{
  "id": "plan-and-execute",
  "code": "PTL-0141",
  "term": "Plan-and-Execute",
  "aliases": [
    "planner-executor",
    "plan-then-execute",
    "ReWOO",
    "LLMCompiler"
  ],
  "category": "agents",
  "definition": "Plan-and-execute is an agent design in which a planner model first writes a complete multi-step plan and executor calls then carry out the steps, rather than deciding one action at a time after every observation as in ReAct.",
  "description": "Planning up front reduces repeated reasoning over the growing history, cuts tokens and latency, and lets independent steps run in parallel. ReWOO plans with placeholders for tool results, and LLMCompiler schedules tool calls as a dependency graph. The trade-off is adaptability, so designs usually add replanning when a step fails or returns something unexpected.",
  "example": null,
  "broader": [
    "agentic-workflow"
  ],
  "narrower": [],
  "related": [
    "react",
    "plan-and-solve-prompting",
    "orchestrator-workers",
    "parallelization"
  ],
  "introduced": 2023,
  "added": "2026-10-10",
  "sources": [
    {
      "title": "ReWOO: Decoupling Reasoning from Observations for Efficient Augmented Language Models",
      "authors": "Xu et al.",
      "year": 2023,
      "url": "https://arxiv.org/abs/2305.18323"
    },
    {
      "title": "An LLM Compiler for Parallel Function Calling",
      "authors": "Kim et al.",
      "year": 2023,
      "url": "https://arxiv.org/abs/2312.04511"
    }
  ],
  "url": "https://protologue.com/t/plan-and-execute/",
  "citation": "Protologue. (2026). Plan-and-Execute. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.4.0, PTL-0141). https://protologue.com/t/plan-and-execute/"
}