[
  {
    "id": "uitars-2025",
    "type": "article-journal",
    "title": "UI-TARS: Pioneering Automated GUI Interaction with Native Agents",
    "author": [
      { "family": "Qin", "given": "Yujia" },
      { "literal": "ByteDance Seed" }
    ],
    "issued": { "date-parts": [[2025, 1]] },
    "URL": "https://arxiv.org/abs/2501.12326",
    "note": "Sections 1, 2 — 5계층 아키텍처와 4축 혁신의 원전"
  },
  {
    "id": "uitars2-2025",
    "type": "article-journal",
    "title": "UI-TARS-2 Technical Report: Advancing GUI Agent with Multi-Turn Reinforcement Learning",
    "author": [
      { "literal": "ByteDance Seed" }
    ],
    "issued": { "date-parts": [[2025, 9]] },
    "URL": "https://arxiv.org/abs/2509.02544",
    "note": "Exec Summary, Section 1 — 'data flywheel' 공식 어휘의 출처"
  },
  {
    "id": "cogagent-2023",
    "type": "paper-conference",
    "title": "CogAgent: A Visual Language Model for GUI Agents",
    "author": [
      { "family": "Hong", "given": "Wenyi" },
      { "literal": "THUDM" }
    ],
    "issued": { "date-parts": [[2023, 12]] },
    "URL": "https://arxiv.org/abs/2312.08914",
    "container-title": "CVPR 2024",
    "note": "Section 2 — 1120×1120 듀얼 인코더 패턴의 시조"
  },
  {
    "id": "seeclick-2024",
    "type": "paper-conference",
    "title": "SeeClick: Harnessing GUI Grounding for Advanced Visual GUI Agents",
    "author": [
      { "family": "Cheng", "given": "Kanzhi" }
    ],
    "issued": { "date-parts": [[2024, 1]] },
    "URL": "https://arxiv.org/abs/2401.10935",
    "container-title": "ACL 2024",
    "note": "Section 2 — GUI 그라운딩 표준화"
  },
  {
    "id": "ferretui-2024",
    "type": "paper-conference",
    "title": "Ferret-UI: Grounded Mobile UI Understanding with Multimodal LLMs",
    "author": [
      { "family": "You", "given": "Keen" },
      { "literal": "Apple" }
    ],
    "issued": { "date-parts": [[2024, 4]] },
    "URL": "https://arxiv.org/abs/2404.05719",
    "container-title": "ECCV 2024",
    "note": "Section 2 — Apple의 모바일 UI 이해 모델"
  },
  {
    "id": "osatlas-2024",
    "type": "article-journal",
    "title": "OS-ATLAS: A Foundation Action Model for Generalist GUI Agents",
    "author": [
      { "family": "Wu", "given": "Zhiyong" }
    ],
    "issued": { "date-parts": [[2024, 10]] },
    "URL": "https://arxiv.org/abs/2410.23218",
    "note": "Section 2 — Foundation Action Model 카테고리"
  },
  {
    "id": "showui-2024",
    "type": "paper-conference",
    "title": "ShowUI: One Vision-Language-Action Model for GUI Visual Agent",
    "author": [
      { "family": "Lin", "given": "Kevin Qinghong" }
    ],
    "issued": { "date-parts": [[2024, 11]] },
    "URL": "https://arxiv.org/abs/2411.17465",
    "container-title": "CVPR 2025",
    "note": "Section 2 — VLA를 GUI에 적용한 직전 단계"
  },
  {
    "id": "osworld-2024",
    "type": "paper-conference",
    "title": "OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments",
    "author": [
      { "family": "Xie", "given": "Tianbao" }
    ],
    "issued": { "date-parts": [[2024, 4]] },
    "URL": "https://arxiv.org/abs/2404.07972",
    "container-title": "NeurIPS 2024",
    "note": "Section 3 — 핵심 벤치마크, 인간 기준선 72.36%"
  },
  {
    "id": "webarena-2023",
    "type": "article-journal",
    "title": "WebArena: A Realistic Web Environment for Building Autonomous Agents",
    "author": [
      { "family": "Zhou", "given": "Shuyan" }
    ],
    "issued": { "date-parts": [[2023, 7]] },
    "URL": "https://arxiv.org/abs/2307.13854",
    "note": "Section 3 — 웹 환경 벤치마크"
  },
  {
    "id": "screenspotpro-2025",
    "type": "article-journal",
    "title": "ScreenSpot-Pro: GUI Grounding for Professional High-Resolution Computer Use",
