Sophia Kim.

Creative Direction 
Experiential I Spatial I Retaill I Creative AI

Strategy 

What if AI could respond to a wildfire before we even realized it had become a disaster?

I've been thinking about a project that sits at the intersection of AI, robotics, computer vision, swarm intelligence, and spatial design. I'm calling it SKYWEAVE.

 

The question I'm exploring is:

Can AI detect a problem in the physical world, understand what's happening, predict what will happen next, and coordinate autonomous machines to respond?

​

Wildfire is my first use case.

Imagine a system that continuously analyzes:

• Satellite imagery
• Thermal and visual camera feeds
• Weather and wind conditions
• Terrain and vegetation
• Fire behavior

​

AI detects the earliest signs of a wildfire, determines its location, predicts how it may spread, and recommends an intervention.

Then a swarm of autonomous drones is deployed.

 

Not one drone.

Hundreds or potentially thousands of coordinated agents.

 

The interesting part is that the drones wouldn't simply follow predetermined paths. The AI could continuously optimize the swarm based on changing conditions:

→ Where is the fire moving?
→ Which area is most critical to protect?
→ How many drones are needed?
→ How should the swarm reorganize itself?
→ How can we minimize energy while maximizing coverage?
→ What happens if some drones fail?

​

And this is where I'm exploring an even more speculative idea.

 

What if the drones didn't need to carry water?

​

A drone already moves enormous amounts of air through its propulsion system.

Could we use that airflow as part of a miniature atmospheric water-harvesting system—cooling/dehumidifying the air, extracting existing atmospheric moisture, converting it into droplets, and releasing those droplets directly where they could have an impact?

I'm not assuming this is solved.

In fact, that's the point of the project.

 

I want to investigate the feasibility scientifically and build progressively:

AI fire detection
↓
Fire-spread prediction
↓
Digital twin simulation
↓
Autonomous swarm coordination
↓
AI resource optimization
↓
Physical drone prototype
↓
Atmospheric water-harvesting research

​

The system would also remain human-in-the-loop: AI recommends and coordinates; humans retain authorization, oversight, and the ability to intervene.

​

What excites me most isn't actually the drone.

It's the larger question:

How do we design AI systems that can perceive, reason about, and responsibly act in the physical world?

​This is the beginning of the experiment.

​