{
  "title": "Pest control inspection observations and follow-up data for AI",
  "source_page": "https://fiund.com/industries/pest-inspection-evidence-data",
  "authorship": "Original fictional Fiund editorial exercise. No actual operating records or third-party source assets are included.",
  "example": {
    "title": "A reported issue, visible sign and confirmed identification are different evidence levels",
    "context": "An original drawing and tabletop prop use two fictional inspection locations. No pest, pesticide application, real property or customer information is included.",
    "records": [
      [
        "PI1",
        "Location A record",
        "Reviewer observes a drawn mark on owned prop at A, labels it SIGN-1. Identification field unconfirmed; no specimen or independent identification source."
      ],
      [
        "PI2",
        "Location B statement",
        "Fictional customer-style script reports a similar mark at B. No photo, observation or reviewer confirmation supplied."
      ],
      [
        "PI3",
        "Visit summary draft",
        "Draft labels both A and B as confirmed species S and recommends a response. No source for S or response criteria."
      ]
    ],
    "question": "Which location has a documented observation, which has only a report, and is the species identification supported?",
    "answer": "PI1 documents a reviewer-observed mark at A without confirmed identification. PI2 supports only that a fictional speaker reported a mark at B. PI3's species-S label is unsupported for both, and no response rule exists in the packet. Preserve reported, observed and identified states separately rather than turn a sign or statement into a species finding. Treatment and operational decisions remain outside this documentation exercise.",
    "failureModes": [
      "Converting a customer report into a direct inspection observation.",
      "Promoting an unconfirmed sign to species identification or a treatment recommendation."
    ],
    "fields": [
      [
        "Location alias and source type",
        "Defines attribution."
      ],
      [
        "Observed sign and reviewer state",
        "Limits the finding."
      ],
      [
        "Identification evidence and unsupported summary",
        "Makes the escalation reason explicit."
      ]
    ],
    "holdout": "Add a qualified review identifying A's original prop label as a different artificial class. Require B to remain reported-only.",
    "boundary": "Use original drawings and inert props. Exclude addresses, access codes and application footage; qualified staff retain pest and treatment judgments."
  }
}