{
  "total": 49,
  "limit": 50,
  "offset": 0,
  "records": [
    {
      "fields": {
        "buyer": "frontier labs and AI companies",
        "category": "rlhf-expert",
        "funding_usd": {
          "availability": "undisclosed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "reported round at a $3.2B valuation (Forbes, 2026-09-01, amount not disclosed); $30M Series A at $300M valuation, April 2026"
        },
        "hq": "San Francisco, CA",
        "name": "AfterQuery",
        "note": "YC W25. Claimed $100M annualized revenue run rate in April 2026 (company claim). The $3.2B round was reported by Forbes and not confirmed by the company at publication.",
        "orgs": [
          "frontier labs and AI companies"
        ],
        "sells": "Expert-produced training data and environments teaching models and agents to work like professionals (doctors, lawyers and other specialists)",
        "status": "operating",
        "url": "https://techcrunch.com/2026/09/01/afterquery-reportedly-becomes-y-combinators-fastest-ever-unicorn-now-valued-at-3-2b",
        "verticals": [
          "LLM/RLHF",
          "agents"
        ],
        "workforce": "expert"
      },
      "id": "vendor:afterquery",
      "name": "AfterQuery",
      "notes": {
        "_org_roles": {
          "frontier labs and AI companies": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-17",
        "last_seen": "2026-09-17",
        "method": "refinery-explore",
        "source_hosts": [
          "techcrunch.com"
        ],
        "source_urls": [
          "https://techcrunch.com/2026/09/01/afterquery-reportedly-becomes-y-combinators-fastest-ever-unicorn-now-valued-at-3-2b"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.727,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "prose-funding_usd",
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://techcrunch.com/2026/09/01/afterquery-reportedly-becomes-y-combinators-fastest-ever-unicorn-now-valued-at-3-2b"
    },
    {
      "fields": {
        "buyer": "Labelbox's customers",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "rlhf-expert",
        "name": "Alignerr (Labelbox)",
        "note": "Header reads 'Alignerr | Powered by Labelbox'. Not found as a separately incorporated company.",
        "orgs": [
          "Labelbox's customers"
        ],
        "sells": "Labelbox's expert contributor network for evaluation, red-teaming, prompt writing and code review",
        "status": "operating",
        "status_note": "Reported by workers or reporters, not the vendor. Sample: 81. Reported average about $37/hr vs advertised '$80/hr average pay': the vendor's average is 2.2x the community average.",
        "update_note": "updated 2026-09-17 from breakingeven.online: Reported average about $37/hr vs advertised '$80/hr average pay': the vendor's average is 2.2x the community average.",
        "url": "https://www.alignerr.com",
        "verticals": [
          "coding",
          "STEM",
          "AI safety review",
          "languages",
          "video/audio annotation"
        ],
        "worker_pay": {
          "availability": "listed",
          "basis": "per-hour",
          "currency": "USD",
          "kind": "point",
          "max": 80,
          "min": 80,
          "note": "parsed from the quoted excerpt: $80/hr average pay",
          "raw": "vendor page: '$80/hr average pay', '$1M paid weekly'; listings show $30-50/hr, $40-120/hr and $30-35/hr depending on role"
        },
        "worker_pay_reported": "community-reported range widens to $15-$60/hr with an average around $37/hr; voice work billed Per Finished Hour 'about $9 per real hour worked'",
        "worker_pay_reported_sample": 81,
        "worker_pay_reported_source": "reddit aggregator (single-author blog), 90 days ending 2026-08-16",
        "workforce": "expert"
      },
      "id": "vendor:alignerr-labelbox",
      "name": "Alignerr (Labelbox)",
      "notes": {
        "_org_roles": {
          "Labelbox's customers": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "alignerr.com",
          "breakingeven.online",
          "pulse2.com"
        ],
        "source_urls": [
          "https://breakingeven.online/blog/alignerr-review-2026",
          "https://www.alignerr.com",
          "https://pulse2.com/labelbox-acquires-agentic-sales-automation-startup-upcraft-to-scale-alignerr-expert-network"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 1,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 3,
        "independent_hosts": 3,
        "sources": 3,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.alignerr.com"
    },
    {
      "fields": {
        "buyer": "researchers and enterprises needing human annotation",
        "category": "crowd-platform",
        "hq": "Seattle, WA",
        "name": "Amazon Mechanical Turk",
        "note": "Closed to new customers 2026-07-30 (TechCrunch, 2026-07-05); closes 2026-09-30 after 21 years, and the MTurk worker type disappears from SageMaker labeling jobs and A2I review workflows the same day, per AWS's closure FAQ as reported by Tech Times. Status 'shutting down' is new to the vocabulary; move it to 'closed' after 2026-09-30.",
        "orgs": [
          "researchers and enterprises needing human annotation"
        ],
        "sells": "Crowdsourced micro-task marketplace, including the MTurk workforce option inside SageMaker Ground Truth and Augmented AI",
        "status": "shutting down",
        "url": "https://www.techtimes.com/articles/325645/20260826/amazon-mechanical-turk-will-close-september-30-shutting-down-sagemaker-ground-truth-too.htm",
        "verticals": [
          "annotation",
          "survey research"
        ],
        "workforce": "crowd"
      },
      "id": "vendor:amazon-mechanical-turk",
      "name": "Amazon Mechanical Turk",
      "notes": {
        "_org_roles": {
          "researchers and enterprises needing human annotation": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-17",
        "last_seen": "2026-09-17",
        "method": "refinery-explore",
        "source_hosts": [
          "techtimes.com"
        ],
        "source_urls": [
          "https://www.techtimes.com/articles/325645/20260826/amazon-mechanical-turk-will-close-september-30-shutting-down-sagemaker-ground-truth-too.htm"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.727,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.techtimes.com/articles/325645/20260826/amazon-mechanical-turk-will-close-september-30-shutting-down-sagemaker-ground-truth-too.htm"
    },
    {
      "fields": {
        "buyer": [
          "LLM builders",
          "enterprises"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "crowd-platform",
        "hq": "Sydney, Australia and Kirkland, WA (dual HQ)",
        "name": "Appen",
        "note": "Listed on ASX (APX). Google terminated its contract (reported about US$82.8M), services ceased March 19, 2024, shares fell over 40%. About page still names Meta and Microsoft as clients. Public-company revenue exists in filings but was not read this session.",
        "orgs": [
          "LLM builders",
          "enterprises"
        ],
        "sells": "AI data platform and crowd workforce for alignment, speech, multimodal, physical-AI annotation and evaluation",
        "status": "operating",
        "status_note": "Reported by workers or reporters, not the vendor. Sample: 298. Lowest median of the five platforms ILO surveyed. Stale (CrowdFlower was acquired by Appen in 2019), useful only as a historical baseline. Other reports: Fairwork Cloudwork score 3/10; Appen 'provided evidence about a minimum wage policy, but w (fairwork, 2024-02 to 2025-01) | pay increase from $10 to $14.50 an hour (contract-wide rais",
        "update_note": "updated 2026-09-17 from ilo.org: Lowest median of the five platforms ILO surveyed. Stale (CrowdFlower was acquired by Appen in 2019), useful only as a historical baseline.",
        "url": "https://appen.com/about-us",
        "verticals": [
          "speech",
          "machine translation",
          "search evaluation",
          "computer vision",
          "LLM/agentic"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published on corporate pages read"
        },
        "worker_pay_reported": "median US$1.50/hr paid work (mean 2.65, 298 observations); median US$1.11/hr paid and unpaid work (mean 1.95)",
        "worker_pay_reported_sample": 298,
        "worker_pay_reported_source": "academic/ILO survey, 2017 survey",
        "workforce": "crowd"
      },
      "id": "vendor:appen",
      "name": "Appen",
      "notes": {
        "_org_roles": {
          "LLM builders": [
            "buyer"
          ],
          "enterprises": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "appen.com",
          "ilo.org",
          "kalkine.com.au"
        ],
        "source_urls": [
          "https://www.ilo.org/sites/default/files/wcmsp5/groups/public/@dgreports/@dcomm/@publ/documents/publication/wcms_645337.pdf",
          "https://kalkine.com.au/news/announcements/appen-reports-17-revenue-rise-and-narrowed-loss-in-half-year-to-june-2026",
          "https://appen.com/about-us"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 1,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 3,
        "independent_hosts": 3,
        "sources": 3,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://appen.com/about-us"
    },
    {
      "fields": {
        "acquirer": "Protege",
        "buyer": "AI video-generation companies",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "licensed-corpora",
        "name": "Calliope Networks",
        "note": "Site banner: 'Calliope is now a part of Protege!' Founding member of the Dataset Providers Alliance.",
        "orgs": [
          "AI video-generation companies"
        ],
        "sells": "Aggregated TV/film/news/sports catalog (35,000+ hours) licensed to AI video companies under a 'License to Scrape' model",
        "status": "acquired",
        "status_note": "Confirmed by independent trade press covering Protege's $30M Series A1, which states the Calliope Networks acquisition as fact. An acquirer-side release from 2024-12-18 is indexed but could not be opened, so it is not cited.",
        "update_note": "updated 2026-09-17 from alleywatch.com: previously only the target's own banner; now independent",
        "url": "https://calliopenetworks.ai/",
        "verticals": [
          "video/film/TV",
          "news",
          "sports"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "licensed-corpora"
      },
      "id": "vendor:calliope-networks",
      "name": "Calliope Networks",
      "notes": {
        "_org_roles": {
          "AI video-generation companies": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "alleywatch.com",
          "calliopenetworks.ai"
        ],
        "source_urls": [
          "https://www.alleywatch.com/2026/01/protege-ai-training-data-licensing-real-world-data-bobby-samuels",
          "https://calliopenetworks.ai/"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.909,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://calliopenetworks.ai/"
    },
    {
      "fields": {
        "buyer": [
          "medical device",
          "life sciences and AI companies"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "annotation",
        "name": "Centaur.ai",
        "note": "Formerly Centaur Labs (centaurlabs.com redirects). Claims 100,000+ subject matter experts.",
        "orgs": [
          "medical device",
          "life sciences and AI companies"
        ],
        "sells": "Expert annotation and model evaluation, strongest in healthcare, using competitive performance-ranked labelers",
        "status": "operating",
        "url": "https://centaur.ai/",
        "verticals": [
          "medical devices",
          "life sciences",
          "healthcare",
          "LLMs",
          "robotics"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published; labelers are 'measured and rewarded based on performance', no figures"
        },
        "workforce": "expert"
      },
      "id": "vendor:centaur-ai",
      "name": "Centaur.ai",
      "notes": {
        "_org_roles": {
          "life sciences and AI companies": [
            "buyer"
          ],
