ProMoAI is an AI-powered suite for Process Mining that leverages Large Language Models (LLMs) to bridge the gap between natural language and technical process analysis. The framework now includes PMAx, an autonomous agentic system for data-driven insights.

The suite consists of two primary modules:

1. ProMoAI (Model Generation): Transforms text or event logs into formal process models (BPMN/Petri nets).
2. PMAx (Agentic Analytics): An autonomous multi-agent framework that functions as a virtual process analyst to query event logs and generate data-grounded reports.

• Text-to-Model: Generate BPMN or PNML models from natural language descriptions.
• Model Refinement: Upload an existing BPMN or Petri net and use AI to modify or extend it via chat.
• Discovery Baseline: Start with an XES event log to discover an initial model, then refine it using the LLM.

• Autonomous Reasoning: Uses a "divide-and-conquer" architecture with specialized Engineer and Analyst agents.
• Privacy-Preserving: Only lightweight metadata (column names/types) is sent to the LLM. Raw event data never leaves your local environment.
• Deterministic Accuracy: The system generates and executes local Python code (using whitelisted data preprocessing libraries) to compute exact metrics, avoiding LLM hallucinations.
• Comprehensive Reporting: Automatically generates tables, statistical charts, and narrative insights from high-level business questions.

Access the unified suite directly at: https://promoai.streamlit.app/

1. Clone this repository.
2. Install the required environment and packages (see Requirements).
3. Run the application:

• Unified Suite (ProMoAI + PMAx):

  streamlit run app.py

• Standalone ProMoAI (Legacy Interface):

  streamlit run promoai_standalone.py

You can install the core ProMoAI components via pip:

pip install promoai

• Environment: the app is tested on both Python 3.9 and 3.10.
• Dependencies: all required dependencies are listed in the file 'requirements.txt'.
• Packages: all required packages are listed in the file 'packages.txt'.

If you use this suite in your research, please cite the relevant papers:

@inproceedings{DBLP:conf/ijcai/KouraniB0A24,
  author       = {Humam Kourani and
                  Alessandro Berti and
                  Daniel Schuster and
                  Wil M. P. van der Aalst},
  title        = {ProMoAI: Process Modeling with Generative {AI}},
  booktitle    = {Proceedings of the Thirty-Third International Joint Conference on
                  Artificial Intelligence, {IJCAI} 2024},
  pages        = {8708--8712},
  publisher    = {ijcai.org},
  year         = {2024},
  url          = {https://www.ijcai.org/proceedings/2024/1014}
}