Agricultural intelligence · Portugal
Siren predicts plant disease and water stress before the crop shows them, and tells growers when to act. Our tools are trained on nine crops, tested on our own land and now running on a client farm.
First client pilot
running since October 2026
9 crops
Plantologist and Pluvora, trained on the same nine
90.3%
top-1 disease detection across 64 classes
2 signed partners
Universidade Politécnica de Viseu · ViTEB, Visabeira's testbed
Finalist
AI Summit Barcelona 2026
Four things that hold for every crop and every client.
Somebody walks out and looks. Every alert gets checked in the field, blind, and so do places the model said were fine, because otherwise you never find out what it missed. Each model is scored on both.
A satellite pixel covers several plants and the soil between them, so a block average quietly hides the one tree that is dying. We keep a record for each plant instead: where it is, how it has done, what has been done to it.
We can tell you a surprising amount about a farm before setting foot on it, from public satellite, terrain and weather records. Then we come out, and field checks, drones and sensors fill in everything open data cannot see.
We own our land in Tondela, which means we can afford to be wrong on it. Every new tool and sensor runs there first, at our own cost, before it goes near a client's crop.
Five layers, all joined on the same plant, season after season. The join is the point: any one of them on its own tells you very little.
Genetic profiling where it changes a decision: variety and rootstock identity for growers, whole genomes on our research crop.
Moisture, temperature and drainage, logged through the season and modelled from the terrain.
Satellite, drone and ground imaging of vigour, stress and phenology, plant by plant.
On-site and regional weather with forecasts, so alerts arrive before conditions turn.
Lab confirmation of disease, and assays at harvest, closing the loop between prediction and outcome.
Our research farm is where the technology is tested. The platform is what growers use.
Research farm · our land
Our own land in Tondela, where tools and sensors get tried before anyone else has to live with them. Once licensed, its outdoor medical cannabis crop funds the science and produces records no one can buy.
Platform · running since October 2026
The same tools, pointed at other growers' crops: farm records, irrigation intelligence and disease caught early. It started with our first client pilot.
Once the licence comes through, what the farm sells pays the engineers.
Every season on our own land adds labelled records nobody can sell us.
Our land and our clients' farms are what the models get measured against, and where they get proven wrong.
Sensors and machines take their chances on our crop before they take chances on anyone else's.
Updated October 2026
Growers first, and everyone who helps us get there.
Start with a short pilot: a first map of your farm from open data, checked on the ground, with our tools running on your land. You see what we get right and what we miss, then we decide together whether to run a full season.
We are building a genetic and environmental record that tracks genotype, phenotype and conditions plant by plant. We welcome research collaborations, joint programmes and field trials under formal agreement.
We are building the intelligence layer for high-value crops, with a client pilot running and our own land to test on. Write to us for the deck.
Once our INFARMED licence is issued, our research farm will grow outdoor medical cannabis flower to EU-GACP, aiming at EU-GMP compatibility through controlled post-harvest curing. Talk to us about supply.