LESSON 7.4 · Theory · 120 min

Planning, reflection, and multi-agent systems

Planning, reflection, and multi-agent systems. Learn decomposition, search, critics, and coordination cost and complete: Run ablations on a verifiable task. Part of the “From Model

Learning objectives

  1. Explain what problem “Planning, reflection, and multi-agent systems” solves without hiding behind terminology.
  2. Trace the variables and causal links across decomposition, search, critics.
  3. Complete “Run ablations on a verifiable task” and judge the result with evidence rather than intuition.

Core concepts

decomposition

decomposition is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment before moving to an optimized implementation.

search

search is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment.

critics

critics is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment.

and coordination cost

and coordination cost is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment.

Build and verify

Run ablations on a verifiable task

  • Predict: write the expected output, trend, or failure before running code.
  • Build: implement only the minimum components needed to answer the question.
  • Verify: compare with a baseline or trusted implementation; save seeds, parameters, and raw outputs.
  • Transfer: change one shape, dataset, scale, or workload condition and explain whether the conclusion still holds.

Open the complete interactive lesson