          "medical device": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-corroborate",
        "source_hosts": [
          "centaur.ai",
          "itnonline.com"
        ],
        "source_urls": [
          "https://www.itnonline.com/content/ensuring-clinical-ai-safety-scale-deepc-centaur-labs-and-segmed-form-radiology-ai-safety",
          "https://centaur.ai/"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.909,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://centaur.ai/"
    },
    {
      "fields": {
        "buyer": "enterprises",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": "per-hour",
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published; described only as 'predictable hourly pricing' with no rate"
        },
        "category": "annotation",
        "name": "CloudFactory",
        "note": "Founded in Nepal (2010); current HQ not stated on pages read. Managed, not open crowd.",
        "orgs": [
          "enterprises"
        ],
        "sells": "Managed workforce for data curation and annotation plus an AI platform (acquired Hasty)",
        "status": "operating",
        "url": "https://www.cloudfactory.com/",
        "verticals": [
          "computer vision",
          "AV/robotics",
          "insurance",
          "healthcare",
          "manufacturing",
          "e-commerce"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "hybrid"
      },
      "id": "vendor:cloudfactory",
      "name": "CloudFactory",
      "notes": {
        "_org_roles": {
          "enterprises": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-corroborate",
        "source_hosts": [
          "cloudfactory.com",
          "technologyreview.com"
        ],
        "source_urls": [
          "https://www.cloudfactory.com/",
          "https://www.technologyreview.com/2019/08/07/133845/cloudfactory-ddd-samasource-imerit-impact-sourcing-companies-for-data-annotation"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.909,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.cloudfactory.com/"
    },
    {
      "fields": {
        "buyer": "AI companies",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "licensed-corpora",
        "name": "Created by Humans",
        "note": "Launched January 2025 with 50+ named bestselling authors backing it. Member of the Dataset Providers Alliance.",
        "orgs": [
          "AI companies"
        ],
        "sells": "Book/author AI-rights licensing (training, reference and transformative rights) to AI companies",
        "status": "operating",
        "url": "https://www.createdbyhumans.ai/",
        "verticals": [
          "books/text"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published; authors 'get compensated when your work is licensed'"
        },
        "workforce": "licensed-corpora"
      },
      "id": "vendor:created-by-humans",
      "name": "Created by Humans",
      "notes": {
        "_org_roles": {
          "AI companies": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-corroborate",
        "source_hosts": [
          "createdbyhumans.ai",
          "techcrunch.com"
        ],
        "source_urls": [
          "https://www.createdbyhumans.ai/",
          "https://techcrunch.com/2024/06/25/backed-by-david-sacks-garry-tan-and-walter-isaacson-created-by-humans-helps-authors-license-their-creative-work-to-ai-models"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.909,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.createdbyhumans.ai/"
    },
    {
      "fields": {
        "buyer": "AI agents and their operators",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not captured (storefront has its own flow, not read)"
        },
        "category": "broker",
        "name": "Dappier",
        "note": "Adjacent: real-time agentic content monetization rather than bulk corpus licensing. DPA member; listed on DataLicenses.org.",
        "orgs": [
          "AI agents and their operators"
        ],
        "sells": "Agent storefront where AI agents buy publisher content and ad placements, plus conversational AI for publishers",
        "status": "operating",
        "url": "https://www.dappier.com/",
        "verticals": [
          "publishing/text"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "licensed-corpora"
      },
      "id": "vendor:dappier",
      "name": "Dappier",
      "notes": {
        "_org_roles": {
          "AI agents and their operators": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-corroborate",
        "source_hosts": [
          "dappier.com",
          "techcrunch.com"
        ],
        "source_urls": [
          "https://techcrunch.com/2024/06/26/dappier-is-building-a-marketplace-for-publishers-to-sell-their-content-to-llm-builders",
          "https://www.dappier.com/"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.909,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.dappier.com/"
    },
    {
      "fields": {
        "buyer": "not disclosed on vendor page",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "rlhf-expert",
        "name": "DataAnnotation",
        "note": "These are ADVERTISED rates. Third-party review sites report contributors clustering lower (about $14 to $20/hr general, $20 to $40/hr coding); the advertised-vs-reported gap is itself a data point. Buyer identity not disclosed on the site.",
        "orgs": [
          "not disclosed on vendor page"
        ],
        "sells": "Contributor platform for chatbot evaluation, coding and STEM/professional AI training tasks",
        "status": "operating",
        "status_note": "Reported by workers or reporters, not the vendor. Sample: 127. Reported sits ABOVE the published floor ($20/hr) at roughly $29-54/hr: consistent with $20 being a floor and coding/expert tracks paying $40+. Not a contradiction.",
        "update_note": "updated 2026-09-17 from glassdoor.com: Reported sits ABOVE the published floor ($20/hr) at roughly $29-54/hr: consistent with $20 being a floor and coding/expert tracks paying $40+. Not a contradiction.",
        "url": "https://www.dataannotation.tech/faq",
        "verticals": [
          "LLM evaluation",
          "coding",
          "STEM",
          "law",
          "finance",
          "medicine",
          "multilingual"
        ],
        "worker_pay": {
          "availability": "listed",
          "basis": "per-hour",
          "currency": "USD",
          "kind": "point",
          "max": 20,
          "min": 20,
          "note": null,
          "raw": "about $20/hr for general work (also reported: $40 per coding task)"
        },
        "worker_pay_reported": "'based on 127 salaries submitted' (AI Trainer, band $60,499-$111,669/yr); Remote AI Trainer '$112,134 per year or $54 per hour' (glassdoor.ca)",
        "worker_pay_reported_sample": 127,
        "worker_pay_reported_source": "glassdoor, to 2026-09",
        "workforce": "hybrid"
      },
      "id": "vendor:dataannotation",
      "name": "DataAnnotation",
      "notes": {
        "_org_roles": {
          "not disclosed on vendor page": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "dataannotation.tech",
          "glassdoor.com",
          "time.com"
        ],
        "source_urls": [
          "https://time.com/6962608/data-annotation-legit-tech-jobs-ai",
          "https://www.glassdoor.com/Salary/DataAnnotation-AI-Trainer-Salaries-E8605843_D_KO15,25.htm?payPeriod=HOURLY",
          "https://www.dataannotation.tech/faq"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 1,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 3,
        "independent_hosts": 3,
        "sources": 3,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.dataannotation.tech/faq"
    },
    {
      "fields": {
        "buyer": "frontier labs",
        "category": "rl-environments",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "point",
          "max": 43000000,
          "min": 43000000,
          "note": null,
          "raw": "$43M Series A led by Andreessen Horowitz, March 2026"
        },
        "hq": "New York, NY",
        "name": "Deeptune",
        "note": "Acquired by Mercor, announced 2026-07-09, terms undisclosed. Mercor CEO Brendan Foody had invested personally in the Series A with a purchase in mind. Category 'rl-environments' is new to the vocabulary.",
        "orgs": [
          "frontier labs"
        ],
        "sells": "Simulated copies of enterprise software ('training gyms') used as reinforcement-learning environments for AI agents",
        "status": "acquired",
        "url": "https://siliconangle.com/2026/07/09/mercor-buys-deeptune-build-training-environments-ai-agents",
        "verticals": [
          "agents",
          "RL environments"
        ]
      },
      "id": "vendor:deeptune",
      "name": "Deeptune",
      "notes": {
        "_org_roles": {
          "frontier labs": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-17",
        "last_seen": "2026-09-17",
        "method": "refinery-explore",
        "source_hosts": [
          "siliconangle.com"
        ],
        "source_urls": [
          "https://siliconangle.com/2026/07/09/mercor-buys-deeptune-build-training-environments-ai-agents"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.636,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://siliconangle.com/2026/07/09/mercor-buys-deeptune-build-training-environments-ai-agents"
    },
    {
      "fields": {
        "buyer": [
          "AI teams",
          "enterprises"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "licensed-corpora",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "'over $85 million raised' (vendor about page)"
        },
        "hq": "Seattle, WA",
        "name": "Defined.ai",
        "note": "Formerly DefinedCrowd. Jan 27, 2026 vendor release reports 65% revenue growth for 2025: operating, not shut down.",
        "orgs": [
          "AI teams",
          "enterprises"
        ],
        "sells": "Marketplace of licensed speech/NLP/multimodal training data plus annotation, collection and evaluation services",
        "status": "operating",
        "url": "https://defined.ai/press-room/defined-ai-reports-sixty-five-percent-revenue-growth",
        "verticals": [
          "automotive",
          "financial services",
          "gaming",
          "healthcare",
          "retail",
          "robotics"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published; cites '1.6M+ experts worldwide' and a Neevo.ai task sub-brand"
        },
        "workforce": "hybrid"
      },
      "id": "vendor:defined-ai",
      "name": "Defined.ai",
      "notes": {
        "_org_roles": {
          "AI teams": [
            "buyer"
          ],
          "enterprises": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "defined.ai"
        ],
        "source_urls": [
          "https://defined.ai/press-room/defined-ai-reports-sixty-five-percent-revenue-growth"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "prose-funding_usd",
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://defined.ai/press-room/defined-ai-reports-sixty-five-percent-revenue-growth"
    },
    {
      "fields": {
        "buyer": [
          "AI teams",
          "robotics and physical-AI companies"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published: pricing page shows Starter/Team/Enterprise tiers with no dollar amounts; physical-AI and data-collection service pages show no prices"
        },
        "category": "robotics-data",
        "hq": "US and UK entities (Cord Technologies Inc. / Cord Technologies Ltd.)",
        "name": "Encord",
        "note": "CORRECTION LOGGED: '$80,000-$400,000 programs' and '$4,000-$8,000 pilot' figures attributed to Encord elsewhere come from Truelabel's competitor-ranking blog, NOT from Encord. Verified absent from encord.com/physical-ai-data-services and /data-collection-services on 2026-09-16.",
        "orgs": [
          "AI teams",
          "robotics and physical-AI companies"
        ],
        "sells": "Multimodal annotation and data-management platform plus physical-AI data collection (teleoperation, egocentric capture) and data-as-a-service",
        "status": "operating",
        "url": "https://encord.com/pricing",
        "verticals": [
          "physical AI",
          "robotics/humanoids",
          "autonomous vehicles",
          "surgical video",
