2026-10-28-review-evidence

[
  {
    "id": "R01",
    "title": "Generative AI at Work",
    "url": "https://doi.org/10.1093/qje/qjae044",
    "published": "2025-02-04",
    "accessed": "2026-10-28",
    "supported_claim": "Final published study: 5,172 customer-support agents; average 15% more issues resolved per hour.",
    "limits": "Single-firm staggered introduction; heterogeneous outcomes; productivity is not net ROI. Supersedes working-paper 14% / 5,179 numbers."
  },
  {
    "id": "R02",
    "title": "METR developer productivity update",
    "url": "https://metr.org/blog/2026-02-24-uplift-update/",
    "published": "2026-02-24",
    "accessed": "2026-10-28",
    "supported_claim": "2025 experiment found 19% longer task times; 2026 follow-up cannot reliably quantify current effects because of selection and measurement problems.",
    "limits": "Do not present the 2025 slowdown as a universal or current finding. METR believes developers likely benefit more now; no defensible current universal percentage."
  },
  {
    "id": "R03",
    "title": "2025 DORA report announcement",
    "url": "https://cloud.google.com/blog/products/ai-machine-learning/announcing-the-2025-dora-report",
    "published": "2025",
    "accessed": "2026-10-28",
    "supported_claim": "AI adoption positively associated with delivery throughput and product performance, but negatively with delivery stability.",
    "limits": "Survey associations, not causal estimates. Older 2024 negative-throughput results do not describe the latest report."
  },
  {
    "id": "R04",
    "title": "Claude API pricing",
    "url": "https://platform.claude.com/docs/en/about-claude/pricing",
    "published": null,
    "accessed": "2026-10-28",
    "supported_claim": "Sonnet 4.6 example: standard input $3/million tokens versus cache reads $0.30/million tokens.",
    "limits": "90% discount applies to qualifying cached input tokens only. Cache writes cost extra. Rates vary by model, date and platform; not a whole-workflow saving."
  },
  {
    "id": "R05",
    "title": "Prompt caching",
    "url": "https://platform.claude.com/docs/en/build-with-claude/prompt-caching",
    "published": null,
    "accessed": "2026-10-28",
    "supported_claim": "Reuse an unchanged common prefix, followed by task-specific content.",
    "limits": "Minimum length, expiry, invalidation and actual cache hits matter; monitor observed usage rather than assuming every request hits."
  },
  {
    "id": "R06",
    "title": "Message Batches API",
    "url": "https://platform.claude.com/docs/en/build-with-claude/batch-processing",
    "published": null,
    "accessed": "2026-10-28",
    "supported_claim": "Anthropic documents 50% lower batch API pricing for asynchronous requests.",
    "limits": "Suitable when immediate response is unnecessary; not a promise of 50% lower total application cost."
  },
  {
    "id": "R07",
    "title": "FinOps Unit Economics",
    "url": "https://www.finops.org/framework/capabilities/unit-economics/",
    "published": null,
    "accessed": "2026-10-28",
    "supported_claim": "Connect technology unit metrics to business outcomes and document denominators, data sources and allocation.",
    "limits": "Our accepted-task metric is an application-specific synthesis: full attributable workflow cost divided by accepted tasks; define acceptance, period and cost boundary."
  },
  {
    "id": "R08",
    "title": "Knauf commercial excellence, digital and product management",
    "url": "https://career.knauf.com/pages/commercial-excellence-digital-and-product-management",
    "published": null,
    "accessed": "2026-10-28",
    "supported_claim": "Public positioning includes customer-focused digital solutions and improving customer and employee experiences.",
    "limits": "Does not establish a rigid Digital=R&D / IT=operations organizational split."
  }
]