Mkulima AI

Agriculture accounts for roughly 30% of global greenhouse gas emissions, yet smallholder farmers — who produce over 70% of food in sub-Saharan Africa — are also the people most devastated by the climate change they did not cause. In Kenya, erratic rainfall, prolonged droughts, and soil degradation are already cutting yields by up to 20% per season. The cruel irony is that these same farmers hold enormous potential to reverse the damage. Regenerative farming practices like agroforestry, no-till, and cover cropping can dramatically increase soil carbon storage — but over 7 million Kenyan smallholders have no access to the knowledge, tools, or markets needed to adopt them. They face three compounding barriers: • No climate guidance: farmers cannot access personalised advice on climate-resilient crops, planting calendars aligned to shifting rainfall, or soil health practices — in a language they speak. • No carbon market access: sustainable farmers sequester real carbon but earn nothing from it because carbon credit certification is complex, expensive, and designed for large enterprises. • No early warning systems: crop disease outbreaks and livestock emergencies caused by climate stress go undiagnosed until it is too late, wiping out livelihoods. Without intervention, climate change will push millions of smallholder farming families deeper into poverty while a massive natural carbon sink goes untapped.

Mkulima AI is a multilingual smart farming platform that gives Kenyan smallholder farmers an AI assistant in their own language. It has seven integrated modules: 1. AI Chat Assistant — Farmers ask questions via text, voice note, or photo in English, Swahili, or Kikuyu. The app transcribes speech, translates it, queries Llama 3.1 via Groq, and reads the answer back aloud using gTTS — accessible even for low-literacy users. 2. Crop Disease Scanner — Farmers photograph a sick plant and the vision module identifies the disease (e.g. maize leaf blight, bean rust, coffee berry disease) and returns treatment steps in the farmer's language immediately. 3. Farm Map and Carbon Audit — GPS location provides real-time local weather. Farmers input farm size, soil type, farming practice, and crops to get estimated carbon sequestration (tCO2e/year), potential earnings in USD and KES from Verra VCS, Gold Standard, or Kenya National carbon credit schemes, and a step-by-step certification checklist. 4. Carbon Progress Dashboard — A time-series chart tracks estimated carbon sequestration over months as farmers log practice changes, making the impact of switching to regenerative farming visible and motivating. 5. Market Price Tracker — Daily commodity prices (maize, beans, tea) from Kenya's AMIS are surfaced in chat, so farmers can ask "what is maize selling for today in Nakuru?" before going to market. 6. Personalised Planting Calendar — Using captured GPS coordinates and the farmer's selected crops, the app generates a planting and harvesting calendar based on local rainfall patterns and season, reducing crop failure from mistimed planting. 7. Veterinary Support — Finds nearby vet clinics with ratings, hours, emergency contacts, and WhatsApp links using GPS. Includes emergency hotlines and condition-specific triage guides (cattle, poultry, sheep/goats). Built with: Streamlit, Groq (Llama 3.1), gTTS, Python, Docker. Offline-capable via PWA caching for low-connectivity rural areas.