Why agriculture needs a sector evidence layer.

Walk into a programme office running a generic SME-development platform on agri-cohort intake day. The platform asks the operator to describe the value proposition, the customer segment, and the unique selling point. The operator runs a 40-hectare maize and broiler operation with three buyers, an input-credit line from the local agri-cooperative, and a cash cycle that begins each season in November.

None of the questions the platform asks are wrong in the abstract. They are simply not the questions an agri-operator needs to be asked first. The platform does not know what a planting window is. It cannot read a livestock cycle. It does not know whether broiler margins are typical or anomalous for the region. The mismatch shows up at the operator level immediately, and at the programme-impact-reporting level by the end of the season.

The scale, before the argument

The Food and Agriculture Organization of the United Nations (FAO) estimates approximately 500 million smallholder farms worldwide, supporting an estimated 2 billion people. The International Labour Organization's employment statistics consistently place agriculture as the leading employment sector across sub-Saharan Africa and much of South Asia, with women representing a substantial share of the agricultural labour force in many of those economies.

The agri-MSME finance gap, within the broader $5.2 trillion MSME shortfall estimated by the IFC, is documented in IFAD's and World Bank's rural-finance reviews as among the most severe sub-segments. Smallholder farmers are systematically under-served by formal lenders for reasons OECD's information-asymmetry diagnosis applies to in sharpened form: their evidence is more dispersed, their cash cycles more seasonal, their biological risk less legible to a generalist credit officer.

Major agri-development partners - IFAD, FAO, AGRA (Alliance for a Green Revolution in Africa), the African Development Bank's Feed Africa programme, the World Bank's agriculture practice - all operate against this backdrop. The published programme-effectiveness reviews from each repeat the same operational observation: generic instruments cannot carry the weight of agriculture-specific evidence.

The sector dimensions a generic platform misses

Agri-operators answer a different set of operating questions than generic SMEs. The most consequential ones include:

A platform that cannot capture or reason about these dimensions cannot meaningfully support an agri-operator. Agri-development programmes assume a sector-literate intervention. Generic SME tooling is not sector-literate, and the gap between what the tooling can describe and what the operator actually does is where agri-programme impact reporting falls apart.

A platform that does not know what a planting window is cannot meaningfully support a farmer. Sector literacy is not a nice-to-have. It is the precondition for usefulness.

What a sector layer looks like in operation

A sector layer extends the platform foundation (AI guidance, evidence framework, funding, programmes, learning, network) with sector-specific workflows, evidence schemas, and field operations. For agriculture, the meaningful additions are: commodity diagnostics that understand cycle and margin per commodity, structured farm visits captured in the operator's record, vet-dispatch workflows for animal-health events, production and seasonality tracking month-by-month and commodity-by-commodity, offtake and market-access records with verifiable buyer relationships, agri-funding readiness against seasonal-capital and offtake-backed lending criteria, agri-specific learning that produces applied outputs, and agri programme management that reports outcomes in donor-recognised language.

Each of these connects back to the underlying MGS framework. A farm visit is structured evidence. A vet dispatch is a closed-loop care record. A buyer offtake agreement is commercial proof. The same evidence layer reads them all, at the appropriate confidence rung.

Why this matters at the programme level

Agri-development programmes have historically struggled to report meaningful outcomes at the operator level. Workshops are delivered. Inputs are distributed. Field officers visit. None of these prove that the operator is measurably stronger at season close than at season open. AGRA's public reviews of input-distribution programmes, IFAD's evaluation of rural-finance interventions, and the African Development Bank's Feed Africa progress reporting have all noted, in different language, that operator-level outcome data remains the binding measurement constraint in the sector.

A sector-literate platform produces sector-literate outcomes: cycle yield captured against plan, post-harvest loss measured, offtake evidence accumulated, finance readiness improved per operator, commodity diagnostics updated month-by-month. The programme report describes what changed in language the donor, the lender, and the operator all understand.

Where this leads

A sector layer is not a replacement for sector specialists. The vet still treats the animal. The extension officer still runs the field visit. The buyer still negotiates the contract. The aggregator still consolidates the shipment. The platform captures the record so the work compounds across cycles and across stakeholders.

Agriculture is the lighthouse case for sector specialisation because the gap between generic SME software and sector reality is largest in agri-operations. The same architecture - framework universal, evidence schema sector-specific - extends to manufacturing, services, tourism, and other sectors. The pattern is identical. The evidence schema and the workflows differ. The Mothusi Agriculture layer is the operational answer for the sector that needs it most acutely first.