
Stop guessing when to stop work — and start again the moment it's safe.
Register your sites. We track every storm to them, and tell you when it has passed.
Team chat, webhooks, MQTT or REST — the same warning, in the channels you already run.
SEE EVERY DELIVERY CHANNEL →We track every storm cell, minute by minute, to the site it's heading for.
Every hail, lightning and heavy-rain cell, followed as it moves.
Each cell is resolved against your site's coordinates, not the grid square around it.
The site in its path gets its own warning, with time to act.
We confirm the moment it starts. We call the all clear the moment it's gone.
A warning, a confirmation, the all clear — in the channel your crew already reads.
No code. The warning and the all clear land in the channel your crew already reads.
Prepare — Hail
ETA 11:00 · in 71 min
Same messages in a server channel, with the bot tag so nobody mistakes it for a person.
Intensifying — Rain
ETA 23:10 · in 10 min
A group or a bot chat. Same words, same times, on the phone in the yard.
Detected — Lightning
Now
For crews on Workspace: a space that only ever carries warnings.
All clear — Hail
No app to add. The same warning in the inbox the whole crew already has open.
Prepare — Hail
ETA 11:00 · in 71 min
No account anywhere. The browser asks once, and the warning lands on the screen in front of them.
Detected — Lightning
Now
Minutes to wire. Any system that takes a POST gets the warning as JSON.
POST https://ops.yourcompany.com/weatherx
{
"event": "hazard.hail",
"site_id": "site-4",
"eta_minutes": 71,
"severity": "moderate"
}Broker config. Devices and SCADA subscribe to a topic per site and hazard.
topic: weatherx/site-4/alerts/hail
qos: 1 retained: false
payload: {
"eta_min": 71,
"severity": "moderate",
"action": "safe_mode"
}You call for it. Poll the active warnings for your sites.
$ curl api.weatherx.earth/v2/warnings \
-H "Authorization: Bearer $KEY" \
-d site=your-site-id
200 OK · 0 active warningsAsk an assistant in plain words. It calls WeatherX for the answer.
Any storm risk heading for my sites?
3 of your sites are in lightning danger in about 40 minutes.
Simulated exchange.
Real warnings, replayed from the production feed.
Built for operations with assets standing outside.
Modules, trackers and blade crews.
- HAIL
- LIGHTNING
Orchards, vines and open fields.
- HAIL
- RAIN
Yards, cranes and elevated work.
- LIGHTNING
- RAIN
Public-safety desks and on-air weather.
- HAIL
- LIGHTNING
- RAIN
Every warning is scored against what the sky actually did.
What these numbers beat.
Questions before you sign.
Where in the world does it work?
Minute-level warnings cover Europe (including Turkey and the UK) and the continental United States. Sites elsewhere get hourly forecasts instead. The coverage check tells you which you would get, site by site, before anything is signed.
How accurate is it, and how do you measure that?
Every warning is scored automatically against what the sky actually did — 800,000+ observation points a day, each passed through a quality-control pipeline before it is allowed to judge us, so a drifting station gets excluded rather than averaged in to flatter the numbers. On that record: 97% of lightning risk areas caught, 99.93% of dry calls verified correct, and 9 in 10 work-stopping rain events caught. The scoring runs every day, not once per press release.
What does it take to get started?
Nothing to install. Send us your site list; we run the coverage check and tell you what each one would get. Then one integration conversation, and warnings land in the channels you already run. If we cannot serve a site properly, you hear that first.
