The Kenyan AI tour — seven places you\'re already using it
1. M-Pesa fraud detection
Safaricom uses ML models that look at your transaction patterns — time of day, amount, who you send to, location. When a transaction deviates sharply from your normal patterns, the system pauses it or flags it for review. This is classic "anomaly detection" ML. It's why you sometimes get a "confirm" SMS for a transaction that would have gone through instantly before.
2. Safaricom and Kenya Power chatbots
The chatbot on Safaricom's WhatsApp business line and on Kenya Power's website use natural language AI to understand your query ("my token is not working") and route you to the right resolution. Behind the scenes: a small language model trained on thousands of past support conversations.
3. Google Maps traffic on Thika Road
Maps sees where people are moving (anonymised GPS from millions of phones) and predicts which roads are clogging up. If it says "20 min faster via Kasarani," it's a real-time AI prediction based on thousands of people's phones moving slowly on Thika Rd right now.
4. YouTube and Netflix recommendations
"Watch next" is not random. An ML model scored thousands of possible videos for you based on: what you've watched, what you stopped watching halfway, what people similar to you watched, and what time of day it is. This is why you get sucked in for 2 hours.
5. Your phone's face unlock and photo app
Your phone runs a face-recognition model on-device — never sends your face to the cloud. Your photo gallery's "People" tab also runs clustering AI to group photos by person, automatically.
6. PlantVillage — AI for Kenyan farmers
Kenyan farmers increasingly use apps like Nuru (PlantVillage) to diagnose crop diseases. You photograph a sick maize leaf, the AI identifies the disease (e.g., fall armyworm vs. maize lethal necrosis) and suggests treatment. This is image-recognition AI solving real African agriculture problems — built partly by researchers from Penn State and local African teams.
7. Hello Tractor, Apollo Agriculture, Shamba Pride
Kenyan agritech companies use AI to: predict crop yields from satellite imagery, match smallholder farmers with loans based on behavioural data, and route shared tractors to the fields that need them most.
Where AI is coming next in Kenya
- Healthcare — AI tools are starting to read chest X-rays for TB, a huge deal given radiologist shortages. Kenya has an estimated 1 radiologist per 100,000 people. AI can triage.
- Education — personalised study assistants (Khanmigo is already popular; expect Kenyan-made equivalents within 12 months).
- Legal — document review automation is arriving for Kenyan law firms. Contract drafting that took 3 hours now takes 20 minutes.
- Customer service — every medium-size Kenyan business will have an AI chatbot by 2027. The only question is whether it's useful or terrible.
The exercise that changes how you see the world
For the next 48 hours, every time you use a digital tool, ask yourself: "Is this using AI?" You'll be shocked how often the answer is yes. Try this with:
- Your bank app (categorising transactions → AI)
- Instagram / TikTok (feed order → AI)
- Google Photos search ("show me photos of Mombasa" → AI)
- Spam filter in Gmail (AI)
- Autocorrect on your keyboard (AI)
- Ride-hailing surge pricing in Nairobi (AI)
By the end of 48 hours, you'll never see "AI" as abstract again. It's infrastructure — like electricity was in 1925.