401

SOMMERVILLE, Ian. Engenharia de software. 10.ª ed. Rio de Janeiro: Pearson, 2019.

VALENTE, Marco Tulio. Engenharia de software moderna: Princípios e práticas para desenvolvimento de software com produtividade. [S. l.]: Independente, 2020.

WHITE, Jules; FU, Quchen; HAYS, Sam; SANDBORN, Michael; OLEA, Carlos; GILBERT, Henry; ELNASHAR, Ashraf; SPENCER-SMITH, Jesse; SCHMIDT, Douglas C. A Prompt Pattern Catalog to Enhance Prompt Engineering with ChatGPT. In: CONFERENCE ON PATTERN LANGUAGES OF PROGRAMS, 30., 2023, Anais [...] Monticello, 2023. Disponível em: https://dl.acm.org/doi/10.5555/3721041.3721046. Acesso em: 2 ago. 2026.

Published

2026-09-15

How to Cite

MATOS PEREIRA JÚNIOR, Francisco Afonso; DE SOUZA NETO, João Batista; ROLIM DE SOUSA, Reudismam. Large language models in software architecture: a case study of an intelligent tutoring system. Cadernos UniFOA, Volta Redonda, RJ, v. 21, n. 56, 2026. DOI: 10.47385/cadunifoa.v21.n56.6489. Disponível em: https://revistas.unifoa.edu.br/cadernos/article/view/6489. Acesso em: 20 sep. 2026.

Issue

Section

Tecnologia e Engenharias