Teched2025 BTM161

BTM161 - AI agent mining and governance with SAP Signavio solutions

Description

This repository hosts the (self-contained but simplified) material for the SAP TechEd 2025 session "BTM161 - AI agent mining and governance with SAP Signavio solutions"

Overview

This SAP Sample introduces attendees to analyzing the behavior of LLM-based multi-agent systems with [SAP Signavio Process Intelligence](https://www.signavio.com/products/process-intelligence/). Within three Jupyter notebooks of an agentic coffee shop, specialized agents work together to provide a complete coffee shop experience. Each agent has specific tools and responsibilities, and they intelligently hand off to each other based on the situation to serve the "best" coffee for its customers. The interactions with the coffee shop and the actions taken by the agents are logged and, afterwards, can be analyzed by generating event logs suitable for process mining with SAP Signavio Process Intelligence.

Please be aware that this repository includes a simplified version which generates CSV-files that can be easily consumed by your own SAP Signavio Process Intelligence. This will allow you to try this self-contained version on your own afterwards as well adapt it to your needs but it will not include all features that have been presented at TechEd.

For the TechEd hands-on session, we will host a dedicated Jupyter notebook instance, offer access to SAP Signavio Process Intelligence, and several advanced agent mining features in a dedicated service.

Requirements

This section covers the prerequisites and the installation process required for running the coffee shop MAS.

Prerequisites

• Python >= 3.13
• (Recommended) poetry is used for managing packages and the virtual environment and needs to be installed.
  • Alternative: Use pip to install the dependencies based on the prov