The ReAct Pattern: Reason, Then Act

A single tool call answers a single-step question. But "is Drug X more effective than Drug Y for condition Z, according to the latest three meta-analyses" needs multiple searches, reading several papers, and synthesizing across them. No single LLM call can do that in one shot. ReAct, short for Reasoning and Acting, is the pattern that turns a single tool-using model into something that can work through a genuinely multi-step problem.

Each iteration of a ReAct agent follows the same three-part cycle. Thought: the model reasons in plain language about what it currently knows and what it still needs. Action: the model calls a tool based on that reasoning. Observation: the tool's real result comes back and becomes part of the context for the next thought. This repeats until the model decides it has enough to answer.

You already saw one full trajectory in Tier 7, scheduling a meeting through two tool-call cycles. Here it is again, worth re-watching now that the pattern has a name: step through it and notice that every action is grounded in the observation right before it, not in the model simply guessing.

Without this loop, a model asked a multi-step question either refuses, or worse, confidently fabricates an answer that sounds like it did the research without actually doing any of it. ReAct forces the model to ground each step in a real tool result before moving to the next thought, which is exactly what makes the difference between "sounds plausible" and "is actually true," the same hallucination concern from Phase 1's bridge chapter, now addressed structurally rather than left to hope.

Without an explicit stopping condition, a ReAct agent can loop forever: thinking, failing, retrying, thinking again, burning tokens and money the entire time. The level 22 case brief is specific about this: cap the loop at a fixed number of iterations, commonly around eight for a research-style task, and give the model a concrete criterion for when it has enough, such as "stop once you have at least three sources supporting your conclusion."