          "defense",
          "drones"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "hybrid"
      },
      "id": "vendor:encord",
      "name": "Encord",
      "notes": {
        "_org_roles": {
          "AI teams": [
            "buyer"
          ],
          "robotics and physical-AI companies": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "encord.com"
        ],
        "source_urls": [
          "https://encord.com/pricing"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://encord.com/pricing"
    },
    {
      "fields": {
        "buyer": "frontier labs",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "rlhf-expert",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "parent Handshake: $434M total, $200M Series F January 2022 (PitchBook-reported); no separate raise for the AI division"
        },
        "hq": "San Francisco, CA",
        "name": "Handshake AI",
        "note": "Launched 2025 as a division, not a separate entity. CEO-attributed revenue claims (about $50M in first 4 months, Lenny's Newsletter) excluded from structured fields.",
        "orgs": [
          "frontier labs"
        ],
        "sells": "Expert AI training and evaluation data via a paid fellowship drawn from Handshake's student and alumni network",
        "status": "operating",
        "status_note": "2026-06-30: parent Handshake acquired Uplimit (AI-native corporate learning) to build an AI skills academy (Josh Bersin). August 2026: Handshake AI advertised up to $30,000 for professional written work documents for AI training (Business Insider), raising document-ownership questions.",
        "update_note": "updated 2026-09-17 from joshbersin.com: Uplimit acquisition and document-buying programme",
        "url": "https://joinhandshake.com/ai",
        "verticals": [
          "AI evaluation",
          "energy",
          "finance",
          "gaming",
          "PCB design",
          "software",
          "chemistry",
          "medicine"
        ],
        "worker_pay": {
          "availability": "listed",
          "basis": "per-hour",
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "vendor page: 'Up to $40/hr' AI Evaluation Specialists, 'Up to $80/hr' energy and finance professionals, 'Up to $125/hr' game developers and PCB specialists; headline 'Earn up to $100/hr'"
        },
        "workforce": "expert"
      },
      "id": "vendor:handshake-ai",
      "name": "Handshake AI",
      "notes": {
        "_org_roles": {
          "frontier labs": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "joinhandshake.com",
          "joshbersin.com"
        ],
        "source_urls": [
          "https://joshbersin.com/2026/06/handshake-acquires-uplimit-ai-skilling-takes-a-new-direction",
          "https://joinhandshake.com/ai"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.909,
        "confidence": "high",
        "corroborated": true,
        "flags": [
          "prose-funding_usd",
          "prose-worker_pay"
        ],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://joinhandshake.com/ai"
    },
    {
      "fields": {
        "acquirer": "Cloudflare",
        "buyer": "AI developers",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "broker",
        "hq": "London, UK",
        "name": "Human Native AI",
        "note": "Site now reads 'Human Native is joining Cloudflare' (no terms or date on the page; press dates it about January 2026). Being folded into Cloudflare's AI marketplace. Rights-holder mechanics page 404s.",
        "orgs": [
          "AI developers"
        ],
        "sells": "Marketplace connecting publisher and creator content to AI developers via subscriptions, revenue share and per-use APIs",
        "status": "acquired",
        "status_note": "Acquisition stated as completed by the acquirer on 2026-01-15; earlier 'joining Cloudflare' wording was stale.",
        "update_note": "updated 2026-09-16 from blog.cloudflare.com: acquirer states the acquisition as completed",
        "url": "https://www.humannative.ai/",
        "verticals": [
          "text"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published ('fair compensation', no rate)"
        },
        "workforce": "licensed-corpora"
      },
      "id": "vendor:human-native-ai",
      "name": "Human Native AI",
      "notes": {
        "_org_roles": {
          "AI developers": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-corroborate",
        "source_hosts": [
          "blog.cloudflare.com",
          "humannative.ai"
        ],
        "source_urls": [
          "https://www.humannative.ai/",
          "https://blog.cloudflare.com/human-native-joins-cloudflare"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.humannative.ai/"
    },
    {
      "fields": {
        "buyer": [
          "enterprises in AV",
          "healthcare",
          "geospatial",
          "GenAI"
        ],
        "buyer_price": {
          "availability": "quote-only",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states the value was not disclosed",
          "raw": "quote only"
        },
        "category": "annotation",
        "name": "iMerit",
        "note": "Acquired by EXL: 'EXL to Acquire iMerit' announced June 24, 2026; 'EXL Completes Acquisition of iMerit' August 3, 2026. Claims 25,000+ domain experts, explicitly not crowd.",
        "orgs": [
          "enterprises in AV",
          "healthcare",
          "geospatial",
          "GenAI"
        ],
        "sells": "Data annotation (image, video, LiDAR, DICOM, text), RLHF/GenAI evaluation via its Scholars program, Ango Hub platform",
        "status": "acquired",
        "url": "https://imerit.ai/news",
        "verticals": [
          "generative AI",
          "autonomous vehicles",
          "healthcare",
          "robotics",
          "agriculture"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "expert"
      },
      "id": "vendor:imerit",
      "name": "iMerit",
      "notes": {
        "_org_roles": {
          "GenAI": [
            "buyer"
          ],
          "enterprises in AV": [
            "buyer"
          ],
          "geospatial": [
            "buyer"
          ],
          "healthcare": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-corroborate",
        "source_hosts": [
          "exlservice.com",
          "imerit.ai"
        ],
        "source_urls": [
          "https://www.exlservice.com/about/newsroom/exl-completes-acquisition-of-imerit-accelerating-enterprise-ai-leadership",
          "https://imerit.ai/news"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://imerit.ai/news"
    },
    {
      "fields": {
        "buyer": [
          "enterprises",
          "model builders"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "annotation",
        "hq": "Ridgefield Park, NJ",
        "name": "Innodata",
        "note": "Public company, NASDAQ: INOD. Claims 11,500+ in-house subject matter experts (employed, not crowd).",
        "orgs": [
          "enterprises",
          "model builders"
        ],
        "sells": "Annotation, multimodal collection, SFT datasets, red-teaming, RLHF/DPO and off-the-shelf LLM datasets",
        "status": "operating",
        "url": "https://innodata.com/",
        "verticals": [
          "technology",
          "healthcare",
          "legal",
          "banking",
          "automotive",
          "aerospace/defense"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "expert"
      },
      "id": "vendor:innodata",
      "name": "Innodata",
      "notes": {
        "_org_roles": {
          "enterprises": [
            "buyer"
          ],
          "model builders": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "innodata.com"
        ],
        "source_urls": [
          "https://innodata.com/"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://innodata.com/"
    },
    {
      "fields": {
        "buyer": [
          "frontier labs",
          "enterprises",
          "public sector"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "rlhf-expert",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "$100M raised September 2025 led by Vanara Capital, total $144M at over $2B valuation (Invisible's own fundraise page and Bloomberg)"
        },
        "hq": "San Francisco, CA (per Bloomberg, not stated on its about page)",
        "name": "Invisible Technologies",
        "note": "Join page showed 0 roles and no pay; expert work routes to a separate site (meridial.ai) not read.",
        "orgs": [
          "frontier labs",
          "enterprises",
          "public sector"
        ],
        "sells": "RLHF and model-evaluation data for labs plus enterprise operations-as-a-service automation",
        "status": "operating",
        "url": "https://invisibletech.ai/join-us",
        "verticals": [
          "AI training/eval",
          "banking",
          "insurance",
          "healthcare",
          "private equity",
          "public sector"
        ],
        "workforce": "hybrid"
      },
      "id": "vendor:invisible-technologies",
      "name": "Invisible Technologies",
      "notes": {
        "_org_roles": {
          "enterprises": [
            "buyer"
          ],
          "frontier labs": [
            "buyer"
          ],
          "public sector": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-corroborate",
        "source_hosts": [
          "invisibletech.ai",
          "siliconangle.com"
        ],
        "source_urls": [
          "https://invisibletech.ai/join-us",
          "https://siliconangle.com/2025/09/16/ai-data-provider-invisible-raises-100m-2b-valuation"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.818,
        "confidence": "high",
        "corroborated": true,
        "flags": [
          "prose-funding_usd"
        ],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://invisibletech.ai/join-us"
    },
    {
      "fields": {
        "buyer": [
          "AI teams",
          "enterprises",
          "labs"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": "per-month",
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not currently published: labelbox.com/pricing returns 404; billing docs describe consumption-based Labelbox Units (1 LBU per data row for image/text Annotate, 500 free LBU/month) with no dollar rate"
        },
        "category": "annotation",
        "hq": "San Francisco, CA",
        "name": "Labelbox",
        "note": "Vendr (third party, not vendor) reports a median annual contract of about $44,583; excluded from buyer_price. Public pricing page appears withdrawn, a shift away from self-serve transparency.",
        "orgs": [
          "AI teams",
          "enterprises",
          "labs"
        ],
        "sells": "Labeling platform (Annotate, Catalog, Foundry) plus managed labeling services; contributors via Alignerr",
        "status": "operating",
        "url": "https://docs.labelbox.com/docs/billing",
        "verticals": [
          "LLM post-training",
          "computer vision",
          "general ML"
        ],
        "worker_pay": {
          "availability": "listed",
          "basis": "per-hour",
          "currency": "USD",
          "kind": "point",
          "max": 80,
          "min": 80,
          "note": "parsed from the quoted excerpt: $80/hr average pay",
          "raw": "see Alignerr row: '$80/hr average pay'"
        },
        "workforce": "hybrid"
      },
      "id": "vendor:labelbox",
      "name": "Labelbox",
      "notes": {
        "_org_roles": {
          "AI teams": [
            "buyer"
          ],
          "enterprises": [
            "buyer"
          ],
          "labs": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "docs.labelbox.com"
        ],
        "source_urls": [
          "https://docs.labelbox.com/docs/billing"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://docs.labelbox.com/docs/billing"
    },
    {
      "fields": {
        "buyer": "enterprises",
        "buyer_price": {
          "availability": "quote-only",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states the value was not disclosed",
          "raw": "quote only ('contact sales')"
        },
        "category": "annotation",
        "name": "LXT",
        "note": "Claims 10M+ contributors in 150+ countries. Site offers 'Surveys (via clickworker)', indicating clickworker is part of LXT. HQ not stated on pages read.",
