ReAct
Also: Reasoning + Acting
A pattern for building AI agents that alternates between reasoning and acting — think, then do, then observe the result, then think again. Instead of generating one big response, the agent works iteratively: it reasons about what tool to use, uses it, reads what happened, reasons about the next step, and so on. It's what makes agents feel like thoughtful problem-solvers rather than one-shot responders.
In practice
You give Claude a research task. Instead of one long response, it thinks: "I need to search for X" → searches → reads result → thinks: "I need to also check Y" → searches → synthesizes. That think-act-observe-think loop is ReAct. It's how agents that use tools work reliably — interleaving reasoning with action rather than planning everything upfront.
Related concepts