The Umay Ana Institutional Pilot Package is a configurable reference program for banks, insurers, reinsurers, MRV providers and public institutions. The default scenario uses 10,000 invited farmers over 90 days to test adoption, field-signal generation, data quality, integration and institutional usefulness before a production-scale commitment.
Whether farmer-facing utility can generate structured agricultural context that is complete enough, permissioned enough and operationally useful enough to justify production integration.
The pilot does not require the bank, insurer or MRV partner to accept an unvalidated credit model, automatic claim decision, carbon calculation or predictive risk score.
Scope, cohort, consent, use case, integration and success criteria.
Partner mapping, onboarding flows, dashboard/API sandbox and support process.
Farmer activation, usage education and first data-quality review.
Field-signal generation, portfolio views and institutional workflow testing.
KPI review, outcome-validation analysis and scale/no-scale decision.
WS1 · Onboarding & ActivationMap the cohort, establish purpose/consent, activate farmers and measure funnel friction.
WS2 · Field Signal GenerationGenerate crop, location, season, weather, plant-health, weekly-plan and sustainability context through useful farmer interactions.
WS3 · Data Quality & GovernanceMeasure completeness, provenance, coverage, permission status, evidence maturity and quality exceptions.
WS4 · Institutional IntegrationDeliver agreed outputs through a pilot dashboard, secure export or the proposed institutional API contract.
WS5 · Business & Model ValidationTest whether institutional teams find the data operationally useful and identify which signals deserve outcome validation.
These targets make the pilot concrete enough for procurement and steering committees, while remaining negotiable. A final SOW should replace them with partner-specific thresholds.
| Category | Metric | Illustrative target | Contract note |
|---|---|---|---|
| Onboarding | Cohort mapping coverage | ≥95% of agreed pilot list successfully mapped or exception-coded | Illustrative target — calibrate with partner before contract. |
| Onboarding | Activated participants by Day 30 | ≥40% of invited participants | Illustrative target — calibrate with partner before contract. |
| Onboarding | Purpose/consent status completeness | 100% of externally shared records | Illustrative target — calibrate with partner before contract. |
| Engagement And Signal Generation | Active participant signal coverage | ≥60% have at least one usable agricultural signal during pilot | Illustrative target — calibrate with partner before contract. |
| Engagement And Signal Generation | Meaningful interactions | Target set by use case; e.g. ≥3 structured interactions per active participant/month | Illustrative target — calibrate with partner before contract. |
| Engagement And Signal Generation | Multi-period coverage | ≥50% of active participants have signals in 2+ pilot periods | Illustrative target — calibrate with partner before contract. |
| Data Quality | Core-field completeness | ≥75% on agreed institutional fields | Illustrative target — calibrate with partner before contract. |
| Data Quality | Provenance coverage | 100% of exposed fields identify source class | Illustrative target — calibrate with partner before contract. |
| Data Quality | Evidence-status coverage | 100% for sustainability/photo evidence records | Illustrative target — calibrate with partner before contract. |
| Data Quality | Data coverage reported | 100% of dashboard/API summaries include coverage metadata | Illustrative target — calibrate with partner before contract. |
| Institutional Usefulness | Analyst/underwriter usefulness | ≥70% of pilot reviewers rate agreed views as useful or very useful | Illustrative target — calibrate with partner before contract. |
| Institutional Usefulness | Workflow adoption | At least one defined institutional workflow uses pilot output repeatedly | Illustrative target — calibrate with partner before contract. |
| Institutional Usefulness | Decision-boundary compliance | No automated approve/reject/pay/certify action driven solely by Umay Ana output | Illustrative target — calibrate with partner before contract. |
| Technical | Pilot API/export availability | Target ≥99.5% during agreed monitored windows, excluding planned maintenance | Illustrative target — calibrate with partner before contract. |
| Technical | Traceability | 100% of institutional responses/export batches traceable to request/batch ID | Illustrative target — calibrate with partner before contract. |
| Technical | Critical security incidents | 0 | Illustrative target — calibrate with partner before contract. |
A pilot can succeed operationally even before enough outcome data exists to prove a predictive relationship. For example, a bank may confirm that crop/geography/season context is valuable to relationship managers and that 65% of active farmers generate usable field signals. That does not yet prove a reduction in default.
Outcome validation is a separate workstream. Where the institution lawfully provides historical or pilot-period outcomes — such as arrears, restructuring, claims, inspections or external MRV validation — selected Umay Ana signals can be tested for association, lead time, precision and incremental information value. Only after that should the parties consider predictive or scoring claims.
This sequence protects both parties from turning an attractive data story into an unvalidated risk model.
{
"pilotRef": "pilot_bank_agri_001",
"durationDays": 90,
"invitedParticipants": 10000,
"useCase": "agricultural_credit_monitoring",
"delivery": {
"farmerAccess": "institution_sponsored",
"institutionalInterface": [
"dashboard",
"api_reference"
],
"languages": [
"partner_selected"
]
},
"successFramework": {
"activation": "agreed target",
"dataCoverage": "agreed target",
"quality": "agreed target",
"institutionalUsefulness": "reviewer score + repeated workflow usage",
"outcomeValidation": "only where lawful partner outcome data exists"
},
"decisionBoundary": "Pilot measures context and operational usefulness; no autonomous credit, claim or certification decision."
}The commercial proposal should be built from participant count, analysis allowance, geography/language scope, onboarding work, dashboard requirements, API integration, support, retention/security requirements and any third-party MRV or verification services.
Production-scale commercial models can include institution-sponsored farmer access, per-active-farmer platform pricing, a portfolio/API subscription, enterprise licensing or a custom data/MRV integration. The final model should reflect who receives value and who carries the variable AI and support cost.
The goal of the pilot is not to prove every long-term promise. It is to show whether Umay Ana can create enough farmer value, structured field context and institutional utility to justify the next stage.