Divisi
Our Mission
Know What to Grow,
When and How.
Divisi gives an ambitious smallholder farmer the "superpower" to manage a farm like a Pro: personalised agronomy for one real field, where there is no agronomist or extension officer to ask. It is built to cut the losses a season brings and to make a good yield repeatable. Here is why we built it, what it changes, and how we intend to prove it.
What does "Divisi" mean?
Divisi (dee-VEE-see) is a Shona word for a charm believed to protect a farmer's land and multiply the harvest — a force of abundance and protection. We took the name for something you can check: published data, run on the field you pinned.
Why We Built Divisi
There Was Never Enough Agronomy to Go Around
The training exists; it is simply not where the fields are. One extension officer in Malawi serves more than 4,000 farmers, roughly eight times the maximum the FAO recommends. Generic crop guides are written for estates rather than for 0.4 ha (1 acre) plots, and most digital tools assume reliable internet and a high-end phone. So the farms growing about half of Sub-Saharan Africa's food calories carry the most headroom and the least advice.
Divisi closes that distance, and closes it in a way that leaves the farmer in charge: real agronomy for your own field, free forever, on the phone already in your pocket, with or without network. Not a subscription and not a service you wait on — the knowledge itself, held by the person working the land.
We no longer repeat the claim that smallholders grow 70% of Africa's food. Farms under 5 hectares produce about half of its food calories — the smaller figure, and the true one. Our World in Data on the 70% figure
Theory of Change
The Same Field, a Different Harvest
A smallholder farmer in Sub-Saharan Africa often harvests a fraction of what the same field could produce. The reasons are rarely mysterious: a crop that never suited that soil, a planting date missed because the rains came late, fertiliser applied at the wrong stage, a pest found too late, a storm nobody saw coming.
Each of those is a decision made without information, and the pattern the rains once followed is shifting, so last season's date is a weaker guide every year. The information exists in soil and climate datasets, in agronomic literature and in weather models — but it has never been delivered to the person holding the hoe, for their specific field, on the phone they already own.
That is measured, not asserted. Across 14,773 smallholder fields, a 2024 paper in Nature Communications found management alone — cultivar, nutrient rate and placement, plant density, pest control, earlier sowing — lifts maize from about 2 t/ha to 4.3. No new land, no new technology. Different decisions, better timed. Nature Communications, 2024
That is Divisi's whole theory of change. Put the right decision in front of the right farmer at the right moment, free, and the harvest follows. Behind the harvest is the thing we are actually building for: a household that earns steadily, season after season, off its own land.
Start With a Crop That Fits the Field
A crop that never suited the field is a loss you carry all season. Divisi ranks crops on the soil, terrain and climate of your actual field — the FAO's GAEZ framework run at the point you pinned — so two fields in one district get different answers.
Plant With the Season You Actually Get
Rains that arrive late, a dry spell mid-season, a storm at flowering: a planting date that worked for years can miss now. Divisi sets the planting window against the season actually happening and watches a 7-day forecast for your field, so the warning arrives before the weather does.
Know What to Do Next, and Why
Losses pile up between planting and storage: top-dressing two weeks late, a pest found after it has fed, grain stored damp. Divisi gives the task due now, why it matters and how to do it, stage by stage — plus a diary and ledger that outlive any exercise book, with photos you can review years later.
who this is also for
If You're the One Funding It
A great many farms in this region are paid for by somebody who is not on them. A salary in the city covers the fertiliser. A transfer from another country covers the labour. The ground is family ground, or rented, and it is worked by a brother, a manager, or whoever is there — while the person carrying the cost sees it a few times a year.
That person makes the same decisions as the farmer holding the hoe, with less to go on. Everything reaches them second-hand: a phone call, a photograph of something green, a figure at harvest that cannot be checked against anything. They are not short of commitment or money. They are short of ground truth.
One record serves both. The person doing the work marks each job done — their name, the date, a photograph taken at the time. The person paying for it reads the season as it happens instead of hearing about it afterwards, and sees what each crop cost and earned.
It works the other way too, which is the part we care about most: work that used to vanish into a phone call finally shows up somewhere it counts, attached to the name of the person who did it. Running a farm you're not standing on →
SDG Alignment
Where Divisi Sits in the Global Goals
Two goals are the core of the work, and a third follows directly from them. Each card names the specific target, not just the goal number.
