ReAct: Reasoning & Acting

Pre-Flight Briefing

The Birth of AI Agents

ReAct (Reasoning and Acting) is the paradigm that transforms LLMs from passive text-generators into active Agents.

Instead of just 'thinking', the model is given access to tools (like a search engine or database). It operates in a loop: Thought -> Action -> Observation.

1. Thought: 'I need to know the current weather in Paris.'

2. Action: 'Search[Weather in Paris]'

3. Observation: (The system returns the search results).

4. Thought: 'Now I have the weather, I can formulate the final answer.'

Reference Examples

The ReAct Loop

Thought: I need to find the CEO of Apple.
Action: Search[CEO of Apple]
Observation: Tim Cook is the CEO.
Thought: I have the answer.
Final Answer: Tim Cook.