FireWatch
AI-verified · Open source · Solar

Detect fire
before there is fire.

FireWatch smells a smouldering fire — its carbon monoxide, its hydrogen — minutes before flames become visible. Open sensor mesh, AI verification, autonomous drones.

Pre-launch

Be there when the forests get eyes and noses

  • Early access to the public demo
  • One email at launch — no spam
  • A look at the dashboard, docs & open data

Launch updates only — no spam, unsubscribe anytime.

Built on open data from Copernicus EMS / EFFIS NASA FIRMS & Worldview UFZ drought monitor Sensor.Community OpenStreetMap
The first minutes decide everything

The forest is burning faster every year

3.7Mkm²

burned worldwide in 2024 — larger than India

Carbon Brief, 2025 ↗

+60%

rise in wildfire CO₂ emissions since 2001

Swansea / UC Merced, Science 2024 ↗

€2.5B/yr

in wildfire damage across the EU

EFFIS analysis for the EU Parliament ↗

21min

for a 1-acre fire to grow to 100 acres

Lindsey FireSense ↗

These four figures describe consequences. On causes: 45 % of German wildfires trace back to people — and nobody knows the cause of 43 %.

Responsibility starts before the first spark
Smell beats sight

A smouldering fire is measurable before it is visible

Cameras and satellites only spot the smoke — often too late at night, in fog or under a dense canopy. The chemical signature of a smouldering fire (CO, H₂, VOC) is in the air minutes earlier, right at the forest floor. That is exactly where FireWatch measures.

How it works

From a tree in the forest to an alarm on a screen

Five steps between the first smouldering gas and the alert at the fire station.

01

Sense

Solar sensor nodes on trees measure combustion gases (CO, H₂, VOC), temperature and humidity around the clock.

02

Mesh

Readings hop through a self-healing LoRa mesh — and over MQTT from the gateway — to the cloud. No mobile network or power lines required.

03

Detect

The FireWatch cloud scores the risk from the raw values and classifies each node: ok → elevated → warning → alarm.

04

Verify

On an alarm, an autonomous drone launches to the node and confirms the fire with an onboard vision + thermal AI — killing false alarms. (Next stage: simulated today, the mesh works without it.)

05

Alert

Once verified, an alert with the exact location and a thermal snapshot goes out to fire services within seconds.

Who it’s for

Makers & open source

Build your own node: open hardware, open firmware, a documented API.

Research & municipalities

Open raw data and a live picture — CC-BY snapshots, filter API, EFFIS integration.

Anyone protecting forests

Join the list, sponsor a sensor later or bring FireWatch to your region.

For forest owners & municipalities

Own woodland? Prevention is funded.

Many don’t know that German states and the federal government fund wildfire prevention. We show who can apply for what — with the official contact points per state.

See the funding →
Open to the core

Open hardware, open data, accountable

Every node sends its raw readings — no black-box sensor, no vendor lock-in cloud. Here is what one telegram looks like:

Open source · MITSolarLoRa meshOffene API
POST /api/v1/ingest
{
  "node_id": "fw-node-00A7",
  "co_ppm": 41.8, "h2_ppm": 14.2, "voc_index": 213,
  "temperature": 28.4, "humidity": 31,
  "solar_charging": true, "battery_v": 4.02,
  "rssi": -94, "hops": 3
}
Not claimed — live

Our real demo nodes, right now

No screenshot, no mockup: these nodes sit in the Leipzig floodplain forest and actually transmit — environmental sensors, a soil sensor and the gateway that bundles it all onto the network. The values come live from the same open API anyone can query; the trend line shows the latest readings.

The same data, raw and unfiltered: GET /api/v1/nodes ↗

The situation, live

Dig deeper

Real-time data, open sources — on their own pages, so this one stays short.

Pre-launch

Be there when the forests get eyes and noses

Join the list and be the first to know when the open demo, the live dashboard and the docs/API go live.

Launch updates only — no spam, unsubscribe anytime.