Architecture Selection Guide

Use this decision framework to pick the right architecture for your use case.

Decision Flowchart

flowchart TD
    A[Start: What's your task?] --> B{Single domain?}
    B -->|Yes| C{Needs tools?}
    C -->|No| D[Prompt Chaining]
    C -->|Yes| E{Self-improvement needed?}
    E -->|No| F[Single Agent - ReAct]
    E -->|Yes| G{Quantifiable criteria?}
    G -->|Yes| H[Evaluator-Optimizer]
    G -->|No| I[Reflection]

    B -->|No| J{How many domains?}
    J -->|2-3| K{Central control needed?}
    K -->|Yes| L[Supervisor]
    K -->|No| M{Conversational handoffs?}
    M -->|Yes| N[Swarm]
    M -->|No| O[Routing]

    J -->|4+| P{Predictable subtasks?}
    P -->|Yes| Q[Parallelization]
    P -->|No| R{Many agents?}
    R -->|Yes| S[Hierarchical Teams]
    R -->|No| T[Orchestrator-Worker]

    style A fill:#4CAF50,color:#fff
    style D fill:#2196F3,color:#fff
    style F fill:#2196F3,color:#fff
    style H fill:#2196F3,color:#fff
    style I fill:#2196F3,color:#fff
    style L fill:#2196F3,color:#fff
    style N fill:#2196F3,color:#fff
    style O fill:#2196F3,color:#fff
    style Q fill:#2196F3,color:#fff
    style S fill:#2196F3,color:#fff
    style T fill:#2196F3,color:#fff

Quick Reference Table

Architecture	Complexity	Agents	Best For
ReAct	Low	1	Simple tool-augmented tasks
Prompt Chaining	Low	1	Sequential multi-step pipelines
Routing	Low-Medium	1 + N handlers	Multi-domain classification
Parallelization	Medium	N parallel	Independent concurrent subtasks
Orchestrator-Worker	Medium-High	1 + N dynamic	Complex tasks, unpredictable decomposition
Supervisor	Medium-High	1 + N managed	Coordinated multi-agent with central control
Reflection	Medium	1-2	Self-improving output quality
Evaluator-Optimizer	Medium	2	Scored quality iteration
Swarm	High	N peer	Conversational multi-domain handoffs
Hierarchical Teams	High	N nested	Large-scale enterprise workflows