🗓️ 21032026 2100

AGENTIC DESIGN PATTERNS

Architectural patterns for building AI agent systems, ranging from simple single-agent setups to complex multi-agent orchestrations

Overview

Agentic patterns define how AI agents are structured, coordinated, and controlled to accomplish tasks. Choosing the right pattern involves trade-offs between control, flexibility, cost/latency, and complexity.

Pattern Comparison

PatternUse CaseControlFlexibilityCost & LatencyComplexity
Single AgentSimple prototypesFull (one agent)LowLowLow
Sequential / ParallelStructured workflowsModerateModerateModerateModerate
loop_review_critique_patternTasks with strict criteriaIterativeModerateHigh (loops)High
coordinator_router_patternComplex task routingDelegatedHighHighHigh
agent_as_tool_patternFine control with sub-agentsFull (primary)HighHighHigh

Key Takeaways

  • Single / Sequential / Parallel agents suit most straightforward workflows
  • Loop agents shine when outputs must meet non-negotiable constraints (e.g., budget limits, safety checks)
  • Coordinator agents excel at decomposing complex requests into subtasks handled by specialists
  • Agent-as-tool keeps centralized state management while leveraging specialized capabilities
  • More advanced patterns increase latency and cost — only use them when simpler patterns fall short
  • Foundational concepts: llm_tool_use (how agents interact with external functions) and agent_memory_and_state (how agents retain context across steps)

References