Geospatial AI · Applied space technology · Deep tech science for the Earth
One company, one stack — and the products that sit on top of it.
Geospatial AI · Applied space technology · Deep tech science for the Earth
One company, one stack — and the products that sit on top of it.
The platform
An intelligence layer that reads radar, optical and hyperspectral as one map — and a decision layer where governments, banks and growers act on what the map just said.
The living Earth, mapped by AI
The intelligence layer. Field, district, nation — with every figure traceable to a dated satellite pass.
almamaps.ai →
Scroll what's happening on the ground
The decision layer. Where a department, its growers and the finance behind them see the same record and act on it together.
almaforum.ai →
De-risking agriculture
The product suite for the agriculture economy — ministries, food security, and the finance that has to stand behind it.
Infusing life into soil
Deep tech science — recruiting the desert's own biology under the crop, so arid land holds water and carbon instead of losing both.
The model
Global Earth-observation models are trained to be adequate everywhere, and are therefore expert nowhere. We are building the opposite.
Thirty-seven wheat fields in the Sharjah desert, three seasons of satellite passes, nothing else. Trained and running on Sharjah's own compute — the model, the data and the outputs stay inside the emirate. We have found no published precedent for adapting an Earth-observation foundation model to an area this small.
See the model →
Figures shown are from the completed 2024/25 season and are provisional while the model is calibrated against ground observation.
Sovereign capability
Ten years ago a nation could buy its satellite imagery and its analysis from somewhere else and call that a space programme. That is no longer enough. The imagery is becoming commodity; the model that reads it is becoming sovereignty.
We build it where it is used.The model is trained on the country's own ground, on compute inside its own borders, by people who will still be there next season. The weights stay. The data stays. The capability stays.
And we hand over the keys.Our work is not a subscription to a black box somewhere else. It is a model your institutions own, your agronomists correct, and your auditors can open.
If you are taking your country into space, and into agentic AI on the ground — that is the road we are already on. Come and walk it with us.
Where we work
Flood extent and persistence read from radar for crop insurance on the Jamuna floodplain — through cloud, and at night.
National agriculture monitoring, scoped with the ministry across districts and protected areas.
Marine incident reconstruction off Al Hallaniyat — what was on the water, when, and what the radar saw.
How we got here
Invited to the Hand-in-Hand Initiative Investment Forum in Rome, where the problem we had been circling for years was finally stated at global scale: finance cannot reach land it cannot measure.
Working sessions with agriculture ministries across Africa, Central Asia and the Gulf, with pilots scoped on live platforms rather than proposals.
Alma Earth registered under the SPARK AI Hub at Sharjah Research Technology and Innovation Park, and MleihaEarth — our first sovereign Earth-observation model — went into training on Sharjah compute.
Tell us the decision you need to make, and we will tell you honestly whether we can measure it yet. When we cannot, we will say so.