        "orgs": [
          "enterprises"
        ],
        "sells": "Data collection, annotation, evaluation, transcription and GenAI (SFT, RLHF, red-teaming) services",
        "status": "operating",
        "url": "https://www.lxt.ai/",
        "verticals": [
          "speech/NLP",
          "computer vision",
          "generative AI",
          "physical AI",
          "search relevance"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "crowd"
      },
      "id": "vendor:lxt",
      "name": "LXT",
      "notes": {
        "_org_roles": {
          "enterprises": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "lxt.ai"
        ],
        "source_urls": [
          "https://www.lxt.ai/"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.lxt.ai/"
    },
    {
      "fields": {
        "buyer": "robotics and humanoid developers",
        "category": "robotics-data",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "$60M led by Framework Ventures (announced 2026-06-01); Sequoia-led round at about $500M valuation in talks per TechCrunch (2026-09-11), terms not final"
        },
        "name": "Mecka AI",
        "note": "Founded 2024. Projected a $100M annual run rate by end of 2026 (CEO to Fortune, June 2026, a projection). The Sequoia round is reported talks, not a closed round.",
        "orgs": [
          "robotics and humanoid developers"
        ],
        "sells": "Human motion data for humanoid and general robotics, recorded by paid contributors using body sensors and smartphones",
        "status": "operating",
        "url": "https://techcrunch.com/2026/09/11/mecka-ai-nears-500m-valuation-in-sequoia-led-deal-amid-rush-for-robot-training-data",
        "verticals": [
          "robotics",
          "humanoids"
        ],
        "workforce": "crowd"
      },
      "id": "vendor:mecka-ai",
      "name": "Mecka AI",
      "notes": {
        "_org_roles": {
          "robotics and humanoid developers": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-17",
        "last_seen": "2026-09-17",
        "method": "refinery-explore",
        "source_hosts": [
          "techcrunch.com"
        ],
        "source_urls": [
          "https://techcrunch.com/2026/09/11/mecka-ai-nears-500m-valuation-in-sequoia-led-deal-amid-rush-for-robot-training-data"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.727,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "prose-funding_usd",
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://techcrunch.com/2026/09/11/mecka-ai-nears-500m-valuation-in-sequoia-led-deal-amid-rush-for-robot-training-data"
    },
    {
      "fields": {
        "buyer": [
          "frontier labs",
          "enterprises"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "rlhf-expert",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "$350M Series C at a $10B valuation led by Felicis (TechCrunch); July 2026 talks for about $500M at $20B reported by TechCrunch, not confirmed closed"
        },
        "hq": "San Francisco, CA",
        "name": "Mercor",
        "note": "Revenue figures in press are not vendor-stated and are excluded. Other role pages list rates too but only the software-engineer page was verified.",
        "orgs": [
          "frontier labs",
          "enterprises"
        ],
        "sells": "Vetted expert marketplace supplying professionals for AI training and evaluation work",
        "status": "operating",
        "status_note": "Reported by workers or reporters, not the vendor. Sample: 3. Same article: 'He added that Mercor's average pay rate is $95 an hour' (speaker appears to be a Mercor representative, but attribution is ambiguous in the rendered text; verify before use). Reported $16/hr is about 1/6 of that average and 1/5 of the lowest advertised engineer rate. Rate cut corroborated by Forbes the same day (independen",
        "update_note": "updated 2026-09-17 from businessinsider.com: Same article: 'He added that Mercor's average pay rate is $95 an hour' (speaker appears to be a Mercor representative, but attribution is ambiguous in the rendered text; verify before use). Reported $16/hr is about 1/6 of that average and 1/5 of the lowest advertised engineer rate. Rate cut corrobor",
        "url": "https://www.mercor.com/experts/software-engineers",
        "verticals": [
          "software engineering",
          "consulting",
          "finance",
          "medicine",
          "law"
        ],
        "worker_pay": {
          "availability": "listed",
          "basis": "per-hour",
          "currency": "USD",
          "kind": "range",
          "max": 200,
          "min": 80,
          "note": "parsed from the quoted excerpt: get paid $80 to $200/hr to help train frontier AI systems",
          "raw": "vendor page: software engineers 'get paid $80 to $200/hr to help train frontier AI systems'. Mercor's 'AI trainer salary' article ($12 to $200+/hr) is a market estimate sourced from Glassdoor/ZipRecruiter/Indeed, NOT Mercor's own pay, and is excluded"
        },
        "worker_pay_reported": "$16 an hour (Nova), down from $21/hr (Musen); some contractors previously earned as much as $60 an hour on other projects",
        "worker_pay_reported_sample": 3,
        "worker_pay_reported_source": "news investigation (email seen + workers), 2025-11",
        "workforce": "expert"
      },
      "id": "vendor:mercor",
      "name": "Mercor",
      "notes": {
        "_org_roles": {
          "enterprises": [
            "buyer"
          ],
          "frontier labs": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "businessinsider.com",
          "glassdoor.com",
          "mercor.com",
          "siliconangle.com",
          "techcrunch.com"
        ],
        "source_urls": [
          "https://www.glassdoor.com/Salary/Mercor-Salaries-E9031572.htm",
          "https://techcrunch.com/2025/10/27/mercor-quintuples-valuation-to-10b-with-350m-series-c",
          "https://siliconangle.com/2026/07/09/mercor-buys-deeptune-build-training-environments-ai-agents",
          "https://www.mercor.com/experts/software-engineers",
          "https://www.businessinsider.com/mercor-cuts-contractors-meta-project-less-money-musen-nova-ai-2025-11"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 1,
        "confidence": "high",
        "corroborated": true,
        "flags": [
          "prose-funding_usd"
        ],
        "grade": "A",
        "hosts": 5,
        "independent_hosts": 5,
        "sources": 5,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.mercor.com/experts/software-engineers"
    },
    {
      "fields": {
        "buyer": "AI labs",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "rlhf-expert",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "$35M Series A at $500M valuation led by 01 Advisors, September 2025 (TechCrunch)"
        },
        "hq": "Palo Alto, CA",
        "name": "Micro1",
        "note": "Experts page mentions 'competitive compensation' and 'your hourly rate' but gives no figure.",
        "orgs": [
          "AI labs"
        ],
        "sells": "Domain-expert AI training data across 30+ languages, experts certified by an AI interviewer",
        "status": "operating",
        "status_note": "TechCrunch (2026-08-20) reports $500M GROSS run rate with net of $150M to $200M. The gross figure alone overstates the business; both are recorded.",
        "update_note": "updated 2026-09-17 from techcrunch.com: gross and net distinguished",
        "url": "https://www.micro1.ai/experts",
        "verticals": [
          "engineering",
          "finance",
          "healthcare",
          "legal",
          "linguistics"
        ],
        "workforce": "expert"
      },
      "id": "vendor:micro1",
      "name": "Micro1",
      "notes": {
        "_org_roles": {
          "AI labs": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "micro1.ai",
          "techcrunch.com"
        ],
        "source_urls": [
          "https://www.micro1.ai/experts",
          "https://techcrunch.com/2026/08/20/ai-data-startup-micro1-reaches-500m-gross-run-rate-amid-ai-training-boom"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.818,
        "confidence": "high",
        "corroborated": true,
        "flags": [
          "prose-funding_usd"
        ],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.micro1.ai/experts"
    },
    {
      "fields": {
        "buyer": "AI music companies",
        "buyer_price": {
          "availability": "listed",
          "basis": null,
          "currency": null,
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "mechanic published, no figure: 'priced per attribution event; costs scale with generation volume', with commit and overage terms"
        },
        "category": "broker",
        "name": "Musical AI",
        "note": "Legal entity SOMMS AI Inc. Rights partners named: Symphonic, APM Music, Soundstripe, Songtradr, Pro Sound Effects.",
        "orgs": [
          "AI music companies"
        ],
        "sells": "Consent and attribution layer for generative music, licensing rights-holder catalogs to AI music companies",
        "status": "operating",
        "url": "https://www.wearemusical.ai/",
        "verticals": [
          "music/audio"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "licensed-corpora"
      },
      "id": "vendor:musical-ai",
      "name": "Musical AI",
      "notes": {
        "_org_roles": {
          "AI music companies": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "wearemusical.ai"
        ],
        "source_urls": [
          "https://www.wearemusical.ai/"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "prose-buyer_price",
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.wearemusical.ai/"
    },
    {
      "fields": {
        "buyer": [
          "advertisers",
          "brands"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "broker",
        "hq": "New York, NY (2024 coverage)",
        "name": "Narrativ",
        "note": "status is 'unverified', outside the enum on purpose: narrativ.com did not resolve on 2026-09-16 (a strong sign of churn), but no shutdown notice was found. An unrelated retail-media 'Narrativ' rebranded to Howl.",
        "orgs": [
          "advertisers",
          "brands"
        ],
        "sells": "AI voice-replica marketplace for advertisers using licensed talent voices (as announced with SAG-AFTRA, 2024)",
        "status": "unverified",
        "url": "https://www.sagaftra.org/sag-aftra-and-narrativ-announce-new-agreement",
        "verticals": [
          "voice/likeness"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "no flat rate: per the SAG-AFTRA agreement, talent negotiate per project with a floor at the SAG-AFTRA commercials minimum"
        },
        "workforce": "licensed-corpora"
      },
      "id": "vendor:narrativ",
      "name": "Narrativ",
      "notes": {
        "_org_roles": {
          "advertisers": [
            "buyer"
          ],
          "brands": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "sagaftra.org"
        ],
        "source_urls": [
          "https://www.sagaftra.org/sag-aftra-and-narrativ-announce-new-agreement"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.sagaftra.org/sag-aftra-and-narrativ-announce-new-agreement"
    },
    {
      "fields": {
        "buyer": "enterprises",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "robotics-data",
        "hq": "Arizona, USA (delivery centers in Tamil Nadu, India)",
        "name": "Objectways",
        "note": "Claims 2,200+ trained annotators, managed not crowd.",
        "orgs": [
          "enterprises"
        ],
        "sells": "Annotation (text, image, audio, video, LiDAR), data collection, GenAI evaluation and physical-AI teleoperation/egocentric data",
        "status": "operating",
        "url": "https://www.objectways.com",
        "verticals": [
          "autonomous vehicles",
          "healthcare",
          "media",
          "financial services",
          "robotics/physical AI"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "hybrid"
      },
      "id": "vendor:objectways",
      "name": "Objectways",
      "notes": {
        "_org_roles": {
          "enterprises": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "objectways.com"
