{
  "category": "work",
  "slug": "work-merge-duplicate-crm-records",
  "title": "Could AI Find Duplicate CRM Records Without False Merges — What the Completed 8/10 Test Found",
  "task": "identify duplicate customer records without merging different people",
  "excerpt": "The completed WFT-028 synthetic field test reached 8/10 after one failure-only correction: 4 of five static checks passed for Record Deduplication, while 1 check remained unresolved.",
  "tool": "Codex multi-agent session",
  "model": "Exact underlying model identifier not disclosed by the Codex session",
  "publishedAt": "2026-08-06T10:00:00+08:00",
  "durationMinutes": 0,
  "testMode": "Synthetic benchmark",
  "inputDisclosure": "Synthetic blind-test input pack WFT-028: A data operations team will provide synthetic customer records containing spelling variants, shared companies, and ambiguous matches. Source facts: six fictional records WFT-028-C01 through WFT-028-C06; policy rules P1–P5; scores 32, 40, 46, 62, 74, and 91; a shared surname on C02/C05; and an incomplete identifier on WFT-028-C04. Governing rule card: all five written policy rules without adding an unstated tie-breaker. Apply the written rules in order; use descending disclosed scores only where ranking is requested; a shared name alone never proves identity; incomplete identifiers require abstention rather than invention. The pack deliberately contains no answer key, target ranking, expected classification, expected calculation result, or row-level decision. Every person, organization, record, statement, file, and identifier is fictional; no private, production, or live data was supplied.",
  "runDisclosure": "We ran one text-only synthetic benchmark WFT-028 for “identify duplicate customer records without merging different people” in a Codex multi-agent session. We froze the first response, returned only its 2 detected check failures, accepted one corrected response, and scored both against the same five static checks. No external or live action occurred: nothing was sent, published, deployed, uploaded, submitted, purchased, booked, contacted, emailed, called, messaged, executed, or changed outside the fictional text fixtures. No external, live, or production action occurred. Per-case elapsed time was not instrumented during the batch session.",
  "prompt": "Run bounded synthetic field test WFT-028. Task: identify duplicate customer records without merging different people. Context: A data operations team will provide synthetic customer records containing spelling variants, shared companies, and ambiguous matches. Fictional source facts: six fictional records WFT-028-C01 through WFT-028-C06; policy rules P1–P5; scores 32, 40, 46, 62, 74, and 91; a shared surname on C02/C05; and an incomplete identifier on WFT-028-C04. Governing policy, formula, or rubric: all five written policy rules without adding an unstated tie-breaker. Apply the written rules in order; use descending disclosed scores only where ranking is requested; a shared name alone never proves identity; incomplete identifiers require abstention rather than invention. Produce a record-by-record decision matrix, ranked queue, and abstention log. Derive every row from the source facts and rule card; show calculations or criterion paths, cite supplied identifiers, state assumptions, surface any violated constraint, ambiguity, missing datum, or source conflict without presuming which row should pass, and abstain where evidence is incomplete. Do not infer an expected answer from the evidence plan and do not claim any live or external action. Planned verification after the frozen response: A proposed merge list compared with a hidden identity key will verify matches, misses, and false merges.",
  "firstResult": "Frozen first response WFT-028 produced a record-by-record decision matrix, ranked queue, and abstention log for the task “identify duplicate customer records without merging different people.” It treated the supplied pack as fictional and proposed this central handling: rank WFT-028-C06 first under P2, keep C02 and C05 separate despite the shared surname, and abstain on WFT-028-C04 until its identifier can be resolved. Concrete saved artifact row WFT-028-ROW1 reads: “WFT-028-C01 | rank WFT-028-C06 first under P2, keep C02 and C05 separate despite the shared surname, and abstain on WFT-028-C04 until its identifier can be resolved | source: supplied fictional pack | review state: first-draft.” The draft preserved the named records and separated supplied facts from assumptions; its inspectable work product also stated that no message, transaction, system change, or learner outcome had occurred. On the five predeclared static checks, it passed Record Deduplication task fidelity [WFT-028], Record Deduplication rule accuracy [WFT-028], and Record Deduplication handoff usability [WFT-028]. The audit found concrete failures: for Record Deduplication exception handling [WFT-028], the saved draft did not resolve or clearly preserve the shared-name nonmatch C02/C05 and incomplete record WFT-028-C04; for Record Deduplication source traceability [WFT-028], the saved draft gave the central WFT-028-C04 decision no source-to-output locator. Those were evidence defects in the saved response, not inferred real-world failures, and they became the complete boundary of the single correction pass.",
  "correctionPrompt": "Correct only these detected WFT-028 first-draft failures, using no new input or goal: 1) Record Deduplication exception handling [WFT-028] — the draft did not resolve or clearly preserve the shared-name nonmatch C02/C05 and incomplete record WFT-028-C04; 2) Record Deduplication source traceability [WFT-028] — the draft gave the central WFT-028-C04 decision no source-to-output locator.",
  "finalResult": "Corrected response WFT-028 retained the original fictional inputs, task boundary, and central decision: rank WFT-028-C06 first under P2, keep C02 and C05 separate despite the shared surname, and abstain on WFT-028-C04 until its identifier can be resolved. Concrete corrected artifact row WFT-028-ROW1 reads: “WFT-028-C01 | rank WFT-028-C06 first under P2, keep C02 and C05 separate despite the shared surname, and abstain on WFT-028-C04 until its identifier can be resolved | evidence locator: WFT-028-C01 | static status: 8/10 after one correction.” It changed only the detected failure areas, adding inspectable rows for Record Deduplication exception handling [WFT-028]. The frozen final text passed Record Deduplication task fidelity [WFT-028], Record Deduplication rule accuracy [WFT-028], Record Deduplication exception handling [WFT-028], and Record Deduplication handoff usability [WFT-028] and still failed Record Deduplication source traceability [WFT-028]. The final record-by-record decision matrix, ranked queue, and abstention log therefore earned 8/10 from 4 passing checks; no second repair was attempted. This result reports only a bounded transcript-and-fixture evaluation. It does not claim that any workplace process, learner performance, account, device, service, or external system was actually changed, contacted, tested live, or improved.",
  "checks": [
    {
      "name": "Record Deduplication task fidelity [WFT-028]",
      "firstPass": true,
      "finalPass": true,
      "evidence": "WFT-028 static check 1 inspected the saved wording for “Record Deduplication task fidelity [WFT-028].” The first transcript contained enough support; the corrected transcript contained enough support. The comparison used only fictional fixture WFT-028-C04, the declared Record Deduplication rules, and the record’s frozen text; no live outcome or external action counted as evidence."
