{
  "title": "Tellseed live internal automation pilot",
  "date": "2026-10-03",
  "type": "Operator-run validation with synthetic answers",
  "customerStudy": false,
  "plan": "Hosted Free",
  "n8nVersion": "2.41.6",
  "transport": "Public HTTPS to Tellseed MCP and production signed webhooks to a temporary n8n receiver",
  "procedure": "https://tellseed.com/guides/automation#pilot",
  "outcome": "passed",
  "counts": {
    "questions": 10,
    "savedAnswers": 10,
    "idempotentSubmissionRetries": 10,
    "completedRounds": 1,
    "finalizedReviews": 10,
    "verifiedExportFiles": 3,
    "businessEventEffects": 23,
    "testEventEffects": 1,
    "replayedEventAttemptsWithoutAdditionalEffects": 1
  },
  "checks": [
    "Draft creation through the native n8n MCP node; retry returned the same draft",
    "Operator approved the exact synthetic draft before the publication workflow",
    "Campaign-scoped credentials used for publication, answer retrieval and exports",
    "Production webhook worker delivered through public DNS and HTTPS",
    "Native n8n Crypto node verified event signatures before a durable destination write",
    "All ten questions completed; repeated submissions did not create extra answers",
    "Reviewed targets reached the destination alongside original answers",
    "All three frozen export file SHA256 hashes matched their manifest",
    "Destination outage caused a production retry without a duplicate destination record",
    "Revoked key could no longer download an export"
  ],
  "cleanup": "Pilot campaign closed; webhook disabled; automation keys, owner sessions and participant access revoked; temporary repository secret removed.",
  "limitations": [
    "The operator provisioned the workspace and temporary session; email signup was not part of this run.",
    "Answers and review decisions were scripted synthetic material, not customer research or human labels.",
    "The receiver and SQLite destination ran on a disposable runner; persistent customer destinations and n8n Cloud remain unverified.",
    "The real payment test was not performed.",
    "This is functional evidence, not a performance benchmark, adoption claim or model-quality evaluation."
  ]
}