{
  "schema_version": "proof-evidence-v1",
  "id": "proof-cold-start-comparison",
  "slug": "cold-start-comparison",
  "title": "Cold start vs. system-backed work",
  "audience": [
    "buyer",
    "technical evaluator",
    "operator"
  ],
  "proof_type": "comparison",
  "maturity": "pilot evidence",
  "claim_status": "measured pilot",
  "visibility": "public summary",
  "sanitization_status": "passed",
  "use_case": "A team wants to know whether a new agent produces better work when it starts from the project system instead of a bare task prompt.",
  "claim": "System-backed runs preserve prior decisions, reuse existing structures, and surface clearer next actions than cold runs on the same scenario set.",
  "diff_summary": "The visible difference is not prose polish. It is continuity: the second run starts with project memory and keeps the work attached to it.",
  "review_manifest": {
    "evidence_room_url": "/proof/cold-start-comparison#evidence-room",
    "evidence_packet_url": "/proof/cold-start-comparison/evidence.json",
    "fact_count": 3,
    "visible_value_count": 8,
    "data_point_count": 8,
    "source_count": 3,
    "proof_quality_score": 4.8,
    "weakest_quality_dimension": "visual-hierarchy"
  },
  "quality_review": {
    "overall_score": 4.8,
    "overall_label": "Launch-readable",
    "overall_note": "This is a cold-review rubric, not a sales claim. It checks whether the case can be understood, inspected, bounded, and acted on.",
    "dimensions": [
      {
        "id": "cold-audience-clarity",
        "label": "Cold-audience clarity",
        "score": 5,
        "signal": "A reviewer gets the situation, contrast, and buyer decision without reading the JSON.",
        "next": "Tighten the opening latch or before/after language when the case still needs explanation."
      },
      {
        "id": "evidence-visibility",
        "label": "Evidence visibility",
        "score": 5,
        "signal": "The case exposes proof rows, values, source handles, and excerpts before the JSON export.",
        "next": "Add openable artifacts, row links, or public excerpts when the proof still feels referenced rather than shown."
      },
      {
        "id": "limitation-clarity",
        "label": "Limitation clarity",
        "score": 5,
        "signal": "The page names what the case proves and what it does not prove.",
        "next": "Add a sharper boundary when a reader could confuse a demonstrator with a benchmark."
      },
      {
        "id": "visual-hierarchy",
        "label": "Visual hierarchy",
        "score": 4,
        "signal": "The route uses visuals, storyboards, or surface frames to break the proof into inspectable layers.",
        "next": "Promote the proof row into a diagram, table, screenshot projection, or openable artifact."
      },
      {
        "id": "purchase-motivation",
        "label": "Purchase motivation",
        "score": 5,
        "signal": "The proof sends the buyer to an evidence room, artifact link, pricing path, or local run.",
        "next": "Tie the proof to local inspection, pricing, account support, or a narrower try/reject decision."
      }
    ],
    "strongest_dimension": {
      "id": "purchase-motivation",
      "label": "Purchase motivation",
      "score": 5,
      "signal": "The proof sends the buyer to an evidence room, artifact link, pricing path, or local run.",
      "next": "Tie the proof to local inspection, pricing, account support, or a narrower try/reject decision."
    },
    "weakest_dimension": {
      "id": "visual-hierarchy",
      "label": "Visual hierarchy",
      "score": 4,
      "signal": "The route uses visuals, storyboards, or surface frames to break the proof into inspectable layers.",
      "next": "Promote the proof row into a diagram, table, screenshot projection, or openable artifact."
    },
    "next_improvement": "Promote the proof row into a diagram, table, screenshot projection, or openable artifact."
