{
  "title": "The agentic internet: a demand-driven scenario model, 2026–2036",
  "research_date": "2026-10-09",
  "kind": "conditional scenario forecast",
  "scope": "Public-facing application requests to websites and internet-facing APIs, counted once per request. Not bandwidth, video bytes, internal model reasoning, spending or human attention.",
  "caveat": "Conditional scenario forecasts, not measured probabilities or an industry census. Annual rows are reference snapshots. The 2026 starting split (47% human, 1% task agents, 52% other automation) is an assumed normalization.",
  "planning_case": [
    {
      "year": 2026,
      "status": "assumed normalization",
      "adoption_pct": 0.8,
      "jobs": 1,
      "fanout": 1,
      "requests_per_worker": 1,
      "business_ratio": 0.2,
      "workload_index": 1,
      "human_pct": 47,
      "other_pct": 52,
      "agent_pct": 1,
      "automated_pct": 53,
      "total_index": 100,
      "human_volume": 47,
      "agent_volume": 1,
      "agent_yoy_pct": null
    },
    {
      "year": 2027,
      "status": "conditional forecast",
      "adoption_pct": 4,
      "jobs": 1.4,
      "fanout": 1.4,
      "requests_per_worker": 0.93,
      "business_ratio": 0.28,
      "workload_index": 1.944,
      "human_pct": 41.92,
      "other_pct": 49.19,
      "agent_pct": 8.89,
      "automated_pct": 58.08,
      "total_index": 111.6,
      "human_volume": 46.78,
      "agent_volume": 9.92,
      "agent_yoy_pct": 892
    },
    {
      "year": 2028,
      "status": "conditional forecast",
      "adoption_pct": 12,
      "jobs": 2,
      "fanout": 1.8,
      "requests_per_worker": 0.85,
      "business_ratio": 0.37,
      "workload_index": 3.494,
      "human_pct": 28.47,
      "other_pct": 36.86,
      "agent_pct": 34.67,
      "automated_pct": 71.53,
      "total_index": 157.2,
      "human_volume": 44.77,
      "agent_volume": 54.52,
      "agent_yoy_pct": 450
    },
    {
      "year": 2029,
      "status": "conditional forecast",
      "adoption_pct": 25,
      "jobs": 2.6,
      "fanout": 2.2,
      "requests_per_worker": 0.78,
      "business_ratio": 0.46,
      "workload_index": 5.428,
      "human_pct": 14.46,
      "other_pct": 21.7,
      "agent_pct": 63.84,
      "automated_pct": 85.54,
      "total_index": 282,
      "human_volume": 40.77,
      "agent_volume": 180.02,
      "agent_yoy_pct": 230
    },
    {
      "year": 2030,
      "status": "conditional forecast",
      "adoption_pct": 40,
      "jobs": 3.2,
      "fanout": 2.6,
      "requests_per_worker": 0.72,
      "business_ratio": 0.55,
      "workload_index": 7.738,
      "human_pct": 6.9,
      "other_pct": 12.44,
      "agent_pct": 80.66,
      "automated_pct": 93.1,
      "total_index": 519.2,
      "human_volume": 35.83,
      "agent_volume": 418.77,
      "agent_yoy_pct": 133
    },
    {
      "year": 2031,
      "status": "conditional forecast",
      "adoption_pct": 53,
      "jobs": 3.7,
      "fanout": 2.9,
      "requests_per_worker": 0.67,
      "business_ratio": 0.64,
      "workload_index": 9.825,
      "human_pct": 3.84,
      "other_pct": 8.33,
      "agent_pct": 87.82,
      "automated_pct": 96.16,
      "total_index": 818.3,
      "human_volume": 31.45,
      "agent_volume": 718.66,
      "agent_yoy_pct": 72
    },
    {
      "year": 2032,
      "status": "conditional forecast",
      "adoption_pct": 63,
      "jobs": 4.2,
      "fanout": 3.2,
      "requests_per_worker": 0.62,
      "business_ratio": 0.72,
      "workload_index": 11.944,
      "human_pct": 2.42,
      "other_pct": 6.21,
      "agent_pct": 91.37,
      "automated_pct": 97.58,
      "total_index": 1159.3,
      "human_volume": 28.09,
      "agent_volume": 1059.23,
      "agent_yoy_pct": 47
    },
    {
      "year": 2033,
      "status": "conditional forecast",
      "adoption_pct": 70,
      "jobs": 4.7,
      "fanout": 3.45,
      "requests_per_worker": 0.58,
      "business_ratio": 0.8,
      "workload_index": 14.107,
      "human_pct": 1.7,
      "other_pct": 5,
      "agent_pct": 93.3,
      "automated_pct": 98.3,
      "total_index": 1519.7,
      "human_volume": 25.8,
      "agent_volume": 1417.9,
      "agent_yoy_pct": 34
    },
