AI for Business Process Automation

2 December 2019 in Technology



AI for Business Process Automation

Business Process Automation – global trends

Business process management projects generally aim to reduce costs and improve efficiency, though should also factor in the right kind of innovation to bring a strategic competitive edge. The business process automation (BPA) market is predicted to be worth $12.7 million in 2021, more than twice that of 2016. Central to this is AI, which can improve operations and strategic BPM, with an extra difference.

AI could either be used through the kind of robot process automation (RPA) that notices human activity then automates and improves on it, machine learning that triggers new processes or reroutes current ones, or through machine learning that can make recommendations for new actions.

Data mining techniques cannot be used effectively for processes because behavior such as concurrency, loops, and decisions are too complex.

Wil van der Aalst – Eindhoven University of Technology

AI applied to process automation

Process-based applications that are very advanced can cause delays, though it may be possible to use process-based data to apply the continuous development of process-based apps. This could involve enabling predictions and recommendations for making improvements on a continual basis.

If an existing model can be reviewed against process event logs, then the differences that are found can be used by the model to make predictions, such as for predicting the duration of a process. Live data can be quickly interpreted, and the same algorithm can be used for prediction and recommendation, so process data can be useful for process AI. Suitable process mining algorithms can be found for any process model, such as timestamps for events or time remaining for deadlines.

An example of AI used for a process-based app could be when an organization cannot meet the terms of a time-sensitive service level agreement (SLA). AI enables the app to analyze historical patterns and highlight future limitations. If the manual steps are time-consuming then user interface data can be used to improve predictions and recommendations. The AI can also recommend corrective actions for the current app, as well as making design suggestions for the future app.

With AI applied to process applications, future constraints can be anticipated and aligned with available resources, which can prevent delays. Corrective actions can be guided by process flow pattern detection and business metrics predictions, and current apps can be updated accordingly.

Decision automation can also be based on process mining algorithms, as well as machine learning techniques, and some behaviors may be predicted from studying business process execution, so AI can be enabled to make recommendations on both human and automated decision making. Though similar to recent advancements in science and technology that have enabled cognitive technologies (CT) to be used in BPM, this can improve processes without causing unemployment.

AI is a game-changer

In these ways, AI can be effectively used to improve process execution and user interactions, while also acting as a support to human intelligence. Current process-based apps for BPM are constructive in everyday processes but can be taken further with the help of AI. In the future, BPM apps will use AI techniques for enhanced user experience, operations, and strategy.