  },
  "evidence_room_url": "/proof/cold-start-comparison#evidence-room",
  "evidence_facts": [
    {
      "id": "paired-scenarios-reviewed",
      "label": "Five paired scenarios were reviewed",
      "observed": "The pilot evidence set contains five anonymized tasks run once from a prompt-only starting point and once from a system-backed starting point.",
      "evidence": "Run metadata lists five paired scenarios; the comparison matrix is retained as the public projection source.",
      "supports": "Supports the claim that the comparison is based on paired cases, not unrelated examples.",
      "status": "public projection",
      "review": {
        "row": 1,
        "public_review_url": "/proof/cold-start-comparison#fact-paired-scenarios-reviewed",
        "packet_pointer": "evidence_facts[0]",
        "visible_value_refs": [
          "data-01-scenario-set",
          "data-03-claim-boundary",
          "data-05-sample",
          "data-06-provenance"
        ],
        "data_refs": [
          "data-01-scenario-set",
          "data-03-claim-boundary",
          "data-05-sample",
          "data-06-provenance"
        ],
        "source_ids": [
          "source-02-system-backed-outputs",
          "source-03-comparison-matrix"
        ],
        "case_claim_status": "measured pilot",
        "case_visibility": "public summary"
      }
    },
    {
      "id": "system-backed-outputs-carry-continuity",
      "label": "System-backed outputs carried named context",
      "observed": "The system-backed summaries include inherited decisions, named reusable assets, and explicit limitation language that the cold summaries did not consistently carry.",
      "evidence": "Before/after excerpts and metric notes on the proof detail page name reuse, inherited decisions, and explicit limitations as the primary reviewed signal.",
      "supports": "Supports the narrow claim that continuity improved in the inspected pilot set.",
      "status": "public projection",
      "review": {
        "row": 2,
        "public_review_url": "/proof/cold-start-comparison#fact-system-backed-outputs-carry-continuity",
        "packet_pointer": "evidence_facts[1]",
        "visible_value_refs": [
          "data-02-primary-signal",
          "data-03-claim-boundary",
          "data-07-reproducibility",
          "data-08-claim-status"
        ],
        "data_refs": [
          "data-02-primary-signal",
          "data-03-claim-boundary",
          "data-07-reproducibility",
          "data-08-claim-status"
        ],
        "source_ids": [
          "source-01-cold-outputs",
          "source-02-system-backed-outputs",
          "source-03-comparison-matrix"
        ],
        "case_claim_status": "measured pilot",
        "case_visibility": "public summary"
      }
    },
    {
      "id": "no-productivity-number-published",
      "label": "No productivity number is published",
      "observed": "The public case labels the result as a measured pilot and explicitly avoids broad ROI or productivity lift claims.",
      "evidence": "Claim status, metric note, and limitations all state that this is pilot evidence rather than a statistical productivity result.",
      "supports": "Supports claim discipline and prevents the comparison from being overstated.",
      "status": "publicly inspectable",
      "review": {
        "row": 3,
        "public_review_url": "/proof/cold-start-comparison#fact-no-productivity-number-published",
        "packet_pointer": "evidence_facts[2]",
        "visible_value_refs": [
          "data-03-claim-boundary",
          "data-06-provenance",
          "data-08-claim-status"
        ],
        "data_refs": [
          "data-03-claim-boundary",
          "data-06-provenance",
          "data-08-claim-status"
        ],
        "source_ids": [
          "source-03-comparison-matrix"
        ],
        "case_claim_status": "measured pilot",
        "case_visibility": "public summary"
      }
    }
  ],
  "fact_review_table": [
    {
      "row": 1,
      "id": "paired-scenarios-reviewed",
      "label": "Five paired scenarios were reviewed",
      "status": "public projection",
      "observed": "The pilot evidence set contains five anonymized tasks run once from a prompt-only starting point and once from a system-backed starting point.",
      "evidence": "Run metadata lists five paired scenarios; the comparison matrix is retained as the public projection source.",
      "supports": "Supports the claim that the comparison is based on paired cases, not unrelated examples.",