    {
      "year": 2034,
      "status": "conditional forecast",
      "adoption_pct": 75,
      "jobs": 5.2,
      "fanout": 3.65,
      "requests_per_worker": 0.55,
      "business_ratio": 0.88,
      "workload_index": 16.354,
      "human_pct": 1.28,
      "other_pct": 4.22,
      "agent_pct": 94.5,
      "automated_pct": 98.72,
      "total_index": 1900.9,
      "human_volume": 24.24,
      "agent_volume": 1796.42,
      "agent_yoy_pct": 27
    },
    {
      "year": 2035,
      "status": "conditional forecast",
      "adoption_pct": 79,
      "jobs": 5.6,
      "fanout": 3.85,
      "requests_per_worker": 0.52,
      "business_ratio": 0.94,
      "workload_index": 18.125,
      "human_pct": 1.02,
      "other_pct": 3.77,
      "agent_pct": 95.21,
      "automated_pct": 98.98,
      "total_index": 2246.7,
      "human_volume": 23.03,
      "agent_volume": 2139,
      "agent_yoy_pct": 19
    },
    {
      "year": 2036,
      "status": "conditional forecast",
      "adoption_pct": 82,
      "jobs": 6,
      "fanout": 4,
      "requests_per_worker": 0.5,
      "business_ratio": 1,
      "workload_index": 20,
      "human_pct": 0.85,
      "other_pct": 3.43,
      "agent_pct": 95.72,
      "automated_pct": 99.15,
      "total_index": 2610.5,
      "human_volume": 22.19,
      "agent_volume": 2498.94,
      "agent_yoy_pct": 17
    }
  ],
  "scenarios": {
    "constrained": [
      {
        "year": 2026,
        "users_pct": 0.8,
        "work": 1,
        "agent_pct": 1,
        "automated_pct": 53
      },
      {
        "year": 2027,
        "users_pct": 2,
        "work": 1.2,
        "agent_pct": 2.9,
        "automated_pct": 54.95
      },
      {
        "year": 2028,
        "users_pct": 5,
        "work": 1.5,
        "agent_pct": 8.48,
        "automated_pct": 58.85
      },
      {
        "year": 2029,
        "users_pct": 10,
        "work": 1.8,
        "agent_pct": 18.16,
        "automated_pct": 64.7
      },
      {
        "year": 2030,
        "users_pct": 16,
        "work": 2.1,
        "agent_pct": 29.31,
        "automated_pct": 70.96
      },
      {
        "year": 2031,
        "users_pct": 23,
        "work": 2.5,
        "agent_pct": 41.6,
        "automated_pct": 77.35
      },
      {
        "year": 2032,
        "users_pct": 30,
        "work": 2.9,
        "agent_pct": 51.95,
        "automated_pct": 82.49
      },
      {
        "year": 2033,
        "users_pct": 37,
        "work": 3.3,
        "agent_pct": 60.34,
        "automated_pct": 86.49
      },
      {
        "year": 2034,
        "users_pct": 44,
        "work": 3.7,
        "agent_pct": 67.02,
        "automated_pct": 89.57
      },
      {
        "year": 2035,
        "users_pct": 50,
        "work": 4.1,
        "agent_pct": 71.84,
        "automated_pct": 91.71
      },
      {
        "year": 2036,
        "users_pct": 55,
        "work": 4.5,
        "agent_pct": 75.36,
        "automated_pct": 93.23
      }
    ],
    "central": [
      {
        "year": 2026,
        "users_pct": 0.8,
        "work": 1,
        "agent_pct": 1,
        "automated_pct": 53
      },
      {
        "year": 2027,
        "users_pct": 3,
        "work": 1.7,
        "agent_pct": 5.99,
        "automated_pct": 56.57
      },
      {
        "year": 2028,
        "users_pct": 9,
        "work": 2.8,
        "agent_pct": 23.99,
        "automated_pct": 66.42
      },
      {
        "year": 2029,
        "users_pct": 20,
        "work": 4,
        "agent_pct": 50.54,
        "automated_pct": 79.69
      },
      {
        "year": 2030,
        "users_pct": 34,
        "work": 5.5,
        "agent_pct": 71.12,
        "automated_pct": 89.28
      },
      {
        "year": 2031,
        "users_pct": 47,
        "work": 7,
        "agent_pct": 81.66,
        "automated_pct": 93.92
      },
      {
        "year": 2032,
        "users_pct": 57,
        "work": 8.5,
        "agent_pct": 86.94,
        "automated_pct": 96.11
      },
      {
        "year": 2033,
        "users_pct": 65,
        "work": 10,
        "agent_pct": 89.99,
        "automated_pct": 97.32
      },
      {
        "year": 2034,
        "users_pct": 71,
        "work": 11.5,
