The AI that maps the floods: How SatGPT is building Asia-Pacific’s disaster resilience

[...]

2. We often hear about getting tech “to the last mile.” Picture a rural community leader with a simple smartphone. How does SatGPT’s insight practically reach and help them make a life-saving decision?

SatGPT’s strength lies in enhancing historical risk knowledge. It’s not designed to predict the next disaster, but rather to help communities prepare more effectively for it. For instance, when a rural leader needs to decide whether to evacuate ahead of a flood, she will still rely on early warnings from national meteorological services. What SatGPT can do is support smarter ex-ante planning—so that when early warning information arrives, the community is ready to respond quickly. This includes decisions on where to build shelters, how to lay out evacuation routes, and where to preposition relief supplies. These are all critical elements that must be in place to help avert disasters, as consistently demonstrated in the cyclone response histories of India and Bangladesh.

3. Floods are an urgent threat, but what about slower crises like droughts? Is the vision for SatGPT to eventually help with these less visible, but equally devastating, disasters?

Kareff: ESCAP coordinates the long-standing Regional Drought Mechanism, which has been supporting drought-prone countries in gaining access to satellite data, products, tools, and technical expertise—everything they need to conduct drought monitoring and impact assessments more effectively. Our support goes beyond making data available—we work with countries and partners to strengthen institutions and capacities, converting these data into actionable analytics and insights. We are currently working with three Central Asian countries in establishing their own Earth observation-based agricultural drought monitoring systems.

[...]

6. The region is innovating fast, with countries like Indonesia and Thailand building their own systems. How does SatGPT aim to be a good teammate and connect with these national efforts, rather than just adding another tool to the pile?

Kareff: That’s a good point. And beyond technological innovation, we’re also seeing progress in policy and institutional innovations being put in place. Our intention is not to replace national systems, but to show what’s possible when you make risk knowledge accessible and actionable. We work closely with our national counterparts with a focus on integrating SatGPT insights into existing workflows and systems-not reinventing them.

[...]

8. Looking ahead a year, what would a “win” for SatGPT look like on the ground? Is it a specific number of communities better protected, or a faster warning time?

Kareff: Forecasting and enhancing the forecast lead times remains the responsibility of mandated early warning agencies. SatGPT is well-positioned to support efforts to protect more communities. By enhancing the historical understanding of floods, it can help improve the accuracy of early warning information, help communities proactively plan their response, and reduce disaster risk ex-ante. In that sense, I would say that effective SatGPT roll-out would amount to both gains in space and time – more communities being warned with improved lead times for mitigative response with more reliable historical data for granular risk characterization.