      "source_ids": [
        "source-02-system-backed-outputs",
        "source-03-comparison-matrix"
      ],
      "visible_value_refs": [
        "data-01-scenario-set",
        "data-03-claim-boundary",
        "data-05-sample",
        "data-06-provenance"
      ],
      "data_refs": [
        "data-01-scenario-set",
        "data-03-claim-boundary",
        "data-05-sample",
        "data-06-provenance"
      ],
      "public_review_url": "/proof/cold-start-comparison#fact-paired-scenarios-reviewed",
      "packet_pointer": "evidence_facts[0]"
    },
    {
      "row": 2,
      "id": "system-backed-outputs-carry-continuity",
      "label": "System-backed outputs carried named context",
      "status": "public projection",
      "observed": "The system-backed summaries include inherited decisions, named reusable assets, and explicit limitation language that the cold summaries did not consistently carry.",
      "evidence": "Before/after excerpts and metric notes on the proof detail page name reuse, inherited decisions, and explicit limitations as the primary reviewed signal.",
      "supports": "Supports the narrow claim that continuity improved in the inspected pilot set.",
      "source_ids": [
        "source-01-cold-outputs",
        "source-02-system-backed-outputs",
        "source-03-comparison-matrix"
      ],
      "visible_value_refs": [
        "data-02-primary-signal",
        "data-03-claim-boundary",
        "data-07-reproducibility",
        "data-08-claim-status"
      ],
      "data_refs": [
        "data-02-primary-signal",
        "data-03-claim-boundary",
        "data-07-reproducibility",
        "data-08-claim-status"
      ],
      "public_review_url": "/proof/cold-start-comparison#fact-system-backed-outputs-carry-continuity",
      "packet_pointer": "evidence_facts[1]"
    },
    {
      "row": 3,
      "id": "no-productivity-number-published",
      "label": "No productivity number is published",
      "status": "publicly inspectable",
      "observed": "The public case labels the result as a measured pilot and explicitly avoids broad ROI or productivity lift claims.",
      "evidence": "Claim status, metric note, and limitations all state that this is pilot evidence rather than a statistical productivity result.",
      "supports": "Supports claim discipline and prevents the comparison from being overstated.",
      "source_ids": [
        "source-03-comparison-matrix"
      ],
      "visible_value_refs": [
        "data-03-claim-boundary",
        "data-06-provenance",
        "data-08-claim-status"
      ],
      "data_refs": [
        "data-03-claim-boundary",
        "data-06-provenance",
        "data-08-claim-status"
      ],
      "public_review_url": "/proof/cold-start-comparison#fact-no-productivity-number-published",
      "packet_pointer": "evidence_facts[2]"
    }
  ],
  "visible_values": [
    {
      "id": "data-01-scenario-set",
      "kind": "metric",
      "label": "Scenario set",
      "value": "5",
      "note": "Five anonymized work scenarios were compared as cold and system-backed runs.",
      "public_review_url": "/proof/cold-start-comparison#data-01-scenario-set"
    },
    {
      "id": "data-02-primary-signal",
      "kind": "metric",
      "label": "Primary signal",
      "value": "reuse",
      "note": "Cases were reviewed for inherited decisions, named reusable assets, and explicit limitations.",
      "public_review_url": "/proof/cold-start-comparison#data-02-primary-signal"
    },
    {
      "id": "data-03-claim-boundary",
      "kind": "metric",
      "label": "Claim boundary",
      "value": "pilot",
      "note": "Good enough to inspect; not a broad numeric ROI claim.",
      "public_review_url": "/proof/cold-start-comparison#data-03-claim-boundary"
    },
    {
      "id": "data-04-run-date",
      "kind": "metadata",
      "label": "Run date",
      "value": "June 2026",
      "note": "Date label attached to this public proof case.",
      "public_review_url": "/proof/cold-start-comparison#data-04-run-date"
    },
    {
      "id": "data-05-sample",
      "kind": "metadata",
      "label": "Sample",
      "value": "Five paired scenarios",
      "note": "Sample size or unit under review.",
      "public_review_url": "/proof/cold-start-comparison#data-05-sample"
    },
    {
      "id": "data-06-provenance",
      "kind": "metadata",
      "label": "Provenance",
      "value": "Public projection of internal comparison runs",
      "note": "Where the public projection says the evidence came from.",