        "agent_pct": 91.85,
        "automated_pct": 98.01
      },
      {
        "year": 2035,
        "users_pct": 75,
        "work": 13,
        "agent_pct": 93.01,
        "automated_pct": 98.42
      },
      {
        "year": 2036,
        "users_pct": 78,
        "work": 14.5,
        "agent_pct": 93.83,
        "automated_pct": 98.7
      }
    ],
    "hyper": [
      {
        "year": 2026,
        "users_pct": 0.8,
        "work": 1,
        "agent_pct": 1,
        "automated_pct": 53
      },
      {
        "year": 2027,
        "users_pct": 7,
        "work": 3,
        "agent_pct": 21.02,
        "automated_pct": 64.12
      },
      {
        "year": 2028,
        "users_pct": 25,
        "work": 8,
        "agent_pct": 72.65,
        "automated_pct": 88.84
      },
      {
        "year": 2029,
        "users_pct": 50,
        "work": 16,
        "agent_pct": 91.98,
        "automated_pct": 97.28
      },
      {
        "year": 2030,
        "users_pct": 70,
        "work": 28,
        "agent_pct": 96.76,
        "automated_pct": 99.11
      },
      {
        "year": 2031,
        "users_pct": 80,
        "work": 45,
        "agent_pct": 98.24,
        "automated_pct": 99.59
      },
      {
        "year": 2032,
        "users_pct": 86,
        "work": 65,
        "agent_pct": 98.86,
        "automated_pct": 99.76
      },
      {
        "year": 2033,
        "users_pct": 90,
        "work": 90,
        "agent_pct": 99.2,
        "automated_pct": 99.85
      },
      {
        "year": 2034,
        "users_pct": 92,
        "work": 120,
        "agent_pct": 99.4,
        "automated_pct": 99.89
      },
      {
        "year": 2035,
        "users_pct": 93,
        "work": 150,
        "agent_pct": 99.51,
        "automated_pct": 99.92
      },
      {
        "year": 2036,
        "users_pct": 94,
        "work": 180,
        "agent_pct": 99.59,
        "automated_pct": 99.93
      }
    ]
  },
  "scenario_summary": [
    {
      "key": "constrained",
      "name": "Constrained adoption",
      "short": "Constrained",
      "agent_2030": 29.3,
      "agent_2036": 75.4,
      "first_99_agent_year": null
    },
    {
      "key": "central",
      "name": "Central judgment",
      "short": "Central",
      "agent_2030": 71.1,
      "agent_2036": 93.8,
      "first_99_agent_year": null
    },
    {
      "key": "planning",
      "name": "Planning case (slightly above central)",
      "short": "Planning case",
      "agent_2030": 80.7,
      "agent_2036": 95.7,
      "first_99_agent_year": null
    },
    {
      "key": "hyper",
      "name": "Hyper-adoption, heavier workloads",
      "short": "Hyper-adoption",
      "agent_2030": 96.8,
      "agent_2036": 99.6,
      "first_99_agent_year": 2033
    }
  ],
  "sensitivities": [
    {
      "label": "Starting task-agent share 0.25% instead of 1%",
      "agent_2036": 84.7,
      "automated_2036": 97
    },
    {
      "label": "Requests per worker fall 95% instead of 50%",
      "agent_2036": 69.1,
      "automated_2036": 93.9
    }
  ],
  "workload_2036": {
    "jobs": 6,
    "fanout": 4,
    "requests_per_worker": 0.5,
    "business_factor": 1.67,
    "total": 20
  },
  "headlines": {
    "agent_2029": 63.84,
    "agent_2030": 80.66,
    "agent_2036": 95.72,
    "automated_2036": 99.15,
    "total_index_2036_multiple": 26.1,
    "adoption_2029": 25,
    "human_volume_2029_vs_2026_pct": -13,
    "human_volume_2036_of_2026_pct": 47
  },
  "model": {
    "base_year": 2026,
    "baseline": {
      "human": 47,
      "agent": 1,
      "other": 52
    },
    "base_adoption": 0.008,
    "base_business_ratio": 0.2,
    "human_displacement": 0.75,
    "user_growth": 0.02,
    "other_growth": 0.035
  },
  "equations": [
    "K = J × F × E × (1 + b) / (1 + b₂₀₂₆)",
    "N = 1.02^t",
    "A = A₂₀₂₆ × N × (p / p₂₀₂₆) × K",
    "H = H₂₀₂₆ × N × (1 − 0.75p) / (1 − 0.75p₂₀₂₆)",
    "O = O₂₀₂₆ × N × 1.035^t",
    "Total = A + H + O; each share = category / Total; all automated = (A + O) / Total"
  ],
  "evidence": [
    {
      "stat": "5M",
      "label": "estimated U.S. downloads of Meta's Muse in 22 days",
      "detail": "Sensor Tower's comparable milestones: 56 days for ChatGPT, 103 for Grok, 492 for Claude. Downloads, not retained agent users.",
