Architectures Overview

This section covers 70 agentic design patterns organised into 14 parts. Use the Selection Guide to find the right pattern for your use case.

---

Part I — Core Patterns (1–10)

The foundational building blocks. Master these first — every subsequent pattern builds on one or more of these.

#	Architecture	Complexity	Agents	Best For
1	Single Agent (ReAct)	Low	1	Simple tool-augmented tasks
2	Prompt Chaining	Low	1	Sequential multi-step pipelines
3	Routing	Low-Med	1+N	Multi-domain classification
4	Parallelization	Medium	N	Independent concurrent subtasks
5	Orchestrator-Worker	Med-High	1+N	Unpredictable task decomposition
6	Supervisor	Med-High	1+N	Coordinated multi-agent control
7	Reflection / Self-Correction	Medium	1-2	Self-improving output quality
8	Evaluator-Optimizer	Medium	2	Scored quality iteration
9	Swarm	High	N	Conversational multi-domain handoffs
10	Hierarchical Teams	High	N	Large-scale enterprise workflows

---

Part II — Quality, Verification & Oversight (11–13)

#	Architecture	Complexity	Agents	Best For
11	Generator-Verifier	Medium	2	Execution-based verification (code, math)
12	Agent-as-a-Judge	Medium	2	One-shot quality evaluation & ranking
13	Human-in-the-Loop	Medium	1+	High-stakes approval workflows

---

Part III — Parallelism, Pipelines & Cost Optimization (14–18)

#	Architecture	Complexity	Agents	Best For
14	Map-Reduce Agents	Medium	N	Processing large documents/datasets
15	Mixture-of-Agents	High	N models	Maximum quality via model diversity
16	DAG / Graph Orchestration	Medium	N	Complex dependency workflows
17	Plan-and-Execute	Medium	1	Strategic multi-step tasks with replanning
18	Cascading Agents	Medium	2-3	Cost-optimized inference

---

Part IV — Multi-Agent Collaboration (19–26)

#	Architecture	Complexity	Agents	Best For
19	Blackboard Pattern	High	N	Incremental collaborative problem-solving
20	Market-Based / Bidding	Medium	N	Dynamic self-organizing task allocation
21	Contract Net Protocol	Medium	N	Formal task delegation with accountability
22	Debate / Adversarial	Medium	2-3	Rigorous decision analysis
23	Red-Team Agent	Medium	2	Security & safety testing
24	Role-Based Collaboration	Medium	N	Persona-driven team pipelines
25	Conversational Multi-Agent	Medium	N	Emergent group problem-solving
26	Event-Driven Multi-Agent	High	N	Reactive async multi-agent systems

---

Part V — Reasoning Patterns (27–30)

#	Architecture	Complexity	Agents	Best For
27	Inner Monologue	Low	1	Clean output with hidden reasoning
28	Speculative Execution	Medium	1	Low-latency branching decisions
29	Skeleton of Thought	Medium	1	Fast long-form generation
30	ReAcTree	High	1	Multi-strategy tree exploration

---

Part VI — Domain Applications (31–34)

#	Architecture	Complexity	Agents	Best For
31	Agentic RAG	Medium	1	Intelligent document retrieval
32	Agentic Coding	Medium	1	Self-healing code generation
33	Self-Tooling Agent	Medium	1	Runtime tool creation
34	Dynamic Tool Generation	Medium	1	Systematic tool generation pipeline

---

Part VII — Cognitive Architectures (35–36)

#	Architecture	Complexity	Agents	Best For
35	Neuro-Symbolic Agent	High	1	LLM + formal logic/rules
36	Dual-Paradigm Framework	Medium	1	System 1 / System 2 routing

---

Part VIII — Advanced Planning & Search (37–41)

#	Architecture	Complexity	Agents	Best For
37	LATS	High	1	LLM + Monte Carlo Tree Search
38	Introspective MCTS	High	1	Per-simulation introspective search
39	Reflective MCTS	High	1	Cross-episode reflective search
40	Collaborative Tree Search	High	N	Multi-agent collaborative tree search
41	Beam Search for Agents	Medium	1	Top-K solution exploration

---

Part IX — Agent Protocols (42–45)

#	Architecture	Complexity	Best For
42	Model Context Protocol (MCP)	Low	Tool integration standard
43	Agent-to-Agent Protocol (A2A)	Medium	Cross-vendor agent communication
44	Agent Communication Protocol (ACP)	Medium	Framework-agnostic agent messaging
45	Agent Network Protocol (ANP)	High	Internet-scale agent networking

---

Part X — Agent Memory System (46–52)

#	Architecture	Complexity	Best For
46	Episodic Memory	Medium	Learning from past experiences
47	Semantic Memory	Medium	Storing facts and relationships
48	Procedural Memory	Medium	Remembering how to do tasks
49	Agentic Memory (A-MEM)	High	Interconnected knowledge notes
50	Collaborative Memory	Medium	Shared team knowledge
51	Contextual Experience Replay	Medium	Prioritized experience replay
52	Meta-Memory	High	Memory system optimization

---

Part XI — Agent Safety & Resilience (53–58)

#	Architecture	Complexity	Best For
53	Guardrail Agent	Low	Content safety filtering
54	Circuit Breaker	Low	Preventing cascading failures
55	Saga Pattern	Medium	Multi-step rollback
56	API Gateway / Gatekeeper	Low	Centralized access control
57	Least-Privilege Ephemeral Identity	Medium	Per-task scoped credentials
58	Dead Letter / Escalation	Low	Handling failed tasks

---

Part XII — Agent Infrastructure (59–62)

#	Architecture	Complexity	Best For
59	Agentic Mesh	High	Enterprise multi-agent infrastructure
60	Agent Registry and Discovery	Medium	Large-scale agent management
61	Agent Control Plane	High	Agent governance and lifecycle
62	Data Flywheel	Medium	Continuous improvement from usage

---

Part XIII — Embodied / Physical (63–65)

#	Architecture	Complexity	Best For
63	VLA Models	Very High	Vision + language + robot actions
64	Embodied AI	Very High	Physical world agents
65	World Models	Very High	Internal environment simulation

---

Part XIV — Frameworks & Meta-Approaches (66–70)

#	Architecture	Complexity	Best For
66	CoALA	—	Agent architecture taxonomy
67	AEGIS Framework	High	Comprehensive layered safety
68	ADAS	High	Automated agent design search
69	Self-Evolving Agents (MASE)	High	Self-improving agents
70	Nested Learning	High	Multi-timescale optimization