# Agentic AI for Intent-Based Industrial Automation

**Authors:** Marcos Lima Romero, Ricardo Suyama  
**arXiv:** [2506.04980](https://arxiv.org/abs/2506.04980)  
**Date:** 2025-06-05  
**Summary page:** https://www.summarizepaper.com/en/paper/2506.04980v1/

## Key points

- Marcos Lima Romero and Ricardo Suyama explore the potential of Agentic AI systems in revolutionizing industrial automation
- Agentic AI systems are powered by autonomous large language models (LLMs) with planning and tool-usage capabilities
- A novel conceptual framework integrates Agentic AI with the intent-based paradigm from network research to align with Industry 5.0 principles
- Intent-based processing allows human operators to communicate high-level goals in natural language, decomposed into actionable components for sub-agents
- Proof of concept using CMAPSS dataset and Google Agent Developer Kit (ADK) validates intent decomposition, agent orchestration, and autonomous decision-making in predictive maintenance scenarios
- Agentic AI coupled with intent-based processing can overcome technical barriers and facilitate scalable automation driven by human intent, prioritizing user-friendly interactions and sustainable practices

## Summary

In their paper titled "Agentic AI for Intent-Based Industrial Automation," Marcos Lima Romero and Ricardo Suyama explore the potential of Agentic AI systems in revolutionizing industrial automation. These systems are powered by autonomous large language models (LLMs) equipped with planning and tool-usage capabilities, offering a new frontier in simplifying human-machine interaction (HMI) and streamlining complex processes introduced by Industry 4.0. The authors propose a novel conceptual framework that integrates Agentic AI with the intent-based paradigm originally developed in network research. This integration aims to align automation systems with the principles of Industry 5.0, focusing on human-centric, sustainable, and resilient practices. By leveraging intent-based processing, the framework enables human operators to communicate high-level business or operational goals in natural language. These intents are then decomposed into actionable components such as expectations, conditions, targets, context, and information that guide sub-agents equipped with specialized tools to efficiently execute domain-specific tasks. To validate their approach, a proof of concept was implemented using the CMAPSS dataset and Google Agent Developer Kit (ADK). The results showcased the feasibility of intent decomposition, agent orchestration, and autonomous decision-making in predictive maintenance scenarios. This study highlights the potential of Agentic AI coupled with intent-based processing to overcome technical barriers and facilitate scalable automation driven by human intent. Despite challenges related to data quality and explainability concerns, this innovative framework demonstrates promise in enhancing industrial automation processes while prioritizing user-friendly interactions and sustainable practices. The authors' work sets a foundation for future advancements in intelligent automation systems that prioritize alignment with human goals and values within industrial settings.