Zero Hunger
Targets 2.3 — double the productivity and incomes of small-scale food producers · 2.4 — resilient agricultural practices
This is the goal Divisi was built for. Crop suitability is scored from the soil, terrain and climate of one farmer's actual field, so the season starts with a crop that can succeed there rather than one that was popular at the market. From planting onward, every stage carries the specific task, the timing and the reason behind it — the agronomy an extension officer would give, at a scale extension has never been able to reach. The aim is fewer losses along the way, and a yield the farm can repeat.
Climate Action
Target 13.1 — strengthen resilience and adaptive capacity to climate-related hazards
Climate change reaches a farmer as uncertainty: what to grow, when to plant, and what next week will bring. Adaptation, in practice, is a run of small timely decisions. Divisi watches the forecast for one field and speaks up before frost, a heat spike, a storm or heavy rain arrives, and as soon as a dry spell sets in, with the steps that fit the crop and its current growth stage. Planting windows shift with the season that is actually happening, not the one the calendar remembers. Saving a crop that is already in the ground is worth as much as growing a bigger one, and it costs the farmer far less.
Decent Work & Economic Growth
Target 8.2 — higher productivity through technological upgrading
Farming becomes a business when it is measured like one. A season of diary entries and ledger records is also the first financial record many smallholders have ever held — the kind of evidence a buyer, a co-operative or a lender will actually look at. It is also what the aggregator, processor or exporter at the other end needs: a supply base that plans the crop, the date and the work, instead of a volume nobody can see until it arrives.
Our Principles
You End Up in Charge of Your Own Season
Each of these follows from the same commitment: the farmer ends up in charge. A season should leave them with more skill, better records and more of their own money — not a deeper dependence on us.
Built for Your Farm, Not Everyone Else's
A national average doesn't help a farmer in Mashonaland East. Which crops to grow, when the window opens and what the weather is about to do are all worked out for your actual field, its real climate and the season actually happening.
Works Where the Signal Does Not
Records, guides and the last forecast Divisi fetched stay on the phone, and anything you write is saved there first and synced when the network comes back. Advice that stops when the data bundle does is advice you cannot use standing in a field.
Farmers Are Not the Product
Your records are yours. We do not sell farm data — not field boundaries, not yields, not to anyone. There is a version of this business that does, and taking it would make every other claim on this page worthless.
Nobody Is Priced Out
Every feature, for every farmer — no tier, no trial clock. Good farming advice shouldn't depend on income, so the organisations that run programmes on Divisi pay for it, and farmers never do.
How We Measure
What We Count, and What We're Setting Out to Prove
What we measure today
Things Divisi observes directly, and can therefore stand behind:
- Farms and fields with a location-specific crop recommendation
- Seasons planned, and tasks completed against those plans
- Extreme-weather warnings delivered before the event arrived
- Diary and ledger records kept — a farm's own evidence base
- Questions asked and answered in the Farmer Hive
- Countries, languages and crops covered
An open invitation
We are looking for a research partner to prove — or disprove — that this works. Divisi publishes no yield or income uplift figures yet, because the evidence worth having is the kind that survives peer review: a randomised trial, on real farms, against a comparison group. Uplift claims are easy to make and hard to check, so we would rather publish a number somebody else has verified than one we assembled ourselves.
So we are actively seeking a research institution, NGO, development agency or commercial partner to design and run that trial with us. We will open the methodology, the parameter tables and the anonymised field data to make it possible, and we will publish what it finds either way — because a result that sharpens the model is worth as much to farmers as one that flatters it.
Running a programme, an outgrower scheme or a supply base that needs measured, farm-level evidence? That is what Divisi Partners produces: field records captured by members as they farm, with their consent, rather than recall surveys collected months later. Where a crop was planted on a Divisi recommendation, its field also carries a projection of what it is on course to deliver, updated as the weather is confirmed and the work is logged. Explore Divisi Partners →
Impact Questions
What Funders and Partners Ask First
What a research, development or commercial partner needs to know about impact before anything else.
Data Sources
Where the Numbers Come From
Divisi holds no soil survey, no weather station and no map of its own. Every crop score, planting window and yield figure is computed from published datasets read at the coordinates of one field, and every district and ward we group a report by is somebody else's boundary. Each is credited below with its publisher, its licence and the citation that publisher asks for.
Soil, terrain and climate
Read at the pinned coordinates of a field, and mirrored so the app can reach them offline.
-
Global Agro-Ecological Zones v5 — elevation and slope class
FAO & IIASA. 2025. Global Agro-Ecological Zones version 5 (GAEZ v5): elevation and median slope class. Rome and Laxenburg. Licence: CC-BY-4.0.
