What is ReAct (Reasoning + Acting) prompting?
LLM & AILevel: IntermediateAdvanced Prompting
What is ReAct (Reasoning + Acting) prompting?
Hints
- Identify the stage: pretraining, adaptation, generation or evaluation.
- Name the inputs, constraints and a case where the approach fails.
Answer
ReAct interleaves reasoning traces and action calls:
Thought: I need to find the current AAPL stock price.
Action: search("AAPL stock price today")
Observation: AAPL is trading at $212.50 as of 2:15pm EST.
Thought: Now I have the data I need to answer.
Answer: AAPL is currently trading at $212.50.
Forcing the model to externalize its reasoning before each action:
- Improves tool selection accuracy
- Makes debugging straightforward (full trace is visible)
- Reduces circular loops (model can recognize it's not making progress)
ReAct is the foundation pattern for most production AI agents.
Source and licence: JustInCache & contributors — Awesome AI Interviews · Apache-2.0