What is ReAct prompting and why is it the foundation of most agents?
LLM & AILevel: IntermediateAgent Architectures
What is ReAct prompting and why is it the foundation of most agents?
Hints
- Identify the stage: pretraining, adaptation, generation or evaluation.
- Name the inputs, constraints and a case where the approach fails.
Answer
ReAct (Yao et al., 2022) interleaves reasoning and acting in a single prompt:
Thought: I need to find the current weather in Tokyo.
Action: weather_api({"city": "Tokyo"})
Observation: {"temp": 22, "condition": "Sunny", "humidity": 45}
Thought: I have the weather data. I can now answer the user.
Answer: Tokyo is currently 22°C and sunny.
Why it works as a foundation:
- Explicit reasoning trace = easier debugging
- Model articulates intent before acting = better tool selection
- Observations are visible in trace = model can recognize failure
- Standardized format works across LLMs and tools
Variants: ReAct + self-reflection (Reflexion), ReAct + planning (Plan-and-Execute), ReAct + RAG (Agentic RAG).
Source and licence: JustInCache & contributors — Awesome AI Interviews · Apache-2.0