In private beta — not open yet

Aviation weather is written for cruise.
You don't fly there.

A forecast that's clear and smooth at eight thousand feet tells you nothing about the ridge you're crossing at four hundred. The weather that decides a rotorcraft day — mechanical turbulence, rotor off the lee slope, valley fog that burns off two hours later than the field — happens in a layer most aviation weather barely resolves.

Weather in the app today Where this is

METAR, TAF, winds aloft, radar and G-AIRMETs already ship in the app.

The gap

The layer you work in is the layer nobody forecasts.

What you get today
  • Winds aloft starting at 3,000 ft — above most of your working day
  • A field METAR eleven miles from the ridge that matters
  • Turbulence forecast as a region the size of a state
  • Nothing about rotor, lee waves or valley channelling
  • A TAF for an airport, when you're going to a pad
  • Judgement calls made from experience and a phone call
What we want to build
  • Wind and turbulence modelled at 500 and 1,000 ft AGL along your route
  • Terrain-aware: which side of the ridge, and how bad
  • Rotor and mountain-wave risk flagged where terrain and wind combine
  • Conditions at the destination, not the nearest airport
  • Go / marginal / no-go per leg, with the reason stated
  • The same picture for the pilot and the operations desk
Why this is plausible for us

We already model the terrain.

Route planning in the app is terrain-aware today — it knows the elevation profile along your route and already derives a turbulence estimate from terrain and wind. Weather isn't a new discipline for us so much as a deepening of something that's shipping.

Terrain, already loaded

Elevation along any route, at the resolution the planner uses now.

Routes, already known

Forecasting a corridor is far more tractable than forecasting everywhere.

Tracks, already recorded

Every flight is a measurement. Actual wind and ride quality, at altitude, over time — the ground truth to check a model against.

The part that isn't engineering

The hard problem here is licensing.

Building the model is the cheap half. Aviation-grade weather — vendor radar, lightning, satellite, high-resolution model output — is commercially licensed and priced for companies with revenue. The free and open sources that make a prototype possible often carry terms that don't permit a commercial product.

So the beta is built on sources whose terms permit it, and the licence gets scoped before each next layer rather than after. That is the right order, and it is why some of what you would expect to see is not in it yet.

If you operate somewhere the standard products serve badly — mountain work, coastal, anywhere the valley does its own thing — that's exactly the input that would shape this. We'd genuinely like to hear it.

It's built. It isn't open yet.

Radar, satellite, echo tops, turbulence down to a hundred feet, gridded flight category, icing and the GFA layers all run today — as a private beta, on our own machines. What stands between that and a link you can open is hosting: the low-level layers are rendered from model files by a tile server, and that has to run somewhere before anyone else can reach it. We would rather say that than put a button here that opens nothing. The weather in the app today — METAR, TAF, winds aloft, radar, G-AIRMETs, decoded and tied to each leg — ships now.