        ],
        "source_urls": [
          "https://www.objectways.com"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.objectways.com"
    },
    {
      "fields": {
        "buyer": [
          "AI labs",
          "labeling companies"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "broker",
        "name": "OpenTrain AI",
        "note": "A robot-teleoperator listing at $30-55/hr surfaced in search but the job page returned 410, so excluded.",
        "orgs": [
          "AI labs",
          "labeling companies"
        ],
        "sells": "Marketplace connecting AI companies with vetted experts for evaluation, RLHF, red-teaming and labeling",
        "status": "operating",
        "url": "https://www.opentrain.ai",
        "verticals": [
          "LLM evaluation",
          "RLHF",
          "medical",
          "multilingual",
          "robot teleoperation (unverified listing)"
        ],
        "worker_pay": {
          "availability": "listed",
          "basis": "per-hour",
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "published example role rates: Agentic AI Evaluation Lead $65-95/hr, Medical RLHF Reviewer $55-80/hr, Math Reasoning Specialist $50-75/hr, Code Generation Evaluator $60-90/hr, Safety Red Team Analyst $70-100/hr, Multilingual AI Trainer $30-50/hr"
        },
        "workforce": "expert"
      },
      "id": "vendor:opentrain-ai",
      "name": "OpenTrain AI",
      "notes": {
        "_org_roles": {
          "AI labs": [
            "buyer"
          ],
          "labeling companies": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-corroborate",
        "source_hosts": [
          "opentrain.ai",
          "robottraining.jobs"
        ],
        "source_urls": [
          "https://www.opentrain.ai",
          "https://robottraining.jobs/"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "prose-worker_pay",
          "single-source"
        ],
        "grade": "B",
        "hosts": 2,
        "independent_hosts": 1,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.opentrain.ai"
    },
    {
      "fields": {
        "buyer": "Scale AI's lab customers",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "rlhf-expert",
        "name": "Outlier (Scale AI)",
        "note": "Page claims 900K+ contributors. Third-party ranges ($15 to $100/hr) exist on aggregator sites but are not vendor-published and are excluded.",
        "orgs": [
          "Scale AI's lab customers"
        ],
        "sells": "Contributor platform for prompt writing, rubric grading and output ranking for LLM training",
        "status": "operating",
        "status_note": "Reported by workers or reporters, not the vendor. Sample: 118. Vendor publishes no rate, so the reported figure is the only value; about $29-54/hr implied by the annual band, $37/hr on the Trainer page.",
        "update_note": "updated 2026-09-17 from glassdoor.com: Vendor publishes no rate, so the reported figure is the only value; about $29-54/hr implied by the annual band, $37/hr on the Trainer page.",
        "url": "https://outlier.ai",
        "verticals": [
          "coding",
          "STEM",
          "languages"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "no rate card; site states only 'competitive pay + quality-based rewards, paid weekly' and an aggregate '$500M+ paid to experts'"
        },
        "worker_pay_reported": "$60,499 (25th percentile) - $111,669 (75th percentile) annually, 'based on 118 salaries submitted'; Trainer title: '$37 per hour' (13 salaries submitted)",
        "worker_pay_reported_sample": 118,
        "worker_pay_reported_source": "glassdoor, to 2026-09 (118 salaries); 2025-04 snapshot for Trainer (13)",
        "workforce": "crowd"
      },
      "id": "vendor:outlier-scale-ai",
      "name": "Outlier (Scale AI)",
      "notes": {
        "_org_roles": {
          "Scale AI's lab customers": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "glassdoor.com",
          "outlier.ai"
        ],
        "source_urls": [
          "https://outlier.ai",
          "https://www.glassdoor.com/Salary/Outlier-AI-AI-Trainer-Salaries-E2858115_D_KO11,21.htm?payPeriod=HOURLY"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 1,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://outlier.ai"
    },
    {
      "fields": {
        "buyer": "AI labs",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "rlhf-expert",
        "name": "Pareto AI",
        "note": "Experts page lists recruiting fields but no pay figures; home page gives no HQ or product detail.",
        "orgs": [
          "AI labs"
        ],
        "sells": "Human training signal for AI labs from a recruited expert network",
        "status": "operating",
        "url": "https://pareto.ai/experts",
        "verticals": [
          "medicine",
          "law",
          "STEM",
          "finance",
          "software/ML",
          "humanities"
        ],
        "workforce": "expert"
      },
      "id": "vendor:pareto-ai",
      "name": "Pareto AI",
      "notes": {
        "_org_roles": {
          "AI labs": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "pareto.ai"
        ],
        "source_urls": [
          "https://pareto.ai/experts"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.818,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://pareto.ai/experts"
    },
    {
      "fields": {
        "buyer": [
          "academic and corporate researchers",
          "AI companies"
        ],
        "buyer_price": {
          "availability": "listed",
          "basis": null,
          "currency": null,
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "platform fee on top of participant reward: 42.8% corporate, 33.3% academic/non-profit; pay-as-you-go, no subscription"
        },
        "category": "crowd-platform",
        "name": "Prolific",
        "note": "The ONLY vendor found publishing both sides exactly. Worker floor from researcher-help.prolific.com payment principles. At the $12/hr recommended rate a corporate buyer pays about $17.14/hr, so the platform take is about $5.14/hr. HQ commonly reported as London, not verified.",
        "orgs": [
          "academic and corporate researchers",
          "AI companies"
        ],
        "sells": "Participant recruitment for research studies and AI human-feedback tasks",
        "status": "operating",
        "status_note": "Reported by workers or reporters, not the vendor. Sample: 450. In 2017 Prolific asked researchers for at least US$6.50/hr; surveyed workers' median was $4.55 on paid time and $3.56 including unpaid search time (about 70% and 55% of the floor). Rates cover time on the study, not time spent finding studies. Current floor is higher; no recent survey found. Other reports: Fairwork Cloudwork score 5/10",
        "update_note": "updated 2026-09-17 from ilo.org: In 2017 Prolific asked researchers for at least US$6.50/hr; surveyed workers' median was $4.55 on paid time and $3.56 including unpaid search time (about 70% and 55% of the floor). Rates cover time on the study, not time spent finding studies. Current floor is higher; no recent survey found.",
        "url": "https://www.prolific.com/pricing",
        "verticals": [
          "academic research",
          "UX research",
          "RLHF/AI feedback",
          "surveys"
        ],
        "worker_pay": {
          "availability": "listed",
          "basis": "per-hour",
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "minimum '£6.00 / $8.00 per hour'; recommended 'at least £9.00 / $12.00 per hour'; underpaying studies are blocked"
        },
        "worker_pay_reported": "median US$4.55/hr paid work (mean 5.45, 450 observations); median US$3.56/hr paid and unpaid (mean 4.26, 446 observations)",
        "worker_pay_reported_sample": 450,
        "worker_pay_reported_source": "academic/ILO survey, 2017 survey",
        "workforce": "crowd"
      },
      "id": "vendor:prolific",
      "name": "Prolific",
      "notes": {
        "_org_roles": {
          "AI companies": [
            "buyer"
          ],
          "academic and corporate researchers": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "ilo.org",
          "prolific.com"
        ],
        "source_urls": [
          "https://www.ilo.org/sites/default/files/wcmsp5/groups/public/@dgreports/@dcomm/@publ/documents/publication/wcms_645337.pdf",
          "https://www.prolific.com/pricing"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 1,
        "confidence": "high",
        "corroborated": true,
        "flags": [
          "prose-buyer_price",
          "prose-worker_pay"
        ],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.prolific.com/pricing"
    },
    {
      "fields": {
        "buyer": "AI model builders",
        "buyer_price": {
          "availability": "quote-only",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states the value was not disclosed",
          "raw": "quote only"
        },
        "category": "licensed-corpora",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "$25M Series A (Aug 2025) plus $30M extension (Jan 2026) led by a16z, per Protege press materials"
        },
        "hq": "New York, NY",
        "name": "Protege",
        "note": "Acquired Calliope Networks in 2026. Member of the Dataset Providers Alliance.",
        "orgs": [
          "AI model builders"
        ],
        "sells": "Licensed real-world data (healthcare, video, audio, motion capture, spatial) under non-perpetual training-window licenses",
        "status": "operating",
        "url": "https://withprotege.ai/data-providers",
        "verticals": [
          "healthcare",
          "video",
          "audio",
          "motion capture",
          "spatial",
          "text"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": null,
          "raw": "revenue-share with data vendors stated, percentage not published"
        },
        "workforce": "licensed-corpora"
      },
      "id": "vendor:protege",
      "name": "Protege",
      "notes": {
        "_org_roles": {
          "AI model builders": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "withprotege.ai"
        ],
        "source_urls": [
          "https://withprotege.ai/data-providers"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "prose-funding_usd",
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://withprotege.ai/data-providers"
    },
    {
      "fields": {
        "buyer": "Scale AI's customers",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "crowd-platform",
        "name": "Remotasks (Scale AI)",
        "note": "Still live and accepting signups, but Scale consolidated contractor work under Outlier in 2024 and Remotasks is geo-restricted in some countries. Largely superseded, not shut down.",
        "orgs": [
          "Scale AI's customers"
        ],
        "sells": "Crowd task platform for image labeling, transcription and annotation",
        "status": "operating",
        "status_note": "Reported by workers or reporters, not the vendor. Sample: 70. No dollar figure, but an independent finding that minimum-wage pay could not be evidenced. Scale's published reply calls the evaluation 'overly subjective'. Other reports: often earn far below the minimum wage, which in the Philippines ranges from $6 to $10 a da (news investigation (interviews + payment records), 2021 to 2023-08) | aver",
        "update_note": "updated 2026-09-17 from fairwork.oii.ox.ac.uk: No dollar figure, but an independent finding that minimum-wage pay could not be evidenced. Scale's published reply calls the evaluation 'overly subjective'.",
        "url": "https://www.remotasks.com",
        "verticals": [
          "image/video annotation",
          "audio transcription"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "no rate published; pay 'based on the quality and number of tasks', weekly via PayPal or AirTM; site claims '$15M total earnings' across '240,000+ taskers' (aggregate, not a rate)"
        },
        "worker_pay_reported": "Fairwork Cloudwork score 2/10; no Fair Pay points (threshold 1.1 paid on time for all work, and 1.2 paid at least local minimum wage, not evidenced)",
        "worker_pay_reported_sample": 70,
        "worker_pay_reported_source": "fairwork, 2024-02 to 2025-01 (survey window)",
