{
  "schemaVersion": 3,
  "id": "article:multi-agent-systems",
  "slug": "multi-agent-systems",
  "title": "Multi-Agent Systems: When More Than One AI Worker Is Involved",
  "canonicalPath": "/articles/multi-agent-systems/",
  "sourcePath": "content/articles/2026/multi-agent-systems/article.md",
  "agentBriefPath": "content/articles/2026/multi-agent-systems/agent.md",
  "thesis": "Multi-agent systems split complex work across specialized AI agents, but the engineering value comes from coordination, communication, evaluation, and cost tradeoffs, not from simply adding more agents.",
  "status": "published",
  "maturity": "seed",
  "publishedAt": "2026-06-29",
  "updatedAt": "2026-06-29",
  "audiences": [
    "general",
    "students",
    "builders"
  ],
  "topics": [
    "ai-agents",
    "ai-literacy"
  ],
  "series": {
    "slug": "ai-demystified",
    "title": "AI, De-Mystified",
    "order": 12,
    "role": "chapter"
  },
  "claims": [
    {
      "id": "claim-001",
      "claim": "A multi-agent system solves a problem by dividing it among specialized agents and coordinating their work, not by simply running several models in parallel.",
      "confidence": "high",
      "status": "core",
      "evidence": [
        {
          "sourceId": "source-wooldridge-mas",
          "snippet": "A multi-agent system is one composed of multiple interacting, autonomous agents, where the key problem is coordinating those agents so that they can work together.",
          "supports": "background",
          "assessedAt": "2026-06-29"
        }
      ],
      "counterevidence": [
        {
          "summary": "Some implementations call themselves multi-agent while mostly running the same prompt across several outputs and picking the best one, which is closer to ensembling than true division of labor.",
          "assessedAt": "2026-06-29"
        }
      ]
    },
    {
      "id": "claim-002",
      "claim": "Multi-agent systems borrow ideas from distributed systems, ensemble methods, and organizational design, not just recent AI research.",
      "confidence": "high",
      "status": "landscape",
      "evidence": [
        {
          "sourceId": "source-wooldridge-mas",
          "snippet": "Research in multi-agent systems draws on ideas from economics, game theory, and distributed computing, treating coordination as a central problem.",
          "supports": "background",
          "assessedAt": "2026-06-29"
        }
      ],
      "counterevidence": [
        {
          "summary": "Classical multi-agent work often assumes formally specified agents and environments, whereas LLM-based agents are defined informally by prompts, making direct transfer of older theory difficult.",
          "assessedAt": "2026-06-29"
        }
      ]
    },
    {
      "id": "claim-003",
      "claim": "Common multi-agent patterns include sequential pipelines, manager-and-workers, and debate-and-review, and each pattern carries different coordination risks.",
      "confidence": "medium-high",
      "status": "design",
      "evidence": [
        {
          "sourceId": "source-autogen",
          "snippet": "Conversational programming with multiple agents enables flexible patterns such as sequential chats, group chats, and nested chats for task decomposition.",
          "supports": "direct",
          "assessedAt": "2026-06-29"
        },
        {
          "sourceId": "source-multi-agent-debate",
          "snippet": "Multiple instances of a language model debate their answers over several rounds, improving reasoning and factual accuracy.",
          "supports": "direct",
          "assessedAt": "2026-06-29"
        }
      ],
      "counterevidence": [
        {
          "summary": "These patterns are not standardized; different frameworks implement them with different interfaces, error-handling rules, and observability guarantees.",
          "assessedAt": "2026-06-29"
        }
      ]
    },
    {
      "id": "claim-004",
      "claim": "Adding agents increases coordination cost, ambiguity, and failure modes; a multi-agent design should be justified by a specific division of labor, not by default.",
      "confidence": "medium",
      "status": "risk",
      "evidence": [
        {
          "sourceId": "source-multi-agent-debate",
          "snippet": "Debate improves accuracy on some reasoning tasks but increases the number of model calls, latency, and the complexity of aggregating conflicting outputs.",
          "supports": "indirect",
          "assessedAt": "2026-06-29"
        }
      ],
      "counterevidence": [
        {
          "summary": "For some simple tasks, a carefully written single-agent prompt can match or beat a multi-agent setup at lower cost and latency.",
          "assessedAt": "2026-06-29"
        }
      ]
    }
  ],
  "sources": [
    {
      "id": "source-wooldridge-mas",
      "title": "Wooldridge: An Introduction to MultiAgent Systems",
      "url": "https://www.cs.ox.ac.uk/people/michael.wooldridge/pubs/imas/IMAS2e.pdf",
      "type": "book",
      "accessed": "2026-06-29"
    },
    {
      "id": "source-multi-agent-debate",
      "title": "Du, Li, Tarlow, et al.: Improving Factuality and Reasoning in Language Models through Multiagent Debate",
      "url": "https://arxiv.org/abs/2305.14325",
      "type": "paper",
      "accessed": "2026-06-29"
    },
    {
      "id": "source-react-paper",
      "title": "ReAct: Synergizing Reasoning and Acting in Language Models",
      "url": "https://arxiv.org/abs/2210.03629",
      "type": "paper",
      "accessed": "2026-06-29"
    },
    {
      "id": "source-autogen",
      "title": "Wu, Bansal, Jin, et al.: AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation",
      "url": "https://arxiv.org/abs/2308.08155",
      "type": "paper",
      "accessed": "2026-06-29"
    }
  ],
  "related": [
    {
      "type": "article",
      "id": "article:loops-vs-goals"
    },
    {
      "type": "topic",
      "id": "topic:ai-agents"
    }
  ],
  "agentInstructions": [
    "Use claim IDs as the retrieval unit.",
    "Treat maturity=seed as an explicit uncertainty marker.",
    "Do not present multi-agent systems as automatically better than single-agent systems.",
    "When summarizing, preserve the plain-language-first, technical-depth-later structure."
