FireNow

FireNow simulates wildfires in real time, helping teams prepare, coordinate, and plan. AI uses WebMCP tools for precise advice; humans approve every action.

Comment

Inspiration

Wildfires evolve quickly, while response teams must coordinate people, resources, weather data, and evacuation decisions under pressure. I wanted to explore how AI could support incident commanders without replacing their judgment.

What it does

FireNow is a real-time wildfire simulation and decision-support web app. It helps teams visualize a scenario, simulate fire behavior, compare response strategies, prepare tactical plans, and coordinate resources.

AI can read the operational situation and use WebMCP tools to make relevant recommendations. However, it cannot directly commit a response plan: every proposed plan stays reversible until a human operator reviews and approves it.

How I built it

I built FireNow as a modern web application with React, TypeScript, MapLibre, and a local wildfire simulation engine. The interactive map displays fire progression, weather conditions, available units, threatened areas, and proposed interventions.

I created 21 WebMCP tools with structured JSON inputs and outputs. These tools let an AI agent inspect the situation, list resources, run simulations, compare plans, prepare actions, and request human approval through commit_plan. Plans can also be cancelled with revert_plan.

Challenges I faced

The main challenge was designing a useful human-AI workflow without giving AI uncontrolled operational authority. I solved this with a staged-plan model: AI can analyze, simulate, and prepare actions, but a human must validate the final plan.

Another challenge was making the WebMCP integration clear and robust across different browser environments. I focused on precise tool schemas, structured responses, meaningful errors, and a visible interface that explains the current tool status.

What I learned

I learned that the most valuable AI experiences are not fully autonomous. In high-stakes situations, AI should make complex information easier to understand, compare options, and reduce workload—while humans remain accountable for final decisions.

Built With

Share this project:

Updates

Submission history