Built with AI
Fogos Firewatch
A wildfire alerter built for Portuguese summers. It watches active fire data and pings my phone only when a fire is close enough, and moving the right way, to actually matter.
10 min
How often it checks
Wind
Factored into each alert
Near
Distance-based warnings
1
Alert per fire, no spam
The problem
Wildfire data is public, but a public dashboard still means remembering to go and check it. What I actually wanted was narrower: a timely alert on my phone when an active fire is close enough to affect home, smoke included.
The trick is dodging both silence and noise. A plain distance check warns you about everything. A wind-only check can miss a fire right next door when the wind reading is missing or wrong.
What it does
- Checks active fire data from Fogos.pt on a schedule.
- Measures the distance from a home location I keep private.
- Folds in wind direction to colour or filter the alert, depending on the mode.
- Remembers what it has already told me, so one fire doesn't ping me again and again unless it gets worse.
- Sends the alert to my phone.
How I built it
Like the rest of these, AI wrote the code and I made the design calls. The one I thought hardest about was wind. The tool can treat it as a footnote on the alert, or as a filter that holds the alert back. I set it to fail open: warn me about a nearby fire first, then add wind context once the reading looks trustworthy, instead of going silent because a wind value happened to be missing.
It runs on a timer and keeps track of what it has already told me, so a single fire doesn't buzz me every ten minutes. It also assumes the incoming data might be wrong rather than trusting it blindly.
Made for me, maybe for everyone
Right now it's wired to my own location and alert settings, so it isn't something I can just hand over. But it would work for anyone living near fire-prone areas, so I might build a public version where you plug in your own location and API keys and point the alerts wherever you want them.