Flashcard: when would you pick Plan-and-Execute over ReAct?

Attempt it

Plan-and-Execute has the LLM write a complete plan (an ordered list of steps) once at the start, then a separate executor runs the steps in sequence without re-planning at every step. ReAct interleaves a Thought and an Action at every step, so each new reasoning step sees the latest Observation. The structural difference is when the planning happens: up front in Plan-and-Execute, between every step in ReAct. Plan-and-Execute is faster, cheaper, and lower latency on long, predictable tasks because the expensive planner-model call happens once instead of N times. ReAct adapts better when intermediate results change what should happen next, because the next step is chosen with the latest evidence in hand. Hybrid patterns (LangGraph's plan-and-execute with a replanning node) combine both: plan once, execute several steps, and only call the planner again when an observation contradicts the plan.

Your answer

Quick reference

TL;DR

Plan-and-Execute writes the full plan once and executes it in sequence; ReAct interleaves a Thought and Action at every step. The difference is when planning happens.

Memory aid

Sign in for the mnemonic that makes this stick. See Pro pricing.

Easy to grasp

Imagine planning a road trip. One way is to map out every stop before you leave, then just drive between them without checking the map again. That is Plan-and-Execute. Another way is to drive a bit, look around, decide where to go next, drive a bit more. That is ReAct. The first is faster when the route is predictable because you only think hard once. The second adapts better when something unexpected shows up, like a closed road, because you re-decide with the new information. Real agents often blend the two: plan first, execute several steps, and only re-plan when the world surprises you. Both patterns turn a single model call into an agent loop, just with different rhythms.

Key concepts

Plan-and-Execute ReAct Pattern Replanning Agent Loop

Detailed answer & concept explanation

~7 min read

Everything you need to truly understand this topic: intuition, mechanics, step by step explanation, code, formulas, and worked example. Click to expand.

Everything you need to truly understand this topic: intuition, mechanics, step by step explanation, code, formulas, and worked example.

Sign in for 5 free unlocks/day or go Pro for unlimited. See Pro pricing.

Quick scan · 2–4 min

Everything important, quickly.

Define both patterns by when the planner runs: once up front in Plan-and-Execute, every step in ReAct. State the cost consequence: one planner call versus N. Cover where each wins. Close with the hybrid replanning pattern that most production frameworks ship by default.

Real-world examples

Real products, models, and research that use this idea.

LangGraph's prebuilt plan-and-execute agent runs a planner node once to produce a step list, then loops an executor node with a conditional replanning edge that fires on plan-contradicting observations.
CrewAI's hierarchical-process mode uses a manager agent that plans the task into subtasks and delegates each to a worker agent, with the manager only re-engaging when a worker reports failure.
OpenAI's deep-research mode plans a research outline up front, then executes search and synthesis steps against the plan, replanning only when a critical source is unavailable.
Anthropic's claude-code agent often plans the steps of a refactor up front, runs the edits one by one, and only re-plans when a test failure invalidates a downstream step.

Interviewer follow-ups

What an interviewer would ask next. Try answering before peeking at the approach.

QWhy is Plan-and-Execute often cheaper than ReAct on a 12-step research task?

Sign in or go Pro to see the approach for this follow-up.

QWhat goes wrong if you never re-plan in Plan-and-Execute?

Sign in or go Pro to see the approach for this follow-up.

QHow does LangGraph's hybrid plan-and-execute pattern get the best of both?

Sign in or go Pro to see the approach for this follow-up.

Don't say this

Red flags and common mistakes that signal junior thinking. Click to expand.

Most common mistake

Treating Plan and Execute and ReAct as the same loop with different names. They differ in when the planning step runs: once up front versus at every turn.

Sign in or go Pro to see all red flags and common mistakes. See Pro pricing.

60-second cheat sheet

The night-before-the-interview bullets. Scan these on the way to the call.

Sign in for 5 unlocks/day or go Pro for unlimited revision sheets. See Pro pricing.

References

Primary sources. Skim if you want the original framing.