Decomposition

7 frameworks — Break complex problems into smaller, manageable parts with tree, graph, and recursive approaches.

Overview

Decomposition methods break complex problems into smaller, manageable parts using tree, graph, and recursive approaches. Instead of asking AI to solve everything at once, these techniques structure the problem space into interconnected sub-problems.

Use decomposition when facing problems too complex for a single prompt — multi-part analyses, planning tasks, or problems with interconnected dependencies. These techniques are especially valuable for programming, mathematical proofs, and strategic planning.

Technique Comparison

Side-by-side comparison of all 8 frameworks in this category.

Technique Year Best For Key Strength Complexity
Least-to-Most 2022 Progressive problems Incremental difficulty Medium
DECOMP 2022 Multi-part questions Sub-question delegation Medium
Plan-and-Solve 2023 Sequential tasks Plan-first approach Medium
Tree of Thought 2023 Exploration tasks Branch evaluation High
Graph of Thought 2023 Complex dependencies Merge operations High
Recursion of Thought 2023 Nested problems Recursive solving High
Program of Thought 2022 Math & logic Executable code High
Skeleton-of-Thought 2023 Latency reduction Parallel expansion Medium
Recommended Starting Point

Try Least-to-Most Prompting

Least-to-Most is the most accessible decomposition technique — break any complex problem into a series of progressively harder sub-problems, solving each one before tackling the next.