“Systems thinking helps me put the big picture front and center”

AI researcher Shuai Guo discusses the practical challenges of building deep research agents, noting they often make mistakes by mixing up search results or pre-trained knowledge. He advocates for using guardrails and test-driven development to create production-ready systems. Guo suggests a division of labor where analytical AI handles quantitative tasks like churn prediction, while LLM-based agents tackle qualitative work like personalized communication. He credits his background in systems thinking for helping him design robust AI applications by focusing on components, interfaces, and the overall architecture. While multi-agent systems are advancing rapidly, he warns that the assurance and reliability layers are lagging and could benefit from the rigor applied in industrial systems engineering.