EUVIMEDEUVIMED entdecken
Crossref

A Modular Multi-Agent {LLM} Architecture for Text-to-Diagram Generation and User-Guided Refinement

Hartwig Grabowski

e-Informatica Software Engineering Journal · 2026 · 20 (1) · 260109

DOI: 10.37190/e-inf260109 ↗

Kurz gefasst

Fachartikel in der Zeitschrift e-Informatica Software Engineering Journal, 2026. Verfasst von Hartwig Grabowski.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Context

Recent advances in LLM-based diagram generation increasingly rely on coordinated agent systems rather than single-model prompts.

Objective

This work highlights how modular multi-agent architectures improve reliability, semantic grounding, and iterative refinement in text-to-diagram workflows.

Method

We analyze a pipeline composed of specialized agents for interpretation, synthesis, validation, and correction, each contributing a bounded and inspectable transformation.

Results

The agent system provides deterministic validation, structured reasoning, and controlled refinement loops that outperform monolithic LLM generation.

Conclusions

Multi-agent LLM pipelines represent a robust foundation for precise, verifiable diagram generation and serve as a reproducible alternative to single-pass text-to-diagram models.

Zitieren

APA

Grabowski, H. (2026). A Modular Multi-Agent {LLM} Architecture for Text-to-Diagram Generation and User-Guided Refinement. e-Informatica Software Engineering Journal, 20(1), 260109. https://doi.org/10.37190/e-inf260109

Vancouver

Grabowski H. A Modular Multi-Agent {LLM} Architecture for Text-to-Diagram Generation and User-Guided Refinement. e-Informatica Software Engineering Journal. 2026;20(1):260109. doi:10.37190/e-inf260109

BibTeX
@article{Grabowski2026,
  author = {Grabowski, Hartwig},
  title = {A Modular Multi-Agent LLM Architecture for Text-to-Diagram Generation and User-Guided Refinement},
  journal = {e-Informatica Software Engineering Journal},
  year = {2026},
  volume = {20},
  number = {1},
  pages = {260109},
  doi = {10.37190/e-inf260109}
}
Kostenlos starten →