Farming has always been a game of timing. Plant too early and a late frost wipes out your seedlings. Sell too late and the market price has already dropped. For smallholder farmers across South Africa, these decisions are made with little more than instinct and word of mouth, because the tools that could help are often too expensive, too complicated, or too dependent on a connection that isn't there.
Why Khula
Khula started with a simple question: what does a farmer actually need at the moment they need it? The answer turned out to be four things, always.
- Weather they can trust, both current conditions and a real forecast.
- Market prices so they know what their crop is worth today.
- A planting calendar so they know what to put in the ground and when.
- A calculator to work out resources and costs before committing.
On top of that, we layered AI. Farmers can snap a photo of a crop and ask Khula to analyse it, or load their data and get a plain-language read on what it means. No jargon, no training manual.
Built for low-bandwidth reality
Most apps assume a fast, always-on connection. The field doesn't work that way. We designed Khula to be lightweight and forgiving, loading fast and degrading gracefully when the network is weak. The interface is deliberately simple, with large targets and clear language, because the best tool is the one someone actually uses without thinking about it.
What we learned
The biggest lesson was that AI is only valuable if it's practical. A model that recommends fertiliser ratios is useless if a farmer can't reach it, afford it, or understand it. So we spent more time on the question than the model, and kept the experience focused on the four things farmers ask for every day.
Khula is currently available to our testers for full access. We're continuing to add tools and refine the AI based on what real users tell us, because that's the only way software in this space gets better.