    "author": [
      { "family": "Li", "given": "Kaixin" }
    ],
    "issued": { "date-parts": [[2025, 4]] },
    "URL": "https://arxiv.org/abs/2504.07981",
    "note": "Section 3 — 고해상도 GUI 그라운딩 벤치마크"
  },
  {
    "id": "androidworld-2024",
    "type": "article-journal",
    "title": "AndroidWorld: A Dynamic Benchmarking Environment for Autonomous Agents",
    "author": [
      { "literal": "Google Research" }
    ],
    "issued": { "date-parts": [[2024, 5]] },
    "URL": "https://arxiv.org/abs/2405.14573",
    "note": "Section 3 — UI-TARS-2 AndroidWorld 73.3% 근거"
  },
  {
    "id": "rt2-2023",
    "type": "paper-conference",
    "title": "RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control",
    "author": [
      { "family": "Brohan", "given": "Anthony" },
      { "literal": "Google DeepMind" }
    ],
    "issued": { "date-parts": [[2023, 7]] },
    "URL": "https://arxiv.org/abs/2307.15818",
    "container-title": "CoRL 2023",
    "note": "Section 5 — VLA 패러다임의 원전"
  },
  {
    "id": "openvla-2024",
    "type": "article-journal",
    "title": "OpenVLA: An Open-Source Vision-Language-Action Model",
    "author": [
      { "family": "Kim", "given": "Moo Jin" },
      { "family": "Pertsch", "given": "Karl" }
    ],
    "issued": { "date-parts": [[2024, 6]] },
    "URL": "https://arxiv.org/abs/2406.09246",
    "note": "Section 5 — 오픈소스 VLA. UI-TARS는 'GUI판 OpenVLA'"
  },
  {
    "id": "pi0-2024",
    "type": "article-journal",
    "title": "π0: A Vision-Language-Action Flow Model for General Robot Control",
    "author": [
      { "literal": "Physical Intelligence" }
    ],
    "issued": { "date-parts": [[2024, 10]] },
    "URL": "https://arxiv.org/abs/2410.24164",
    "note": "Section 5 — Dual-System 아키텍처 동형성"
  },
  {
    "id": "groot-n1-2025",
    "type": "article-journal",
    "title": "GR00T N1: An Open Foundation Model for Generalist Humanoid Robots",
    "author": [
      { "literal": "NVIDIA" }
    ],
    "issued": { "date-parts": [[2025, 3]] },
    "URL": "https://arxiv.org/abs/2503.14734",
    "note": "Section 5 — NVIDIA의 휴머노이드 파운데이션 모델"
  },
  {
    "id": "gemini-robotics-2025",
    "type": "article-journal",
    "title": "Gemini Robotics: Bringing AI into the Physical World",
    "author": [
      { "literal": "Google DeepMind" }
    ],
    "issued": { "date-parts": [[2025, 3]] },
    "URL": "https://arxiv.org/abs/2503.20020",
    "note": "Section 5 — Google의 물리 AI 진출"
  },
  {
    "id": "mllm-halluc-survey-2024",
    "type": "article-journal",
    "title": "Hallucination of Multimodal Large Language Models: A Survey",
    "author": [
      { "family": "Bai", "given": "Zechen" }
    ],
    "issued": { "date-parts": [[2024, 4]] },
    "URL": "https://arxiv.org/abs/2404.18930",
    "note": "Section 4 — 멀티모달 환각의 분류 체계"
  },
  {
    "id": "agent-halluc-survey-2025",
    "type": "article-journal",
    "title": "LLM-based Agents Suffer from Hallucinations: A Survey of Taxonomy, Methods, and Directions",
    "issued": { "date-parts": [[2025, 9]] },
    "URL": "https://arxiv.org/abs/2509.18970",
    "note": "Section 4 — 자가강화 오염(recursive memory-conditioning) 핵심 인용"
  },
  {
    "id": "gui-agent-survey-2024",
    "type": "article-journal",
    "title": "Large Language Model-Brained GUI Agents: A Survey",
    "issued": { "date-parts": [[2024, 11]] },
    "URL": "https://arxiv.org/abs/2411.18279",
    "note": "Section 2 — GUI 에이전트 계보 정리"
  },
  {
    "id": "anthropic-computer-use-2024",
    "type": "webpage",
    "title": "Introducing computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku",
    "author": [
      { "literal": "Anthropic" }
    ],