      "public_review_url": "/proof/cold-start-comparison#data-06-provenance"
    },
    {
      "id": "data-07-reproducibility",
      "kind": "metadata",
      "label": "Reproducibility",
      "value": "Representative excerpts are public; raw transcripts remain internal until fully redacted.",
      "note": "What a reviewer can repeat or inspect publicly.",
      "public_review_url": "/proof/cold-start-comparison#data-07-reproducibility"
    },
    {
      "id": "data-08-claim-status",
      "kind": "metadata",
      "label": "Claim status",
      "value": "measured pilot",
      "note": "The maximum strength of the public claim.",
      "public_review_url": "/proof/cold-start-comparison#data-08-claim-status"
    }
  ],
  "source_register": [
    {
      "id": "source-01-cold-outputs",
      "label": "Cold outputs",
      "kind": "run transcript set",
      "status": "internal source retained",
      "public_review_url": "/proof/cold-start-comparison#source-01-cold-outputs"
    },
    {
      "id": "source-02-system-backed-outputs",
      "label": "System-backed outputs",
      "kind": "run transcript set",
      "status": "internal source retained",
      "public_review_url": "/proof/cold-start-comparison#source-02-system-backed-outputs"
    },
    {
      "id": "source-03-comparison-matrix",
      "label": "Comparison matrix",
      "kind": "review table",
      "status": "public projection",
      "public_review_url": "/proof/cold-start-comparison#source-03-comparison-matrix"
    }
  ],
  "metrics": [
    {
      "label": "Scenario set",
      "value": "5",
      "note": "Five anonymized work scenarios were compared as cold and system-backed runs."
    },
    {
      "label": "Primary signal",
      "value": "reuse",
      "note": "Cases were reviewed for inherited decisions, named reusable assets, and explicit limitations."
    },
    {
      "label": "Claim boundary",
      "value": "pilot",
      "note": "Good enough to inspect; not a broad numeric ROI claim."
    }
  ],
  "run_metadata": {
    "date_label": "June 2026",
    "sample_size": "Five paired scenarios",
    "provenance": "Public projection of internal comparison runs",
    "reproducibility": "Representative excerpts are public; raw transcripts remain internal until fully redacted."
  },
  "source_artifacts": [
    {
      "label": "Cold outputs",
      "kind": "run transcript set",
      "status": "internal source retained"
    },
    {
      "label": "System-backed outputs",
      "kind": "run transcript set",
      "status": "internal source retained"
    },
    {
      "label": "Comparison matrix",
      "kind": "review table",
      "status": "public projection"
    }
  ],
  "public_evidence_excerpts": [
    {
      "id": "paired-scenario-review-excerpt",
      "label": "Paired scenario review excerpt",
      "source_label": "Comparison matrix",
      "status": "retained-source summary",
      "excerpt": "Five anonymized work scenarios were reviewed in paired form: one prompt-only run and one project-system-backed run for each scenario.",
      "redaction_note": "Source run text remains retained; the public excerpt preserves the comparison structure without task names or private workspace details.",
      "supports_fact_ids": [
        "paired-scenarios-reviewed",
        "system-backed-outputs-carry-continuity"
      ]
    },
    {
      "id": "claim-boundary-review-excerpt",
      "label": "Claim boundary review excerpt",
      "source_label": "Public proof case",
      "status": "public excerpt",
      "excerpt": "The public case is labeled as measured pilot evidence and does not publish a productivity-lift number.",
      "redaction_note": "This excerpt is visible on the public page and constrains the sales claim.",
      "supports_fact_ids": [
        "no-productivity-number-published"
      ]
    }
  ],
  "limitations": [
    "Small sample size.",
    "Anonymized scenarios only.",
    "The result supports a launch proof, not a statistical productivity claim."
  ],
  "visuals": [
    {
      "label": "Cold prompt vs. system-backed lane",
      "kind": "comparison diagram",
      "alt": "Two lanes compare prompt-only work with a system-backed run that adds project state, decisions, reusable surfaces, and scoped actions.",
      "caption": "The visible signal is continuity: the better run inherits structure before it writes.",
      "source_status": "sanitized projection",
      "component": "cold-start-lanes",
      "src": null
    }
  ],
  "cta": {
    "label": "Read how the system works",
    "href": "/how-it-works"
  }
}