        "workforce": "crowd"
      },
      "id": "vendor:remotasks-scale-ai",
      "name": "Remotasks (Scale AI)",
      "notes": {
        "_org_roles": {
          "Scale AI's customers": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "fairwork.oii.ox.ac.uk",
          "remotasks.com"
        ],
        "source_urls": [
          "https://www.remotasks.com",
          "https://fairwork.oii.ox.ac.uk/wp-content/uploads/sites/11/2025/05/Fairwork-Cloudwork-Report-2025-FINAL.pdf"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 1,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.remotasks.com"
    },
    {
      "fields": {
        "buyer": "AI and music-tech companies",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "licensed-corpora",
        "hq": "Pasadena, CA",
        "name": "Rightsify / Global Copyright Exchange (GCX)",
        "note": "Work-for-hire model means creators get a one-time fee, no ongoing share. Founding DPA member.",
        "orgs": [
          "AI and music-tech companies"
        ],
        "sells": "Copyright-cleared music datasets (4.4M+ hours, stems, MIDI, metadata)",
        "status": "operating",
        "url": "https://www.rightsify.com/gcx",
        "verticals": [
          "music/audio"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not a royalty: content acquired 'as a work for hire, with fair compensation provided to creators', Rightsify 'owns 100% of the copyrights'; no rate disclosed"
        },
        "workforce": "licensed-corpora"
      },
      "id": "vendor:rightsify-global-copyright-exchange-gcx",
      "name": "Rightsify / Global Copyright Exchange (GCX)",
      "notes": {
        "_org_roles": {
          "AI and music-tech companies": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "rightsify.com"
        ],
        "source_urls": [
          "https://www.rightsify.com/gcx"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.rightsify.com/gcx"
    },
    {
      "fields": {
        "buyer": [
          "physical AI and robotics companies",
          "mostly North American"
        ],
        "category": "robotics-data",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "point",
          "max": 22000000,
          "min": 22000000,
          "note": null,
          "raw": "$22M pre-Series A (2026-07-23), $30M total including seed"
        },
        "hq": "Singapore; Mountain View, CA",
        "name": "Ropedia",
        "note": "Founded 2025. Generates its own data rather than annotating client-owned data; positions against teleoperation capture. The Robot Report headline says $22M; a Business Wire release headline reads $30M, the cumulative figure.",
        "orgs": [
          "physical AI and robotics companies",
          "mostly North American"
        ],
        "sells": "Egocentric human movement and object-interaction data captured with its HOMIE head-mounted four-camera rig, processed and annotated end to end",
        "status": "operating",
        "url": "https://www.therobotreport.com/ropedia-raises-22m-scale-data-collection-training-robots",
        "verticals": [
          "robotics",
          "physical AI"
        ]
      },
      "id": "vendor:ropedia",
      "name": "Ropedia",
      "notes": {
        "_org_roles": {
          "mostly North American": [
            "buyer"
          ],
          "physical AI and robotics companies": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-17",
        "last_seen": "2026-09-17",
        "method": "refinery-explore",
        "source_hosts": [
          "therobotreport.com"
        ],
        "source_urls": [
          "https://www.therobotreport.com/ropedia-raises-22m-scale-data-collection-training-robots"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.636,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.therobotreport.com/ropedia-raises-22m-scale-data-collection-training-robots"
    },
    {
      "fields": {
        "buyer": [
          "enterprise AI",
          "ML teams"
        ],
        "buyer_price": {
          "availability": "quote-only",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states the value was not disclosed",
          "raw": "quote only"
        },
        "category": "annotation",
        "name": "Sama",
        "note": "Exited content moderation ('last moderation contract officially ended on March 31, 2023'), now CV annotation only. B-Corp. Employed workforce in East Africa, not open crowd; 'hybrid' used for managed-employed. HQ commonly reported as San Francisco, not verified.",
        "orgs": [
          "enterprise AI",
          "ML teams"
        ],
        "sells": "Managed computer-vision annotation workforce (image, video, 3D point cloud)",
        "status": "pivoted",
        "status_note": "Reported by workers or reporters, not the vendor. Sample: 4. Client (OpenAI) paid Sama $12.50/hr, six to nine times take-home. Sama replied workers could earn $1.46-3.74/hr after tax. Reported take-home is below the $394/month family living cost Sama itself cites.",
        "update_note": "updated 2026-09-17 from time.com: Client (OpenAI) paid Sama $12.50/hr, six to nine times take-home. Sama replied workers could earn $1.46-3.74/hr after tax. Reported take-home is below the $394/month family living cost Sama itself cites.",
        "url": "https://www.sama.com/blog/building-an-ethical-supply-chain",
        "verticals": [
          "computer vision",
          "autonomous vehicles",
          "generative AI",
          "agriculture"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": "per-month",
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "no current wage rate published; living-wage commitment and impact stats ('average of 3.6x increase in their earnings when joining'); cites 2021 Nairobi minimum wage Ksh 13,572 (~$119) vs family living cost Ksh 39,622 (~$394)/month"
        },
        "worker_pay_reported": "take-home wage of between around $1.32 and $2 per hour depending on seniority and performance",
        "worker_pay_reported_sample": 4,
        "worker_pay_reported_source": "news investigation (payslips + interviews), 2021-11 to 2022-02",
        "workforce": "hybrid"
      },
      "id": "vendor:sama",
      "name": "Sama",
      "notes": {
        "_org_roles": {
          "ML teams": [
            "buyer"
          ],
          "enterprise AI": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "sama.com",
          "time.com"
        ],
        "source_urls": [
          "https://time.com/6247678/openai-chatgpt-kenya-workers",
          "https://www.sama.com/blog/building-an-ethical-supply-chain"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 1,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.sama.com/blog/building-an-ethical-supply-chain"
    },
    {
      "fields": {
        "buyer": [
          "frontier labs",
          "AV companies",
          "enterprises",
          "government"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states the value was not disclosed",
          "raw": "quote only for enterprise; self-serve Data Engine is pay-as-you-go with a free first 1,000 labeling units, per-unit price not published"
        },
        "category": "annotation",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "Meta invested roughly $14.3B for a 49% non-voting stake at a $29B valuation, June 2025 (TechCrunch, Axios); earlier $1B Series F at $13.8B, May 2024 (scale.com blog)"
        },
        "hq": "San Francisco, CA",
        "name": "Scale AI",
        "note": "Not an acquisition: Scale remains independent after the Meta stake; founder Alexandr Wang left for Meta. Worker pay is published by neither Scale nor Outlier (see Outlier row).",
        "orgs": [
          "frontier labs",
          "AV companies",
          "enterprises",
          "government"
        ],
        "sells": "Managed data labeling, RLHF and evaluation data, GenAI platform; contributor work runs through its Outlier and Remotasks brands",
        "status": "operating",
        "url": "https://scale.com/pricing",
        "verticals": [
          "LLM/RLHF",
          "autonomous vehicles",
          "computer vision",
          "government"
        ],
        "workforce": "hybrid"
      },
      "id": "vendor:scale-ai",
      "name": "Scale AI",
      "notes": {
        "_org_roles": {
          "AV companies": [
            "buyer"
          ],
          "enterprises": [
            "buyer"
          ],
          "frontier labs": [
            "buyer"
          ],
          "government": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-corroborate",
        "source_hosts": [
          "scale.com",
          "techcrunch.com"
        ],
        "source_urls": [
          "https://techcrunch.com/2025/06/13/scale-ai-confirms-significant-investment-from-meta-says-ceo-alexandr-wang-is-leaving",
          "https://scale.com/pricing"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.818,
        "confidence": "high",
        "corroborated": true,
        "flags": [
          "prose-funding_usd"
        ],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://scale.com/pricing"
    },
    {
      "fields": {
        "buyer": [
          "enterprises",
          "frontier labs"
        ],
        "buyer_price": {
          "availability": "quote-only",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states the value was not disclosed",
          "raw": "quote only (no pricing page exists)"
        },
        "category": "rlhf-expert",
        "hq": "Redwood City, CA",
        "name": "Snorkel AI",
        "note": "Investors named on site without amounts. Third-party estimates of $50K-60K/yr entry contracts are not vendor-published and excluded.",
        "orgs": [
          "enterprises",
          "frontier labs"
        ],
        "sells": "Snorkel Flow programmatic data-development platform, Snorkel Evaluate, and expert data-as-a-service for labs",
        "status": "operating",
        "url": "https://snorkel.ai/about",
        "verticals": [
          "frontier AI",
          "enterprise ML",
          "agentic AI"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "expert"
      },
      "id": "vendor:snorkel-ai",
      "name": "Snorkel AI",
      "notes": {
        "_org_roles": {
          "enterprises": [
            "buyer"
          ],
          "frontier labs": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "snorkel.ai"
        ],
        "source_urls": [
          "https://snorkel.ai/about"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://snorkel.ai/about"
    },
    {
      "fields": {
        "buyer": [
          "enterprises",
          "AI teams"
        ],
        "buyer_price": {
          "availability": "quote-only",
          "basis": null,
          "currency": null,
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "tiers published without dollar prices: Starter (1K compute hours, self-serve), Pro (2.5K compute hours, request demo), Enterprise (10K compute hours, contact sales)"
        },
        "category": "annotation",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "vendor about page: Seed $3M (2020), Series A $14.5M (2021), Series B $50M (Nov 2024) plus $13.5M from Dell Technologies Capital (Jun 2025)"
        },
        "hq": "Silicon Valley, CA (city not stated)",
        "name": "SuperAnnotate",
        "note": "Pricing page structure exists but every tier is effectively quote-gated.",
        "orgs": [
          "enterprises",
          "AI teams"
        ],
        "sells": "Multimodal annotation editor, data curation, RLHF/eval workflows and a managed expert network",
        "status": "operating",
        "url": "https://www.superannotate.com/pricing",
        "verticals": [
          "LLM fine-tuning",
          "computer vision",
          "enterprise AI"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "hybrid"
      },
      "id": "vendor:superannotate",
      "name": "SuperAnnotate",
      "notes": {
        "_org_roles": {
          "AI teams": [
            "buyer"
          ],
          "enterprises": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "superannotate.com"
        ],
        "source_urls": [
          "https://www.superannotate.com/pricing"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "prose-buyer_price",
          "prose-funding_usd",
