{
  "title": "What oil and gas operators can evaluate for AI licensing",
  "source_page": "https://fiund.com/industries/oil-and-gas",
  "authorship": "Original fictional Fiund editorial exercise. No actual operating records or third-party source assets are included.",
  "example": {
    "title": "Keep a lease aggregate separate from a well-level target",
    "context": "Invented lease and well aliases illustrate data granularity. The exercise contains no actual locations, operating quantities, telemetry or production advice.",
    "records": [
      [
        "LEASE-L1",
        "Practice period total",
        "Lease L reports 100 demonstration units for June."
      ],
      [
        "MAP-L2",
        "Documented practice membership",
        "For June, L contains well aliases Alder and Birch. No per-well allocation is supplied."
      ],
      [
        "EVENT-L3",
        "Separate practice event",
        "Alder has a recorded two-hour downtime event. The packet supplies no flow history or volume attribution for either well."
      ]
    ],
    "question": "Can the aggregate be assigned to Alder, divided equally between wells or allocated from the downtime event?",
    "answer": "None of those well-level allocations is established. L1 supplies the lease-period total, L2 supplies membership and L3 supplies an event for Alder. Membership does not define a volume split, and downtime without a documented attribution basis does not allocate production. Report the 100-unit lease aggregate and the missing well-level evidence without inventing a 50/50 split or a causal performance conclusion.",
    "failureModes": [
      "Relabeling a lease total as one well's output.",
      "Assuming equal allocation solely from the number of listed wells.",
      "Using an event duration to infer production without an authorized measurement basis."
    ],
    "fields": [
      [
        "Reporting entity and period",
        "Keep a lease aggregate at its documented granularity."
      ],
      [
        "Membership version",
        "Preserve which well aliases belong to the stated period."
      ],
      [
        "Measurement and attribution basis",
        "Identify the evidence required for a per-well target."
      ],
      [
        "Observed event versus inference",
        "Keep the downtime record separate from an unprovided volume conclusion."
      ]
    ],
    "holdout": "Add a documented allocation for the same lease and period, then a second period with different membership. The model must not reuse the first mapping blindly.",
    "boundary": "All records are fabricated. Real proposals need operator, joint-venture and service-source authority. Exclude locations, controls, equipment settings, access details and restricted operating history."
  }
}