Software AG IUG 2024: Process mining as GenAI democratisation candidate

The Software AG International User Groups conference in Dublin heard about how generative AI large language models could democratise business process mining and management

Brian McKenna,
Senior Analyst, Business Applications

Published: 11 Apr 2024

The Software AG International User Groups conference (IUG24) in Dublin was told this week of the supplier’s distinctive mission for generative artificial intelligence (GenAI) – the democratisation of business process mining and management.

Stefan Sigg, chief product officer at the German enterprise software group, told a gathering of media and analysts ahead of his keynote presentation that organisations are still in the early stages of discovering how GenAI could improve business process mining and modelling, but that it does hold significant promise. And that it needs to be understood better for that to happen.

Sigg is a mathematician by background and spent 22 years at SAP, joining Software AG in 2017.

In the briefing, he said they were announcing “two groundbreaking ways to use generative AI, ie large language models. Number one, we let AI find where the business processes need to be improved. Number two, we let AI create the [business process] models.”

He showed how, using ChatGPT, the supplier’s Aris business process mining and management software can responds to prompts like: “What is my most difficult business process?” and, “Write an email for me to my CEO about how it can be improved”.

He said that these GenAI features profit, like all GenAI, from the wisdom of the crowd: “If you want to do something with AI, you have to have evidence that there is enough world wisdom out there. If you ask the thing something that nobody knows, of course, it will not work. It will not solve you an answer to a new mathematical problem.”

At the conference, which was attended by over 600 people from around from 300 worldwide Software AG customers, the supplier announced updates to its Aris process mining/modelling tool, and its Alfabet enterprise architecture and strategic portfolio management system.

It introduced, within Aris, what it calls, “an AI companion, to allow employees at any level to identify and leverage efficiencies of their operations by assisting them in the mining and modelling business processes. At a time when rapid change and fierce competition are commonplace, this new feature has the capacity to democratise the streamlining of process within organisations”, the company said in a statement.

Sigg added, in that statement: “Businesses ultimately need to be more competitive and differentiated. The barrier is that they often have a sense of what to do, but the complexity of their organisation prevents them from identifying clear areas of improvement. As they attempt to deeper analyse their key business processes, adoption of sophisticated yet accessible, process modelling and process mining tools is necessary. Introducing AI assistance into Aris will accelerate and simplify this initiative. Using natural language prompts to extract specific insights from process data is a really sharp weapon for teams to create sustainable business success.”

In the media and analyst session he went further into what he sees as the value of generative AI for enterprise software, such as business process mining. It is worth quoting him at length because GenAI is not universally understood, or not as well as it might be.

“These large language models are dumb. The intelligence is based on the incredible amount of data that they index. Now, that’s not a new technology. Where this technology comes from is language translation. That’s why they are still called language models.

“In the 1980s or so, people tried to teach a software system what language is. They tried to build software systems that understood language – semantically, syntactically – for the sake of translating, but then there was new approach with the internet. The internet has billions of web pages, probably everything has been translated somewhere. So, rather than teaching software what language is, we just look at what the translation looks like. In a smart way, not the entire text, but token by token. ‘How many times was it translated in this way?’ We take the top of that and then glue it together. And what comes out is the magic. It is incredibly good. What we have here is a humongous network, then there is an algorithm that operates on that network in a very clever way”.

GenAI still going strong

GenAI seems to be holding on as a topic of interest for enterprise IT, and among Software AG-type customers, despite some user weariness. For instance, recent research from TechTarget’s market insights team, based on analysing user activity on our global network of web sites, shows a 183% increase in engagement with generative AI among a business process management-specific audience over the past 12 months. And the 2024 Technology spending intentions survey, from Enterprise Strategy Group and TechTarget Research, showed that 39% of the organisations surveyed worldwide are planning to make significant investments around business process automation over the next 12 months.

Asking business questions

Software AG claims that its AI assistant will allow anyone in an organisation to ask standard and relevant business questions to easily investigate processes. They can ask questions, such as, “Find the anomalies in our purchase-to-pay processes,” or “What are the biggest bottlenecks in our distribution networks”? And they can get comprehensive answers in response, the supplier claims.

That is available now. In Q3 of this year, its process modelling AI assistant, as Sigg demonstrated in the media and analyst session and also in his keynote, promises to enable customers to use natural language prompts to generate process models without manual editing effort.

The supplier also announced a new release of its Alfabet product for enterprise architecture and strategic portfolio management. The release, said the supplier, makes analysis of Alfabet’s “large and interwoven repository of information on the IT landscape far more accessible to non-IT users”.

In that regard, Sigg commented: “We’re in a new era for how business is done. More products and services are digital and nearly everyone in the business has a vested interest in IT. Everyone needs to contribute and collaborate to properly analyse and decide what to do next. This is why we’ve re-designed Alfabet. AI is re-defining what accessibility to information means, so we’re removing the barriers to accessing, analysing and sharing data that’s key to each individual’s role in digital transformation.”

The new features are described like this:

Smart Data Workbench: allows users to fully customise information display, filters and graphics to get the analysis that they need.
Data Quality Rules: allows any user to set the parameters for acceptable data quality, and also shows suggested fixes.
Fast-path Configuration: this is said to apply reasoning-based AI on the Alfabet meta model to reduce product configuration effort.

Takeaway

Software AG’s vision of using GenAI to democratise process mining and management is, admittedly, compelling. I have previously expressed scepticism about the notion of pressing busy business professionals into the use of low-code and no-code tools, for example around the announcement of SAP Build in 2022.

In similar vein, TechTarget industry editor David Essex and Holger Mueller of Constellation Research, had a stimulating discussion in late 2022 about the topic of ERP vendors punting low-code/no-code development that explored the positives and negatives of the trend.

But it is possible that GenAI will, finally, enable the long-hoped for democratisation and greater inclusivity of enterprise software usage among a wider population of business users.

Brian McKenna is a senior analyst at TechTarget’s Enterprise Strategy Group, who focuses on business applications. Previously, he was an editor at ComputerWeekly.

Enterprise Strategy Group is a division of TechTarget. Its analysts have business relationships with vendors.

Read more on Artificial intelligence, automation and robotics

How to craft a generative AI security policy that works