Knowledge Share
Technical articles, tutorials, and insights
Agent Series (2): ReAct — The Most Important Agent Reasoning Paradigm
Deep dive into ReAct (Reasoning + Acting) — how the Thought → Action → Observation loop gives Agents genuine reasoning ability grounded in real-world feedback. Complete runnable demo code with actual execution traces and production safety patterns.
Agent Series (3): Plan-and-Solve — Think First, Then Act
Deep dive into the Plan-and-Solve paradigm — when ReAct's greedy strategy hits a ceiling on complex tasks, how an explicit planning layer breaks through. Full LangGraph implementation with Plan → Execute → Replan → Finalize architecture and 5 key findings from a real demo run.
Agent Series (1): What Is an Agent — It's Not Just an LLM That Can Call Tools
The first article in the Agent series. Clarify the essential differences between LLM, Chatbot, and Agent. Understand the four elements of an Agent and the two paradigms (Workflow-Driven vs AI Native) — so you know when to use an Agent and when a plain LLM call is enough.