  ],
  "provenance": {
    "createdAt": "2026-06-29",
    "createdBy": "human",
    "agents": [
      {
        "role": "drafting",
        "model": "kimi",
        "invokedAt": "2026-06-29",
        "inputHash": "sha256:0000000000000000000000000000000000000000000000000000000000000000",
        "outputHash": "sha256:a98fc69a4f76de44aec09b4053d177a0e5f61c7d61707cac1564e8055aa860d0"
      },
      {
        "role": "review",
        "model": "kimi",
        "invokedAt": "2026-06-29",
        "inputHash": "sha256:0000000000000000000000000000000000000000000000000000000000000000",
        "outputHash": "sha256:a98fc69a4f76de44aec09b4053d177a0e5f61c7d61707cac1564e8055aa860d0"
      }
    ],
    "reviews": [
      {
        "reviewer": "agent",
        "reviewedAt": "2026-06-29",
        "status": "approved",
        "scope": [
          "claims",
          "tone",
          "privacy",
          "scope"
        ],
        "notes": "Sibling-agent review against article-proposal-ideation eval-card. Privacy scan passed. No proprietary or personal content detected.",
        "contentHash": "a98fc69a4f76de44aec09b4053d177a0e5f61c7d61707cac1564e8055aa860d0"
      },
      {
        "reviewer": "human",
        "reviewedAt": "2026-06-29",
        "status": "approved",
        "scope": [
          "thesis",
          "examples",
          "tone",
          "safety"
        ],
        "notes": "Human author approved the draft for publication.",
        "contentHash": "a98fc69a4f76de44aec09b4053d177a0e5f61c7d61707cac1564e8055aa860d0"
      }
    ],
    "policy": {
      "id": "policy:default",
      "version": "1.0.0"
    }
  },
  "contentHash": "a98fc69a4f76de44aec09b4053d177a0e5f61c7d61707cac1564e8055aa860d0",
  "generatedAt": "2026-06-29T00:00:00.000Z",
  "articleUrl": "https://aura-knowledge.github.io/articles/multi-agent-systems/",
  "agentJsonPath": "/agents/articles/multi-agent-systems.json",
  "agentMarkdownPath": "/agents/articles/multi-agent-systems.md",
  "sourceRepoPath": "content/articles/2026/multi-agent-systems/article.md",
  "sourceGitHubUrl": "https://github.com/aura-knowledge/aura-knowledge.github.io/blob/main/content/articles/2026/multi-agent-systems/article.md",
  "tokenEstimate": 489,
  "sectionOutline": [
    {
      "id": "plain-english-meaning",
      "title": "Plain English Meaning"
    },
    {
      "id": "existing-concept-it-resembles",
      "title": "Existing Concept It Resembles"
    },
    {
      "id": "what-is-actually-new",
      "title": "What Is Actually New?"
    },
    {
      "id": "how-it-works-in-practice",
      "title": "How It Works In Practice"
    },
    {
      "id": "where-it-helps",
      "title": "Where It Helps"
    },
    {
      "id": "where-it-fails",
      "title": "Where It Fails"
    },
    {
      "id": "academic-connections",
      "title": "Academic Connections"
    },
    {
      "id": "practical-checklist",
      "title": "Practical Checklist"
    },
    {
      "id": "the-de-hype-check",
      "title": "The De-Hype Check"
    },
    {
      "id": "open-questions",
      "title": "Open Questions"
    }
  ]
}