By Dr. Kidiga (Buni Banda) & Dr. Fred Mutisya (Kenya Medical Association)

Introduction

In rural Kenya, climate change doesn’t arrive as an abstract future. It arrives as floods that wash away homes, droughts that decimate crops, and diseases that spread when water becomes scarce or contaminated. For communities like those in Siaya County, the urgency of resilience is not theoretical — it’s personal.

Why We’re Rethinking AI for Climate Resilience

At Buni Banda, we’ve spent the past two years deeply embedded in the intersections of climate adaptation and public health. Our thinking is evolving — not away from the community, but deeper into it.

We’re asking: What happens when artificial intelligence is guided by local knowledge, not divorced from it?

The Limits of Traditional Systems

For years, early warning systems have relied on static models, top-down alerts, and siloed data. While they’ve played a role, they’ve often fallen short — too slow, too generic, or too disconnected from people’s realities.

Many communities receive alerts too late. Others never receive them at all.

And yet, the data exists: satellite feeds, health surveillance systems, indigenous knowledge, and local indicators.
What’s missing? The intelligence to weave it all together — and the design principles to make that intelligence trusted, useful, and actionable at the grassroots level.

A New Vision for Early Warning Systems

We believe the next generation of early warning systems must do three things:
1. Predict with context — not just forecast floods, but identify which households are most vulnerable, and why.
2. Communicate with empathy — in languages people speak, through channels they use, in formats they trust.
3. Learn from behavior — not just send alerts, but adapt based on how people respond.

We see AI not as a distant algorithm but as a quiet partner — embedded in the rhythms of real life, one that listens, learns, and evolves with the community.

Why Co-Creation Still Matters

If the future is AI-enhanced, it must also be human-informed.

That’s why we’re deeply committed to co-creation, even in the context of AI. We’re not just building models — we’re sitting with mothers, youths, community health workers, and elders, asking:
• What does a useful alert look like to you?
• When do you trust it?
• What would make it more helpful?

These aren’t cosmetic questions. They are the foundation of good design — and good AI.

The Power of a Simple Message

Imagine this: a woman receives a text message in Luo, warning her to move her livestock to higher ground before a storm hits.
The message isn’t generic — it’s tailored to her village, written in her dialect, and timed based on her past responsiveness.

It’s not just an alert. It’s a lifeline.

What We’re Exploring Next

We’re currently prototyping and testing:
• Predictive risk mapping powered by satellite and local data
• Multilingual SMS and voice-based alert systems
• Wearable tech for community-based health surveillance
• Adaptive delivery models that evolve based on user response

All built around three core values:
🔗 Inclusion | 🧑🏽 Participation | 🤝 Shared Ownership

An Invitation

We don’t claim to have all the answers.
But we are committed to asking the right questions — the ones that matter to people on the ground.

And we’re inviting others — technologists, researchers, governments, and communities — to ask them with us.

If you’re working at the intersection of AI, climate, and human dignity, let’s connect.
Let’s imagine — and build — what’s possible, together.