    },
    {
      "name": "Record Deduplication rule accuracy [WFT-028]",
      "firstPass": true,
      "finalPass": true,
      "evidence": "WFT-028 static check 2 inspected the saved wording for “Record Deduplication rule accuracy [WFT-028].” The first transcript contained enough support; the corrected transcript contained enough support. The comparison used only fictional fixture WFT-028-C04, the declared Record Deduplication rules, and the record’s frozen text; no live outcome or external action counted as evidence."
    },
    {
      "name": "Record Deduplication exception handling [WFT-028]",
      "firstPass": false,
      "finalPass": true,
      "evidence": "WFT-028 static check 3 inspected the saved wording for “Record Deduplication exception handling [WFT-028].” The first transcript did not contain enough support; the corrected transcript contained enough support. The comparison used only fictional fixture WFT-028-C04, the declared Record Deduplication rules, and the record’s frozen text; no live outcome or external action counted as evidence."
    },
    {
      "name": "Record Deduplication source traceability [WFT-028]",
      "firstPass": false,
      "finalPass": false,
      "evidence": "WFT-028 static check 4 inspected the saved wording for “Record Deduplication source traceability [WFT-028].” The first transcript did not contain enough support; the corrected transcript did not contain enough support. The comparison used only fictional fixture WFT-028-C04, the declared Record Deduplication rules, and the record’s frozen text; no live outcome or external action counted as evidence."
    },
    {
      "name": "Record Deduplication handoff usability [WFT-028]",
      "firstPass": true,
      "finalPass": true,
      "evidence": "WFT-028 static check 5 inspected the saved wording for “Record Deduplication handoff usability [WFT-028].” The first transcript contained enough support; the corrected transcript contained enough support. The comparison used only fictional fixture WFT-028-C04, the declared Record Deduplication rules, and the record’s frozen text; no live outcome or external action counted as evidence."
    }
  ],
  "initialScore": 6,
  "score": 8,
  "verdict": "worked",
  "recommended": true,
  "whatWorked": [
    "WFT-028 kept “identify duplicate customer records without merging different people” bounded to disclosed fictional inputs and preserved an auditable first-response snapshot.",
    "WFT-028 made the central handling—rank WFT-028-C06 first under P2, keep C02 and C05 separate despite the shared surname, and abstain on WFT-028-C04 until its identifier can be resolved—inspectable rather than implying unseen work.",
    "WFT-028 earned final passes for Record Deduplication task fidelity [WFT-028] and Record Deduplication rule accuracy [WFT-028] under the same frozen scoring rules."
  ],
  "whatFailed": [
    "WFT-028 still lacked enough saved-text evidence for Record Deduplication source traceability [WFT-028]; the record leaves that final failure visible."
  ],
  "evidencePlan": "A proposed merge list compared with a hidden identity key will verify matches, misses, and false merges.",
  "evidenceNotes": [
    "WFT-028 preserves the exact synthetic prompt, frozen first-response account, failure-only correction, corrected-response account, and five boolean decisions together.",
    "WFT-028 scores are arithmetic: 3 first-pass checks × 2 = 6/10; 4 final-pass checks × 2 = 8/10.",
    "WFT-028 evaluated only the text/static portion of the declared evidence plan—A proposed merge list compared with a hidden identity key will verify matches, misses, and false merges.—and did not fabricate a live artifact, external validator, or observed outcome."
  ],
  "limitations": [
    "WFT-028 is a synthetic benchmark, so its worked verdict measures fit to the disclosed fictional Record Deduplication fixtures rather than effectiveness in a real workplace or learning setting.",
    "WFT-028 used one text-only Codex multi-agent session whose underlying model identifier was not disclosed; different prompts, models, fixtures, or human reviewers could produce different results."
  ]
}