      "source": "S03"
    },
    {
      "stat": "3M+",
      "label": "weekly prompting users of Muse; 1M+ daily",
      "detail": "Reported from internal Meta data; not an independently audited census.",
      "source": "S04"
    },
    {
      "stat": "72%",
      "label": "of daily identified agent requests came from Muse in HUMAN's early window",
      "detail": "A share of identified agent traffic, not of the whole internet.",
      "source": "S02"
    },
    {
      "stat": "17.7B",
      "label": "AI requests in Q2 2026, up 45% quarter over quarter",
      "detail": "Includes crawlers and indexers; not 17.7 billion autonomous tasks.",
      "source": "S05"
    }
  ],
  "sources": [
    {
      "id": "S01",
      "type": "Observed, vendor sample",
      "publisher": "Thales, 2026 Bad Bot Report",
      "period": "2025 traffic",
      "finding": "47% human; 53% automated. Used only as a historical normalization reference, not a 2026 global census.",
      "url": "https://cpl.thalesgroup.com/resources/application-security/2026-bad-bot-report"
    },
    {
      "id": "S02",
      "type": "Observed, vendor sample",
      "publisher": "HUMAN, Muse early observations",
      "period": "2026-09-28",
      "finding": "Muse averaged 72% of daily identified AI-agent requests in the early window; not 72% of total internet traffic.",
      "url": "https://www.humansecurity.com/learn/blog/meta-muse-agentic-traffic/"
    },
    {
      "id": "S03",
      "type": "Reported third-party estimate",
      "publisher": "Forbes reporting Sensor Tower",
      "period": "2026-09-30",
      "finding": "Muse estimated 5 million U.S. downloads in 22 days. Downloads are not active users.",
      "url": "https://www.forbes.com/sites/maryroeloffs/2026/09/30/metas-muse-ai-assistant-hits-5-million-downloads-outpacing-growth-of-chatgpt-grok-and-claude/"
    },
    {
      "id": "S04",
      "type": "Reported internal usage",
      "publisher": "TNW reporting The Information",
      "period": "2026-10-02",
      "finding": "Over 3 million weekly prompting users and over 1 million daily prompting users; not independently audited.",
      "url": "https://thenextweb.com/news/meta-muse-3-million-weekly-users-information"
    },
    {
      "id": "S05",
      "type": "Observed, mixed AI category",
      "publisher": "DataDome Q2 AI Traffic Report",
      "period": "Q2 2026",
      "finding": "17.7 billion AI requests, up 45% quarter on quarter. Category includes AI crawlers/indexers; not all autonomous task execution.",
      "url": "https://datadome.co/threat-research/ai-traffic-report-q2-2026/"
    },
    {
      "id": "S06",
      "type": "Product announcement",
      "publisher": "Meta, Introducing Muse",
      "period": "2026-09-08; updated 2026-09-30",
      "finding": "Persistent secure computer, browser, and WhatsApp access; user controls permissions. Distribution is not equivalent to adoption.",
      "url": "https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/"
    },
    {
      "id": "S07",
      "type": "Product announcement",
      "publisher": "OpenAI, Introducing dots",
      "period": "2026-09-29",
      "finding": "Proactive assistants included with eligible Pro/Business Premium plans; work allowances apply.",
      "url": "https://openai.com/index/introducing-dots/"
    },
    {
      "id": "S08",
      "type": "Product documentation",
      "publisher": "xAI, Grok Bot",
      "period": "Accessed 2026-10-09",
      "finding": "Persistent cloud-computer agents and parallel work; no deduplicated global adoption statistic used.",
      "url": "https://docs.x.ai/grok-bot/overview"
    },
    {
      "id": "S09",
      "type": "Technical documentation",
      "publisher": "OpenClaw, Sub-agents",
      "period": "Accessed 2026-10-09",
      "finding": "Supports child agents and controlled concurrency/depth. Spawning is not unlimited replication.",
      "url": "https://docs.openclaw.ai/tools/subagents"
    },
    {
      "id": "S10",
      "type": "Technical documentation",
      "publisher": "Nous Research, Hermes delegation",
      "period": "Accessed 2026-10-09",