-
iSDAsoil extractable soil nutrients, 30 m, Africa
Miller, M.A.E., Shepherd, K.D., Kisitu, B. & Collinson, J. 2021. iSDAsoil: the first continent-scale soil property map at 30 m resolution provides a soil information revolution for Africa. PLOS Biology 19(11): e3001441. https://doi.org/10.1371/journal.pbio.3001441
-
SoilGrids 2.0 — global soil property predictions, 1 km aggregate
Poggio, L., de Sousa, L.M., Batjes, N.H., Heuvelink, G.B.M., Kempen, B., Ribeiro, E. and Rossiter, D. 2021. SoilGrids 2.0: producing soil information for the globe with quantified spatial uncertainty. SOIL 7: 217-240. https://doi.org/10.5194/soil-7-217-2021
CC-BY-4.0 covers ISRIC's derived predictions, which is what this mirrors. It does NOT extend to WoSIS profile data or to ISRIC datasets generally — check the Data and Software Policy per dataset.
-
TerraClimate 1991–2020 climate normals
Abatzoglou, J.T., Dobrowski, S.Z., Parks, S.A. & Hegewisch, K.C. 2018. TerraClimate, a high-resolution global dataset of monthly climate and climatic water balance from 1958–2015. Scientific Data 5: 170191. https://doi.org/10.1038/sdata.2017.191
-
NMME seasonal forecast anomalies, ensemble mean
Kirtman, B.P., Min, D., Infanti, J.M. et al. 2014. The North American Multimodel Ensemble: phase-1 seasonal-to-interannual prediction; phase-2 toward developing intraseasonal prediction. Bulletin of the American Meteorological Society 95(4): 585–601. https://doi.org/10.1175/BAMS-D-12-00050.1 — the NMME project and its data dissemination are supported by NOAA, NSF, NASA and DOE.
Checked against, not computed from
Divisi works out which agro-ecological zone a place sits in from its own rainfall and temperature. This is the publisher's own zone map, mirrored so we can count how often the two disagree. Nothing on a place page is computed from it.
-
Global Agro-Ecological Zones v5 — AEZ classification, 33 classes, ~1 km
FAO & IIASA. 2025. Global Agro-ecological Zoning version 5 (GAEZ v5) Model Documentation. Rome and Laxenburg. https://github.com/un-fao/gaezv5/wiki Licence: CC-BY-4.0.
Map boundaries
The provinces, districts and wards a place page is written about and a partner report is grouped by. Both sets are assembled country by country, so the terms on them are not the same in every country — what is noted under each is the part the headline licence leaves out.
-
OCHA Common Operational Datasets — administrative boundaries (COD-AB)
UN OCHA Field Information Services Section. Subnational Administrative Boundaries (COD-AB). Humanitarian Data Exchange. Licence: CC BY 3.0 IGO. Boundaries are produced by each country's own mapping or statistical authority and endorsed for operational use; Divisi reprojects them to EPSG:4326 and does not otherwise alter them.
CC BY 3.0 IGO asks that the creator be named, and for these datasets that is the national authority behind each country's boundaries — ZIMSTAT for Zimbabwe, the IEBC for Kenya, Ghana Statistical Services for Ghana — 42 distinct authorities across 45 African countries. OCHA publishes and endorses them; it did not draw them.
-
geoBoundaries 6.0.0 (gbOpen) — administrative boundaries
Runfola, D., Anderson, A., Baier, H. et al. 2020. geoBoundaries: A global database of political administrative boundaries. PLoS ONE 15(4): e0231866. https://doi.org/10.1371/journal.pone.0231866
gbOpen carries each contributing source's own licence per country and per level, so this is the licence on the release rather than on every layer in it. Seventy of the 200 African layers are ODbL 1.0 or CC BY-SA, across 34 countries, and most are derived from OpenStreetMap. Check the boundaryLicense field for the country and level you use.
Named here because it is owed, not because it is decoration: none of these publishers has reviewed or endorsed Divisi, and any error in what Divisi computes from their data is ours. The boundaries are the ones humanitarian and statistical agencies work from, drawn for operational use — they are not a legal record of any border, and the names and designations on them imply no position by Divisi or by the publishers on the status of any territory.
Manage your farm like a Pro
Advice for your own field, in your pocket, working offline on the phone you already own, in any country you farm in. Know what to grow, when and how, and keep more of what the season gives you.