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.superannotate.com/pricing"
    },
    {
      "fields": {
        "acquirer": "Warner Music Group",
        "buyer": [
          "AI platforms",
          "collecting societies"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "broker",
        "name": "Sureel AI",
        "note": "Acquisition by Warner Music Group reported about June 2026 by TechCrunch; not mentioned on sureel.ai itself, so single-press-source.",
        "orgs": [
          "AI platforms",
          "collecting societies"
        ],
        "sells": "Attribution/provenance tech tracing copyrighted works in AI training and outputs, enabling attribution-based licensing",
        "status": "acquired",
        "status_note": "Confirmed: Warner Music acquired Sureel AI (announced 2026-06-10, terms undisclosed), carried by TechCrunch and more than 20 other named outlets plus WMG's own release. Sureel continues to operate as an independent platform. Later coverage refers to it as WMG's Sureel, which corroborates closing.",
        "update_note": "updated 2026-09-17 from techcrunch.com: previously unverifiable acquisition now confirmed",
        "url": "https://www.sureel.ai/",
        "verticals": [
          "music/audio"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "licensed-corpora"
      },
      "id": "vendor:sureel-ai",
      "name": "Sureel AI",
      "notes": {
        "_org_roles": {
          "AI platforms": [
            "buyer"
          ],
          "collecting societies": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "sureel.ai",
          "techcrunch.com"
        ],
        "source_urls": [
          "https://techcrunch.com/2026/06/10/warner-music-acquires-ai-attribution-startup-sureel-ai",
          "https://www.sureel.ai/"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.909,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.sureel.ai/"
    },
    {
      "fields": {
        "buyer": "frontier labs",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "rlhf-expert",
        "conflict_note": "independent source inc.com disagrees: The record says Surge had 'a first outside raise reported July 2025'. That is wrong. The July 2025 reports described a SOLICITATION and TALKS, not a closed round. Reuters, 2025-07-01: 'EXCLUSIVE: Scale AI's bigger rival Surge AI SEEKS up to $1 billion capital raise, sources say'. Bloomberg, 2025-07-30: 'Scale Rival Surge AI IN TALKS for Funding at $25 Billion Value'. SiliconANGLE, 2025-07-01: 'Data labeling startup Surge AI REPORTEDLY SEEKING $1B in first ca",
        "hq": "San Francisco, CA",
        "name": "Surge AI",
        "note": "Bootstrapped until a first outside raise reported in July 2025; raise size not confirmed from a vendor source, so null. Press revenue estimates deliberately excluded.",
        "orgs": [
          "frontier labs"
        ],
        "sells": "Expert-sourced RLHF and evaluation data for frontier labs",
        "status": "operating",
        "status_note": "Reported by workers or reporters, not the vendor. Sample: not stated. Only advertised figure is a $200-400/hr specialist listing, 7x to 27x the $15-30/hr workers reported. Not the same role; the gap shows advertised rates are drawn from the top of the distribution.",
        "update_note": "updated 2026-09-17 from theverge.com: Only advertised figure is a $200-400/hr specialist listing, 7x to 27x the $15-30/hr workers reported. Not the same role; the gap shows advertised rates are drawn from the top of the distribution.",
        "url": "https://surgehq.ai/workforce",
        "verticals": [
          "writing/journalism",
          "law",
          "finance",
          "consulting",
          "mathematics",
          "physics",
          "computer science",
          "philosophy"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": "per-hour",
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "no rate card; one published listing on its workforce page: 'Award-Winning Author / Journalist' at $200 to $400/hour (contractor)"
        },
        "worker_pay_reported": "The workers I spoke with earned between $15 and $30 per hour",
        "worker_pay_reported_source": "news investigation (interviews), 2023-06",
        "workforce": "expert"
      },
      "id": "vendor:surge-ai",
      "name": "Surge AI",
      "notes": {
        "_org_roles": {
          "frontier labs": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "inc.com",
          "surgehq.ai",
          "theverge.com"
        ],
        "source_urls": [
          "https://www.theverge.com/features/23764584/ai-artificial-intelligence-data-notation-labor-scale-surge-remotasks-openai-chatbots",
          "https://www.inc.com/sam-blum/bootstrapped-to-1-billion-surge-ai-ceo-edwin-chen-on-how-he-did-it/91207937",
          "https://surgehq.ai/workforce"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 1,
        "confidence": "high",
        "corroborated": true,
        "flags": [
          "conflicted"
        ],
        "grade": "B",
        "hosts": 3,
        "independent_hosts": 3,
        "sources": 3,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://surgehq.ai/workforce"
    },
    {
      "fields": {
        "buyer": [
          "tech companies",
          "automotive",
          "search",
          "ads"
        ],
        "buyer_price": {
          "availability": "quote-only",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states the value was not disclosed",
          "raw": "quote only ('contact us')"
        },
        "category": "annotation",
        "name": "TELUS Digital",
        "note": "Formerly TELUS International; its AI data arm is the former Lionbridge AI, acquired for about US$935M, closed Dec 31, 2020 (TELUS press release). Page claims '1M+' AI community and 'two billion labels annually'.",
        "orgs": [
          "tech companies",
          "automotive",
          "search",
          "ads"
        ],
        "sells": "AI data services: post-training/RLHF data, automotive sensor-fusion labeling, physical-AI egocentric collections, search/ads rating, off-the-shelf datasets",
        "status": "operating",
        "status_note": "Reported by workers or reporters, not the vendor. Sample: 17. Registry says not published, but TELUS's own US Raters posting lists 'Salary range: $11.00 - $40.00' (https://jobs.telusdigital.com/jobs/17318859-us-raters). Glassdoor's annual band ($44-75/hr equivalent) is implausible for part-time raters and exceeds the posted ceiling: modeled, not reported.",
        "update_note": "updated 2026-09-17 from glassdoor.com: Registry says not published, but TELUS's own US Raters posting lists 'Salary range: $11.00 - $40.00' (https://jobs.telusdigital.com/jobs/17318859-us-raters). Glassdoor's annual band ($44-75/hr equivalent) is implausible for part-time raters and exceeds the posted ceiling: modeled, not reported.",
        "url": "https://www.telusdigital.com/solutions/data-and-ai-solutions",
        "verticals": [
          "GenAI/AGI",
          "automotive",
          "physical AI/robotics",
          "search & ads"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "worker_pay_reported": "'based on 17 salaries submitted by TELUS Digital AI rater professionals' (band $92,030-$155,689/yr); AI Search Rater 'estimated total pay ... $34 per hour ... estimated base pay is $27 per hour'",
        "worker_pay_reported_sample": 17,
        "worker_pay_reported_source": "glassdoor, to 2026-09",
        "workforce": "crowd"
      },
      "id": "vendor:telus-digital",
      "name": "TELUS Digital",
      "notes": {
        "_org_roles": {
          "ads": [
            "buyer"
          ],
          "automotive": [
            "buyer"
          ],
          "search": [
            "buyer"
          ],
          "tech companies": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-update",
        "source_hosts": [
          "glassdoor.com",
          "telusdigital.com"
        ],
        "source_urls": [
          "https://www.telusdigital.com/solutions/data-and-ai-solutions",
          "https://www.glassdoor.com/Salary/TELUS-Digital-AI-rater-Salaries-E2841163_D_KO20,28.htm?payPeriod=HOURLY"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 1,
        "confidence": "high",
        "corroborated": true,
        "flags": [],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.telusdigital.com/solutions/data-and-ai-solutions"
    },
    {
      "fields": {
        "buyer": [
          "tech companies",
          "AI labs"
        ],
        "buyer_price": {
          "availability": "quote-only",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states the value was not disclosed",
          "raw": "quote only (toloka.ai/pricing returns 404)"
        },
        "category": "crowd-platform",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "$72M round led by Bezos Expeditions, May 2025 (press); Nebius Group retains majority economic stake"
        },
        "hq": "Amsterdam, Netherlands",
        "name": "Toloka",
        "note": "Originated inside Yandex, moved to Nebius Group after the Yandex split; Russian division divested July 2024. Worker pay verified on https://mindrift.ai/.",
        "orgs": [
          "tech companies",
          "AI labs"
        ],
        "sells": "Human data for SFT, RLHF preference, evaluation and red-teaming from crowd plus vetted experts; worker side runs through Mindrift",
        "status": "operating",
        "url": "https://toloka.ai/about",
        "verticals": [
          "coding",
          "AI safety",
          "reasoning",
          "multimodal generation",
          "robotics"
        ],
        "worker_pay": {
          "availability": "listed",
          "basis": "per-hour",
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "via Mindrift, 'built by Toloka': 'Entry-level: $15-30 per hour', 'Domain experts: $60-100+/hr'"
        },
        "workforce": "hybrid"
      },
      "id": "vendor:toloka",
      "name": "Toloka",
      "notes": {
        "_org_roles": {
          "AI labs": [
            "buyer"
          ],
          "tech companies": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "public",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-17",
        "method": "refinery-corroborate",
        "source_hosts": [
          "siliconangle.com",
          "toloka.ai"
        ],
        "source_urls": [
          "https://toloka.ai/about",
          "https://siliconangle.com/2025/05/07/ai-data-provider-toloka-raises-72m-funding"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.909,
        "confidence": "high",
        "corroborated": true,
        "flags": [
          "prose-funding_usd",
          "prose-worker_pay"
        ],
        "grade": "A",
        "hosts": 2,
        "independent_hosts": 2,
        "sources": 2,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://toloka.ai/about"
    },
    {
      "fields": {
        "buyer": "AI labs",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "licensed-corpora",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "point",
          "max": 4500000,
          "min": 4500000,
          "note": null,
          "raw": "$4.5M seed (per Troveo's own site citing press)"
        },
        "hq": "Austin, TX",
        "name": "Troveo AI",
        "note": "Expected rate card not found; page directs to 'Contact us'. Also publishes a vendor directory (troveo.ai/resources/robotics-training-data-companies), marketing content without pricing.",
        "orgs": [
          "AI labs"
        ],
        "sells": "Rights-cleared video (plus audio, gaming, robotics) with annotations, sourced from a creator network",
        "status": "operating",
        "url": "https://www.troveo.ai/content-owners",
        "verticals": [
          "video",
          "audio",
          "gaming",
          "robotics"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "no per-minute rate on its own content-owners page; states aggregate '$50M+ paid to creators globally'. Trade press cites an industry range of about $1-4/minute, not confirmed on troveo.ai"
        },
        "workforce": "licensed-corpora"
      },
      "id": "vendor:troveo-ai",
      "name": "Troveo AI",
      "notes": {
        "_org_roles": {
          "AI labs": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "troveo.ai"
