Kenya has 47 counties. Each one has a government, a budget, a set of environmental conditions, a food security profile, and a population navigating decisions with incomplete information. Each one is, in principle, a data source.
In practice, the data is scattered across seventeen different government portals, four different NGO data platforms, two donor-funded open data initiatives that may or may not still be maintained, and a set of PDF reports published quarterly that nobody has ever parsed.
The 47-county intelligence layer is the attempt to change that — not by building a data warehouse, but by building a set of small, focused tools that each answer a specific question for a specific user.
The tools, by question
What is the drought risk in my county right now? — WapiMaji ingests NDMA drought early warning data and SPI (Standardized Precipitation Index) calculations, maps them to county boundaries, and sends SMS alerts to NGO administrators in the affected counties. Bilingual: English and Kiswahili. Zero internet required to receive the alert.
Is there a flood risk in my neighbourhood? — Mafuriko combines elevation data, drainage infrastructure records, and rainfall forecasts to produce urban flood risk maps at the sub-county level. Built initially for Nairobi, extensible to any Kenyan city with OpenStreetMap coverage.
What is my county spending my money on? — Hesabu parses Controller of Budget reports and makes county expenditure data searchable and comparable. The COB publishes this data. Almost nobody reads it. Hesabu makes it readable.
What did my MP vote for last session? — Macho-ya-Wananchi tracks National Assembly voting records, bill progression, and committee assignments. Named from the Kiswahili word for "verify" — the tool's core function is making claims about parliamentary behaviour verifiable.
Where should I save? — ChaguaSacco pulls licensed SACCO data from SASRA and maps deposit rates, loan rates, and regulatory standing by county. Most Kenyans who save in a SACCO chose it through word of mouth. Chagua gives them a second opinion.
What is maize selling for in Eldoret today? — JuaMazao tracks crop prices across major markets using WFP VAM data. The price gap between Eldoret and Nairobi for the same commodity on the same day is frequently 40%. Most smallholders don't know this when they sell.
What connects them
Every tool in this layer shares three design constraints.
First, the underlying data must be public. Nothing here requires a data-sharing agreement or a subscription. NDMA publishes drought bulletins. COB publishes budget reports. SASRA publishes SACCO registers. The work is parsing, normalising, and surfacing — not acquiring.
Second, the output must be actionable without a laptop. SMS and USSD are the delivery layer for users who need the information most. A farmer in Baringo County with a 2G phone and a crop to sell should be able to receive a Mazao price alert. A community health worker in Turkana should be able to receive a drought risk update in Kiswahili.
Third, the tools must label uncertainty. Drought risk indices are probabilistic. Budget absorption figures lag by two months. Food price data has known gaps in pastoralist counties. Every output carries a data confidence indicator. The tools are useful precisely because they are honest about what they don't know.
What's not there yet
Health facility data. School performance data. Land transaction records. Water point functionality status. All of these have open data sources in Kenya. None of them have been integrated yet. The layer is partial — intentionally so. Better to have six well-built tools than sixteen poorly maintained ones.
The county intelligence layer is not a platform. It's a practice — the practice of taking public data seriously enough to build infrastructure around it.
All tools: github.com/gabrielmahia
Responses