Agentic AI and the Role of Human Supervision in Complex Systems
Orchestrating automation involving software agents, robots, and people can be a complex challenge for many organizations. At the recent UiPath Agentic Automation Summit in Sydney, Feiran Hao, Vice President of Product Strategy at UiPath, highlighted the difficulties companies face in effectively integrating these technologies. Many businesses are trying to connect systems loosely, which can lead to failures in communication and operation. Hao pointed out that while an organization may have a robot monitoring for new emails and a separate process that sends those emails, a direct link between the two processes often doesn’t exist.
Even as companies show interest in advancing their use of agentic tools, Hao observed that they frequently develop these technologies in isolation. This isolation can hinder the ability to troubleshoot when something goes wrong.
Despite the potential AI brings to the table, most businesses are hesitant to allow AI fully autonomous decision-making. For instance, Hao explained that in the loan processing scenario, organizations prefer AI to gather data and provide recommendations rather than make the final decisions. This ensures that human judgment can still consider additional context that AI may not grasp, a necessity in sensitive areas such as healthcare.
UiPath is investing in improving reliability and version control of its tools by implementing features for traceability and evaluation. They suggest creating scenarios to observe how agents respond and provide feedback on adjusting behaviors. This proactive feedback is integrated early in the development process, allowing for refinements before going live. For organizations apprehensive about experimenting, Hao recommends starting with small projects to build confidence while still striving for innovation. Many businesses still have foundational automation initiatives to complete before attempting more advanced practices.
The ongoing integration of AI in various industries promises significant value in operational areas like supply chain management and pricing strategy. Despite the excitement surrounding agentic AI, Hao cautioned that many enterprises have groundwork to cover before they can leap into more advanced automation.
“Content generated using AI”
We create intelligent software and AI-driven solutions to automate workflows, modernize legacy systems, and sharpen your competitive edge.