        ],
        "source_urls": [
          "https://www.troveo.ai/content-owners"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.troveo.ai/content-owners"
    },
    {
      "fields": {
        "buyer": "robotics and VLA model builders",
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published on its own service pages; its blog self-estimates '$25,000-$60,000' for a 5,000-episode program and '$750-$2,500' for a 200-500 episode pilot"
        },
        "category": "robotics-data",
        "name": "Truelabel",
        "note": "CONFLICT OF INTEREST: this blog ranks Truelabel cheapest and publishes estimated prices for competitors (Scale AI $200K-300K minimum, Encord $80K-120K, Appen $50K-90K, Labelbox $60K-100K per 5,000 episodes). Those are a competitor's claims, never to be recorded as the named vendors' prices.",
        "orgs": [
          "robotics and VLA model builders"
        ],
        "sells": "Broker/marketplace for physical-AI data (egocentric, exocentric, teleoperation) aggregated from '100+ vendors across 100+ countries'",
        "status": "operating",
        "url": "https://truelabel.ai/blog/best-teleoperation-data-providers-2026",
        "verticals": [
          "teleoperation",
          "egocentric video",
          "VLA training",
          "robotics evaluation"
        ],
        "worker_pay": {
          "availability": "listed",
          "basis": "per-hour",
          "currency": "USD",
          "kind": "range",
          "max": 60,
          "min": 25,
          "note": "parsed from the quoted excerpt: typically $25-$60 per hour fully loaded",
          "raw": "blog claims operators 'typically $25-$60 per hour fully loaded' (its estimate, not a payroll disclosure)"
        },
        "workforce": "hybrid"
      },
      "id": "vendor:truelabel",
      "name": "Truelabel",
      "notes": {
        "_org_roles": {
          "robotics and VLA model builders": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "truelabel.ai"
        ],
        "source_urls": [
          "https://truelabel.ai/blog/best-teleoperation-data-providers-2026"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://truelabel.ai/blog/best-teleoperation-data-providers-2026"
    },
    {
      "fields": {
        "buyer": [
          "frontier labs",
          "enterprises"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "rlhf-expert",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "about $247M total, latest Series E $111M at $2.2B, March 2025, per Crunchbase-style aggregators (not a Turing page)"
        },
        "hq": "Palo Alto, CA",
        "name": "Turing",
        "note": "Began as developer staffing, now also a human-data vendor. Pay figures unusable as rates until the unit is known.",
        "orgs": [
          "frontier labs",
          "enterprises"
        ],
        "sells": "LLM post-training data (SFT, RLHF, DPO) plus a vetted remote tech-talent marketplace",
        "status": "operating",
        "url": "https://www.turing.com/talent",
        "verticals": [
          "healthcare",
          "physics/chemistry/math",
          "software engineering",
          "business operations"
        ],
        "worker_pay": {
          "availability": null,
          "basis": "per-hour",
          "currency": null,
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "talent page lists dollar amounts per role with NO unit shown (e.g. '$300' clinical reviewers, '$250' scientific coding, '$1,000' Senior Software Engineer LLM Evaluation); hourly vs per-project unknown"
        },
        "workforce": "expert"
      },
      "id": "vendor:turing",
      "name": "Turing",
      "notes": {
        "_org_roles": {
          "enterprises": [
            "buyer"
          ],
          "frontier labs": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "turing.com"
        ],
        "source_urls": [
          "https://www.turing.com/talent"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "prose-funding_usd",
          "prose-worker_pay",
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.turing.com/talent"
    },
    {
      "fields": {
        "buyer": "AI labs and enterprises",
        "category": "annotation",
        "hq": "San Francisco, CA",
        "name": "Uber AI Solutions",
        "note": "Division of Uber, launched late 2024 as Scaled Solutions. Bloomberg (2026-07-01) reported the dismissal of two top leaders of the data-labeling business; Uber called it a leadership transition. Detail read from TechGolly's summary, not Bloomberg directly.",
        "orgs": [
          "AI labs and enterprises"
        ],
        "sells": "Data labeling, model evaluation and localisation using Uber's gig network, with digital tasks inside the Driver app",
        "status": "operating",
        "url": "https://www.uber.com/us/en/ai-solutions/annotation",
        "verticals": [
          "annotation",
          "evaluation",
          "localization"
        ],
        "workforce": "crowd"
      },
      "id": "vendor:uber-ai-solutions",
      "name": "Uber AI Solutions",
      "notes": {
        "_org_roles": {
          "AI labs and enterprises": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-17",
        "last_seen": "2026-09-17",
        "method": "refinery-explore",
        "source_hosts": [
          "uber.com"
        ],
        "source_urls": [
          "https://www.uber.com/us/en/ai-solutions/annotation"
        ]
      },
      "quality": {
        "age_days": 1,
        "completeness": 0.727,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://www.uber.com/us/en/ai-solutions/annotation"
    },
    {
      "fields": {
        "buyer": "enterprise AI teams",
        "buyer_price": {
          "availability": "quote-only",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states the value was not disclosed",
          "raw": "quote only"
        },
        "category": "licensed-corpora",
        "name": "vAIsual",
        "note": "Founding DPA member.",
        "orgs": [
          "enterprise AI teams"
        ],
        "sells": "Ready-made and custom image/video/audio datasets including biometric and medical, plus synthetic data",
        "status": "operating",
        "url": "https://vaisual.com/",
        "verticals": [
          "image",
          "video",
          "audio",
          "biometric/facial",
          "medical"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published; a Contributor Program exists without rates"
        },
        "workforce": "licensed-corpora"
      },
      "id": "vendor:vaisual",
      "name": "vAIsual",
      "notes": {
        "_org_roles": {
          "enterprise AI teams": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "vaisual.com"
        ],
        "source_urls": [
          "https://vaisual.com/"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://vaisual.com/"
    },
    {
      "fields": {
        "buyer": [
          "rights holders pay for protection",
          "AI developers license NIL"
        ],
        "buyer_price": {
          "availability": "listed",
          "basis": "per-month",
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "published tiers charged TO rights holders for protection (not AI licensing): Free $0/mo, Pro $100/mo (30 takedowns/mo), Pro+ $300/mo (120 takedowns/mo), Premium custom"
        },
        "category": "broker",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "point",
          "max": 16000000,
          "min": 16000000,
          "note": null,
          "raw": "$16M round co-led by Sony Music, March 2025 (musically.com)"
        },
        "name": "Vermillio",
        "note": "Correct domain is vermill.io. The published price is the creator-facing protection product, so it tells you nothing about what AI buyers pay.",
        "orgs": [
          "rights holders pay for protection",
          "AI developers license NIL"
        ],
        "sells": "TraceID: likeness/IP threat monitoring and takedowns for talent, plus licensing enablement of name/image/likeness to AI",
        "status": "operating",
        "url": "https://vermill.io/protection",
        "verticals": [
          "likeness/NIL",
          "music",
          "IP protection"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "no revenue-share published for licensing deals"
        },
        "workforce": "licensed-corpora"
      },
      "id": "vendor:vermillio",
      "name": "Vermillio",
      "notes": {
        "_org_roles": {
          "AI developers license NIL": [
            "buyer"
          ],
          "rights holders pay for protection": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "vermill.io"
        ],
        "source_urls": [
          "https://vermill.io/protection"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "prose-buyer_price",
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://vermill.io/protection"
    },
    {
      "fields": {
        "buyer": [
          "frontier labs",
          "robotics companies"
        ],
        "buyer_price": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "category": "robotics-data",
        "funding_usd": {
          "availability": "listed",
          "basis": null,
          "currency": "USD",
          "kind": "prose",
          "max": null,
          "min": null,
          "note": "multi-clause description; no single figure extracted",
          "raw": "$70M Series A, June 2026 (Thrive, Spark, a16z, Lux, WndrCo) per TechCrunch; Series B talks at about $1.2B reported September 2026, not closed"
        },
        "name": "XDOF",
        "note": "Hires and trains its own teleoperators (managed workforce). Revenue and headcount appear only in press and are excluded. Released ABC open dataset with UC Berkeley.",
        "orgs": [
          "frontier labs",
          "robotics companies"
        ],
        "sells": "Teleoperation rigs (GELLO-style), robot-specific and general teleoperated data, and egocentric human capture via wearables",
        "status": "operating",
        "url": "https://techcrunch.com/2026/06/17/collecting-robot-training-data-is-dirty-unglamorous-work-some-ai-labs-are-already-paying-xdof-to-do-it",
        "verticals": [
          "robot manipulation",
          "teleoperation",
          "egocentric capture"
        ],
        "worker_pay": {
          "availability": "undisclosed",
          "basis": null,
          "currency": null,
          "kind": "none",
          "max": null,
          "min": null,
          "note": "source states no figure is published",
          "raw": "not published"
        },
        "workforce": "expert"
      },
      "id": "vendor:xdof",
      "name": "XDOF",
      "notes": {
        "_org_roles": {
          "frontier labs": [
            "buyer"
          ],
          "robotics companies": [
            "buyer"
          ]
        }
      },
      "provenance": {
        "consent_license": "third-party-derived",
        "first_seen": "2026-09-16",
        "last_seen": "2026-09-16",
        "method": "refinery-explore",
        "source_hosts": [
          "techcrunch.com"
        ],
        "source_urls": [
          "https://techcrunch.com/2026/06/17/collecting-robot-training-data-is-dirty-unglamorous-work-some-ai-labs-are-already-paying-xdof-to-do-it"
        ]
      },
      "quality": {
        "age_days": 2,
        "completeness": 0.909,
        "confidence": "medium",
        "corroborated": false,
        "flags": [
          "prose-funding_usd",
          "single-source"
        ],
        "grade": "B",
        "hosts": 1,
        "independent_hosts": 1,
        "sources": 1,
        "stale": false,
        "stale_after_days": 90
      },
      "type": "vendor",
      "url": "https://techcrunch.com/2026/06/17/collecting-robot-training-data-is-dirty-unglamorous-work-some-ai-labs-are-already-paying-xdof-to-do-it"
    }
  ],
  "_meta": {
    "source": "Blomega Data Refinery",
    "url": "https://data.blomega.com",
    "publisher": "Blomega",
    "publisher_url": "https://blomegalab.com",
    "wikidata": "Q141048865",
    "license": "CC BY 4.0",
    "license_url": "https://creativecommons.org/licenses/by/4.0/",
    "cite_as": "Blomega Data Refinery (https://data.blomega.com), CC BY 4.0. Cite the registry and the record id.",
    "attribution_required": true
  }
}