The problem
If you are responsible for something that sits in a real place (a site, a route, a crew, a home), the warning signs already exist. USGS publishes every earthquake, NOAA and the National Weather Service publish every watch and warning, NASA FIRMS publishes active-fire detections, Blitzortung tracks lightning in near real time. The data is public, authoritative, and free.
The problem is not the data. The problem is that it lives in four separate places, none of which knows where your fence is, and none of which will wake you up at 3am because a fire detection just appeared eight miles upwind of the yard. So somebody ends up refreshing dashboards, or nobody does, and the warning that was sitting right there arrives as a phone call after it already mattered.
How it works
You draw a geofence around what you care about. AYA does the rest, on a loop:
- pulls the four feeds directly (USGS earthquakes, Blitzortung lightning, NASA FIRMS active fire, NOAA/NWS weather alerts),
- merges and de-duplicates them into one event set (so a storm that shows up in two feeds is one event, not two),
- keeps only the events that actually intersect your fence,
- scores each one by severity and drops anything below the threshold you set,
- dispatches an alert to the channel you already use (email, push, SMS, webhook, Slack).
That whole path is one composition. Adding a new region, or a new channel, is configuration, not a new integration project.
What you get
- One geofenced watch across four hazard types, not four browser tabs.
- Alerts filtered to your area and your severity threshold, so the signal-to-noise is set by you, not by a national feed.
- Delivery to a channel you already read.
- A pipeline built from parts that already exist and were verified before this went out (the four feeds, the geofence check, and the multi-channel dispatch are all pre-built patterns; the product is the wiring that binds them).
Who it is for
Operations, safety, and facilities teams responsible for fixed assets in places where the ground shakes, the wind turns, the sky lights up, or the hills burn. Emergency-management and community-resilience organizers who need one multi-hazard watch over a defined area without stitching APIs together. It augments official alerting for people who already act on hazards, and it is honestly not built to be your certified life-safety system of record (that authority stays with NWS and USGS).
Pricing
Prices below are starting hypotheses, not commitments, and we would rather earn the next rung than assume it.
- Watch (about $49/geofence/month): one fence, all four feeds, one channel.
- Metered (about $29/active geofence/month): many fences and channels, once the first watch has proven itself.
- Platform (negotiated): license the pipeline into your own dispatch stack, your channels, your SLAs.
The feed data is free to pull. What you are paying for is the composition, the geofence intelligence, and dispatch you can rely on.
Proof (and what we are dogfooding)
This product is a composition, so we built it the way we tell you it works: the deliverable was authored as data (not code), then quality-scored, and its eight referenced patterns were checked to resolve on disk before this page existed. The build left a receipt. That same watch-filter-alert spine is producer-agnostic about its feeds, which means AYA can point it at its own operational signals (build, deploy, verify) and alert the founder the moment something crosses a line, so we run the watch we sell.
Honest note
This is a Phase-0 draft, and here is what is true and what is not yet.
What is real: the four feed operations, the geofence check, the alert creation, and the multi-channel dispatch span all already exist as verified patterns, and the new early-warning workflow that binds them is authored, quality-scored, and its references resolve.
What is not yet proven: it has not been run against the live feeds, a real geofence, or a live channel. Feed execution depends on the HTTP and websocket runtime handlers being wired (an existing-substrate concern, not new code for this product), and outward dispatch is deliberately fail-closed, so nothing sends live until an authorized person turns it on. No live API calls, no spend, and no alerts were sent to build this. And, plainly, this is not a replacement for official government alerting or a certified life-safety system.
*This page is a specification. The capability it describes is not built yet, and nothing here is a claim that it runs today.*