# Agentic Business Process Management: Practitioner Perspectives on Agent Governance in Business Processes

**Authors:** Hoang Vu, Nataliia Klievtsova, Henrik Leopold, Stefanie Rinderle-Ma, Timotheus Kampik  
**arXiv:** [2504.03693](https://arxiv.org/abs/2504.03693)  
**Date:** 2025-03-23  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2504.03693v2/

## Key points

- Growing interest in software agents driven by generative AI
- Stochastic AI-based agents require robust governance for safe deployment
- Uncertainty among BPM practitioners regarding what qualifies as an agent and associated benefits, risks, and challenges
- Anticipated benefits of AI agents: improved efficiency, data quality, compliance, scalability through automation
- Potential risks of AI agents: bias, over-reliance, cybersecurity threats, job displacement, ambiguous decision-making
- Six key recommendations for responsible adoption of AI agents:
- Defining clear business goals
- Setting legal and ethical guardrails
- Fostering human-agent collaboration
- Customizing agent behavior
- Managing risks effectively
- Ensuring safe integration with fallback options
- Balancing human and agent roles
- Redefining human involvement in processes
- Adapting structures accordingly
- Introducing performance metrics

## Summary

The study "Agentic Business Process Management: Practitioner Perspectives on Agent Governance in Business Processes" explores the growing interest in software agents driven by the emergence of generative AI. These stochastic AI-based agents require robust governance for effective and safe deployment within organizations. The concept of agentic business process management serves as a facilitator for this governance, but there is still uncertainty among BPM practitioners regarding what qualifies as an agent and the associated benefits, risks, and challenges. To shed light on how organizations can govern AI agents effectively, a qualitative study was conducted involving 22 BPM practitioners from various industries. These experts anticipate that AI agents will improve efficiency, data quality, compliance, and scalability through automation. However, they also caution against potential risks such as bias, over-reliance, cybersecurity threats, job displacement, and ambiguous decision-making. In response to these challenges, six key recommendations are proposed for responsible adoption of AI agents: defining clear business goals; setting legal and ethical guardrails; fostering human-agent collaboration; customizing agent behavior; managing risks effectively; and ensuring safe integration with fallback options. Additionally, the paper outlines actions to align traditional BPM practices with agentic AI by balancing human and agent roles; redefining human involvement in processes; adapting structures accordingly; and introducing performance metrics. These insights provide a practical foundation for integrating AI agents into business processes while maintaining oversight mechanisms that preserve flexibility and trust. By proactively addressing concerns surrounding governance issues related to AI agents through these recommendations and actions outlined in the study's findings offer valuable guidance for organizations navigating the evolving landscape of artificial intelligence integration in their operations.