Software AG is enhancing its Aris process mining and Alfabet portfolio management products with generative artificial intelligence capabilities in announcements at its International User Groups conference in Dublin this week.

Both are based on the off-the-shelf Azure OpenAI service from Microsoft Corp. ARIS is used to represent processes within an organization and to model improvements. Alfabet has similar capabilities but is aimed at information technology portfolio management. It helps organizations understand their assets, identify gaps and redundancies and prepare for technology deployments and migrations.

The ARIS AI Companion (pictured) process mining feature, now available, uses generative AI to depict existing processes and identify anomalies graphically. A forthcoming capability planned for the third quarter creates process models based on best practices and recommends improvements in existing processes. The addition of natural language features is intended to broaden both platforms’ appeal and enable more people to participate in efficiency improvements.

Marc Vietor, general manager of Aris and Alfabet, said organizations are often surprised to learn how many exceptions exist to their documented processes. “There are often hundreds of exceptions caused by people working around the rules,” he said.

ARIS AI Companion analyzes existing data in workflows and enterprise resource planning models to derive processes and variations, presenting them in the order of frequency with which they occur. Users can ask questions like “find the anomalies in our purchase-to-pay processes” or “identify the biggest bottlenecks in our distribution networks” and get natural language responses. The forthcoming process modeling feature will present suggested improvements that bring processes in line with best practices in that problem domain.

Modeling the IT landscape

Enhancements in Alfabet 11 provide similar process mining functionality for an organization’s IT landscape. The Smart Data Workbench allows users to customize information display, filters and graphics and share models for collaboration across teams and departments. Users can set the parameters for acceptable data quality and get suggested fixes.

Fast-path Configuration applies reasoning-based AI on the Alfabet meta model to reduce product configuration complexity. It automatically chooses the proper user interface visualization element to ensure consistency across user content and data workbenches.

“We construct the process from the data and then ask AI where there are anomalies and what recommendations for improvement are,” said Stefan Sigg, Software AG’s chief product officer. The answer is in a business process management form anyone can understand.”

Generative AI also helps address what Sigg called the “blank sheet of paper” problem in which users need to document a process but don’t know where to start.

“You can ask the AI model to show the most important processes or generate a process model for a software company’s continuous integration/continuous development process,” he said. “It searches the 30 trillion nodes in its model and can give you maybe an 80% jumpstart on what you think the model should be. You can then iterate on it.”

Software AG chose not to customize the data model for the generative AI features because the ChatGPT model that underlies the Microsoft service already has a wealth of information about best practices, Sigg said. “We are not training the model. We are assuming the knowledge is there and it just needs to be extracted,” he said. “The quality comes through quantity.”

Image: Software AG