      "finding": "Child agents have isolated contexts; inherited tool permissions. Counts are not unique people.",
      "url": "https://hermes-agent.nousresearch.com/docs/user-guide/features/delegation"
    },
    {
      "id": "S11",
      "type": "Provider engineering measurements",
      "publisher": "Anthropic, Multi-agent research system",
      "period": "2025-06-13",
      "finding": "Reported approximately 4x agent and 15x multi-agent token use versus chats. Tokens are not HTTP requests.",
      "url": "https://www.anthropic.com/engineering/multi-agent-research-system"
    },
    {
      "id": "S12",
      "type": "Benchmark research",
      "publisher": "METR, Time horizons",
      "period": "Updated 2026-09-08",
      "finding": "Measures software-task completion at stated reliability levels; not a measure of whole-job replacement.",
      "url": "https://metr.org/time-horizons/"
    },
    {
      "id": "S13",
      "type": "Technical standard announcement",
      "publisher": "Google, Agent2Agent",
      "period": "2025-04-09",
      "finding": "Inter-agent communication standard. Existence does not prove universal adoption.",
      "url": "https://developers.googleblog.com/en/a2a-a-new-era-of-agent-interoperability/"
    },
    {
      "id": "S14",
      "type": "Technical standard announcement",
      "publisher": "Google Cloud, AP2",
      "period": "2025-09-16",
      "finding": "Verifiable intent/cart mandates support bounded payment authority and evidence.",
      "url": "https://cloud.google.com/blog/products/ai-machine-learning/announcing-agents-to-payments-ap2-protocol"
    },
    {
      "id": "S15",
      "type": "Product announcement",
      "publisher": "Cloudflare, Pay Per Use",
      "period": "2026; accessed 2026-10-09",
      "finding": "Content compensation based on use rather than merely crawling; adoption not assumed universal.",
      "url": "https://blog.cloudflare.com/pay-per-use/"
    },
    {
      "id": "S16",
      "type": "Economic research",
      "publisher": "Stanford Digital Economy Lab, Canaries update",
      "period": "2026-08-12; employment through June 2026",
      "finding": "19% kept-pace employment shortfall for young workers in highly exposed occupations; no economy-wide displacement finding. Not a causal AI-only estimate.",
      "url": "https://digitaleconomy.stanford.edu/news/canariesaug26/"
    },
    {
      "id": "S17",
      "type": "Occupational exposure research",
      "publisher": "ILO / NASK",
      "period": "2025-05-20",
      "finding": "Approximately one quarter of jobs show some GenAI exposure; exposure is not job loss.",
      "url": "https://www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-occupational-exposure"
    },
    {
      "id": "S18",
      "type": "External energy projection",
      "publisher": "IEA, Key Questions on Energy and AI",
      "period": "2026",
      "finding": "Data-centre electricity demand projected from 485 TWh in 2025 to around 950 TWh in 2030; not an agent-specific energy forecast.",
      "url": "https://www.iea.org/reports/key-questions-on-energy-and-ai/executive-summary"
    },
    {
      "id": "S19",
      "type": "Research synthesis",
      "publisher": "Stanford AI Index",
      "period": "2025 report; historical 2022–2024 costs",
      "finding": "Fixed GPT-3.5-level benchmark inference costs fell more than 280-fold. Not a universal ongoing cost decline.",
      "url": "https://hai.stanford.edu/ai-index/2025-ai-index-report"
    },
    {
      "id": "S20",
      "type": "Safety / coordination research",
      "publisher": "Anthropic, Multiagent systems",
      "period": "2026-09",
      "finding": "Experiments document both collaboration benefits and coordination/collusion failure modes.",
      "url": "https://www.anthropic.com/research/multiagent-systems"
    }
  ],
  "notice": "Conditional scenario forecasts from Rightshold's planning model, not measurements. The 2026 starting split is an assumed normalization.",
  "page": "https://www.rightshold.com/research/agentic-internet",
  "downloads": {
    "csv": "https://www.rightshold.com/research/agentic-internet.csv",
    "xlsx": "https://www.rightshold.com/research/agentic-internet.xlsx"
  }
}