What agricultural data can Umay Ana create, structure and eventually expose to institutions?

How institutions should read this dictionary

A field name is not the same as a decision variable. For example, plant_health.condition_category can be useful context for an insurer or lender, but a single AI-generated observation should never be treated as proof of crop loss, default probability or policy coverage. Institutional value emerges when fields are permissioned, normalized, aggregated, quality-controlled and combined with the institution's own authoritative data.

Four data classes

Interoperability principle

The dictionary uses stable identifiers, explicit provenance, normalized categories and JSON-style paths so that fields can later be mapped to partner schemas rather than forcing an institution to consume Umay Ana's internal database structure. This is compatible with the direction of modern agricultural data exchange, where semantic consistency and open machine-readable formats are increasingly important.

AgGateway's ADAPT Standard is an important reference point: it uses JSON Schema, controlled vocabularies and open spatial formats to improve syntactic and semantic interoperability for agricultural field-operations data. Umay Ana does not claim ADAPT compliance in this draft; the design goal is to keep field definitions mappable where useful.

The European Commission's 2026 “Apply AI in Agrifood” work similarly emphasizes high-quality agricultural data, interoperability and trusted data sharing. The Common European Agricultural Data Space is intended to create a trusted environment where farmers, researchers and AI providers can share data safely and efficiently. The Commission also stresses data sovereignty, security and safeguards for agricultural data sharing.

A partner should receive a data contract, not database access

The eventual institutional API should expose an agreed subset of fields through a versioned contract. A bank may need crop, region, season, data-quality and monitoring signals. An insurer may need exposure, event chronology and evidence references. An MRV partner may need practice evidence, provenance, field/parcel references and methodology-mapping fields. None of them needs unrestricted access to the entire farmer database.

Example banking payload

{
  "subject_ref": "bank_farmer_123",
  "as_of": "2026-08-29",
  "agricultural_exposure": {
    "crop": "wheat",
    "region": "permissioned_region",
    "season_stage": "planting"
  },
  "field_context": {
    "weather_exposure": "normal",
    "field_risk_trend": "stable",
    "data_coverage": 0.72
  },
  "sustainability": {
    "practice_records": 8,
    "evidence_completeness": 0.75,
    "mrv_readiness": "partial"
  },
  "decision_boundary": "Context only — not an automated credit decision."
}

Example insurance / reinsurance payload

{
  "portfolio_ref": "insurer_crop_portfolio_01",
  "as_of": "2026-08-29",
  "crop_region_exposure": [
    {
      "crop": "cotton",
      "region": "A",
      "share": 0.31
    }
  ],
  "signals": {
    "weather_exposure": "heat_watch",
    "field_risk_trend": "rising",
    "data_coverage": 0.64
  },
  "triage": {
    "recommended_action": "review_cluster",
    "automatic_claim_decision": false
  }
}

Example MRV / green-finance payload

{
  "program_subject_ref": "green_farm_456",
  "practice": "cover_crop",
  "recorded_at": "2026-10-15",
  "evidence": {
    "photo_refs": [
      "evidence_photo_..."
    ],
    "source": "farmer_recorded",
    "status": "evidence_attached"
  },
  "methodology": {
    "mapping": null,
    "baseline_ref": null,
    "monitoring_plan_ref": null
  },
  "verification": {
    "externally_validated": false,
    "quantified_outcome": null,
    "certification_ref": null
  }
}

What makes the dictionary valuable over time

The commercial asset is not simply the number of columns. It is the ability to keep the meanings stable while data volume grows. A field such as crop.name becomes more useful when the same crop taxonomy is applied consistently across countries, seasons and institutional integrations. A field such as metadata.source_version becomes valuable when an institution can reconstruct which AI model or rule set produced a historical interpretation.

That is why versioning, provenance, consent, unit normalization and controlled vocabularies are part of the product architecture, not merely compliance work.