    "issued": { "date-parts": [[2024, 10, 22]] },
    "URL": "https://www.anthropic.com/news/3-5-models-and-computer-use",
    "note": "Section 3 — Computer Use 공식 발표"
  },
  {
    "id": "openai-operator-2025",
    "type": "webpage",
    "title": "Introducing Operator",
    "author": [
      { "literal": "OpenAI" }
    ],
    "issued": { "date-parts": [[2025, 1, 23]] },
    "URL": "https://openai.com/index/introducing-operator/",
    "note": "Section 3 — OpenAI Operator 발표"
  },
  {
    "id": "google-mariner-2024",
    "type": "webpage",
    "title": "Project Mariner",
    "author": [
      { "literal": "Google DeepMind" }
    ],
    "issued": { "date-parts": [[2024, 12, 11]] },
    "URL": "https://deepmind.google/models/project-mariner/",
    "note": "Section 3 — Google Mariner 공식 페이지"
  },
  {
    "id": "microsoft-magentic-ui-2025",
    "type": "webpage",
    "title": "Magentic-UI, an experimental human-centered web agent",
    "author": [
      { "literal": "Microsoft Research" }
    ],
    "issued": { "date-parts": [[2025, 5, 19]] },
    "URL": "https://www.microsoft.com/en-us/research/blog/magentic-ui-an-experimental-human-centered-web-agent/",
    "note": "Section 3 — MIT 라이선스 오픈소스 GUI 에이전트"
  },
  {
    "id": "uipath-agentic-2025",
    "type": "report",
    "title": "UiPath Launches the First Enterprise-Grade Platform for Agentic Automation",
    "author": [
      { "literal": "UiPath" }
    ],
    "issued": { "date-parts": [[2025, 9]] },
    "URL": "https://www.uipath.com/newsroom/uipath-launches-first-enterprise-grade-platform-for-agentic-automation",
    "note": "Section 7 — RPA 진영의 AI Agent 전환"
  },
  {
    "id": "gartner-agentic-hype-2026",
    "type": "report",
    "title": "Hype Cycle for Agentic AI 2026",
    "author": [
      { "literal": "Gartner" }
    ],
    "issued": { "date-parts": [[2026]] },
    "URL": "https://www.gartner.com/en/articles/hype-cycle-for-agentic-ai",
    "note": "Exec Summary, Section 7 — 40% 취소 경고"
  },
  {
    "id": "gartner-press-2025-08-26",
    "type": "webpage",
    "title": "Gartner Predicts 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026",
    "author": [
      { "literal": "Gartner" }
    ],
    "issued": { "date-parts": [[2025, 8, 26]] },
    "URL": "https://www.gartner.com/en/newsroom/press-releases/2025-08-26-gartner-predicts-40-percent-of-enterprise-apps-will-feature-task-specific-ai-agents-by-2026",
    "note": "Section 6 — 엔터프라이즈 에이전트 채택 전망"
  },
  {
    "id": "marketsandmarkets-ai-agents",
    "type": "report",
    "title": "AI Agents Market by Application, Geography, Technology — Global Forecast to 2030",
    "author": [
      { "literal": "MarketsandMarkets" }
    ],
    "issued": { "date-parts": [[2025]] },
    "URL": "https://www.marketsandmarkets.com/Market-Reports/ai-agents-market-15761548.html",
    "note": "Exec Summary, Section 6 — 글로벌 $7.84B→$52.6B, CAGR 46.3%"
  },
  {
    "id": "grand-view-korea-ai-agents",
    "type": "report",
    "title": "South Korea AI Agents Market Size & Outlook, 2024-2033",
    "author": [
      { "literal": "Grand View Research" }
    ],
    "issued": { "date-parts": [[2025]] },
    "URL": "https://www.grandviewresearch.com/horizon/outlook/ai-agents-market/south-korea",
    "note": "Section 6, 7 — 한국 CAGR 59.1%"
  },
  {
    "id": "github-uitars-desktop",
    "type": "webpage",
    "title": "bytedance/UI-TARS-desktop: The Open-Source Multimodal AI Agent Stack",
    "author": [
      { "literal": "ByteDance" }
    ],
    "URL": "https://github.com/bytedance/UI-TARS-desktop",
    "accessed": { "date-parts": [[2026, 5, 13]] },
    "note": "Exec Summary — 33,573 stars GitHub API 직접 조회"
  }
]