Large Language Models to Enhance Business Process Modeling: Past, Present, and Future Trends

2026-04-16 · Source: cs.AI updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Operations & Process Management · Depth: Advanced, extended

Summary

This article reviews AI-driven methods for transforming natural language into Business Process Model and Notation (BPMN) and related workflow models, focusing on Large Language Models (LLMs). The analysis, based on a structured literature review of 41 papers from 2015 onwards, reveals a shift from rule-based and traditional NLP to LLM-based architectures. These LLM approaches leverage prompt engineering, in-context learning, knowledge injection (including Retrieval-Augmented Generation or RAG), domain adaptation via fine-tuning, and intermediate representations like JSON or DSLs for BPMN generation. While LLMs enhance capabilities, challenges persist in semantic correctness, evaluation fragmentation, reproducibility, and real-world validation. The review identifies research gaps, emphasizing the need for integrating contextual knowledge, developing interactive modeling, and establishing standardized evaluation frameworks.

Key takeaway

For Directors of AI/ML overseeing business process automation, recognize that while LLMs offer significant advancements in converting natural language to BPMN, their current implementations often struggle with semantic correctness and real-world complexity. Prioritize solutions that integrate Retrieval-Augmented Generation (RAG) for contextual grounding and support iterative, human-in-the-loop refinement to ensure generated models accurately reflect organizational processes and evolving requirements. Your teams should focus on developing robust validation frameworks beyond simple syntactic checks.

Key insights

LLMs significantly advance text-to-BPMN generation but face challenges in semantic accuracy, evaluation, and real-world applicability.

Principles

• LLMs excel at interpreting text and generating structured outputs.
• Iterative refinement is crucial for complex process modeling.
• Contextual knowledge improves LLM-based process model accuracy.

Method

LLM-based text-to-BPMN transformation often uses prompt engineering, in-context learning, and knowledge injection to generate intermediate representations, which are then converted to BPMN.

In practice

• Use RAG to incorporate organizational documentation for context.
• Employ intermediate representations (JSON, DSLs) for validation.
• Implement feedback loops for iterative model refinement.

Topics

• Business Process Modeling
• BPMN
• Large Language Models
• Retrieval-Augmented Generation
• Prompt Engineering

Best for: AI Scientist, Research Scientist, Director of AI/ML

Related on AIssential

• Large Language Models to Enhance Business Process Modeling: Past, Present, and Future Trends — LLMs are shifting business process modeling from rule-based systems to prompt-engineered, iterative refinement architectures.
• Büyük Dil Modellerinin Hafızasını Güçlendiren Yaklaşım:Retrieval-Augmented Generation (RAG) — RAG enhances LLMs by providing external, real-time context, reducing hallucinations and improving factual accuracy.
• IEEE Rolls Out Large Language Models Virtual Training Course — Large language models are fundamental architectural components demanding deep technical understanding for reliable and secure integration into digital infrastructures.
• A NATO-accredited organization published a report concerning language-based disparities in LLM performance (Published April 25, 2026) — LLM performance varies significantly by language, but adaptation strategies can bridge the gap.
• Retrieval Augmented (Knowledge Graph), and Large Language Model-Driven Design Structure Matrix (DSM) Generation of Cyber-Physical Systems — LLMs, especially with RAG and GraphRAG, can automate Design Structure Matrix generation for complex systems.
• Retrieval-Augmented Generation as a Bridge Between Classical NLP Pipelines and Modern LLM Systems — Retrieval-augmented generation (RAG) addresses the limitation of static knowledge in large language models (LLMs) by connecting them to external, up-to-date knowledge sources.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.AI updates on arXiv.org.