This research introduces Flow-of-Action, a novel multi-agent system that dramatically improves automated root cause analysis in complex microservices architectures.

• Incorporates standard operating procedures (SOPs) into LLM-based agents to guide systematic investigation
• Improves root cause identification accuracy by up to 50% compared to standard ReAct framework
• Reduces problem resolution time from hours to minutes in real-world incident scenarios
• Introduces a hybrid approach combining LLM reasoning with structured operational knowledge

For engineering teams, this represents a significant advancement in automated incident management, enabling faster resolution times and reducing dependency on specialized domain experts during critical outages.

Flow-of-Action: SOP Enhanced LLM-Based Multi-Agent System for Root Cause Analysis