On Earth Day 2026, the Allen Institute for AI published a retrospective marking what it describes as a decade of deploying real-time intelligence tools for conservation. The Ai2 post covers three products — EarthRanger, Skylight, and OlmoEarth — and attributes the original commitment to Ai2 founder Paul Allen, whose stated conviction was that people working at the front lines of conservation deserved better information.
EarthRanger
EarthRanger is Ai2’s platform for coordinating on-the-ground wildlife and ecosystem protection. According to the post, it now operates across more than 900 protected areas in 95 countries.
The post describes a deployment in northern Thailand where wild elephants emerging from protected areas into durian, jackfruit, pineapple, and cassava farms create conflict with farming communities. A programme led by Thailand’s Department of National Parks, conservation partners, and local communities uses EarthRanger to coordinate response. Community members log elephant sightings, crop damage, and conflict incidents from their phones through the platform’s mobile app. At Kuiburi National Park, 64 AI-enabled camera traps across 32 farms transmit real-time alerts into EarthRanger when elephants step out of cover. The post states that 25 community rangers across five response teams mobilise within minutes of a detected movement toward a known crop area.
The post describes the immediate goal as guiding elephants back into the forest before serious damage occurs, and the longer-term goal as moving from reactive response to predictive response, using spatial patterns and individual profiles to anticipate conflict before it happens.
Ai2 also references a 10-Year Impact Report documenting EarthRanger’s operations across hundreds of protected areas worldwide.
Skylight
Skylight targets illegal, unreported, and unregulated fishing. The post cites an estimate of $23 billion a year in stolen marine resources and states that illegal fishing has historically relied on ships turning off tracking systems, operating at night, and transferring catch at sea. Skylight, the post states, was the first system to achieve real-time global detection of vessels using publicly available satellite imagery.
The post describes a recent enforcement action in Argentina: the Naval Prefecture identified a foreign fishing vessel operating within Argentina’s waters, Skylight’s analytics helped validate the detection, and the result was a significant fine. Argentine officials are quoted in the post describing the case as “a historical precedent: effective economic sanctions imposed without ever needing to physically intercept or board the vessel.”
According to the post, Skylight now works with nearly 150 organisations. The post states the platform is integrating capabilities to bring oil spill detection into view and is working to push that visibility closer to real time, though it does not give a timeline.
OlmoEarth
OlmoEarth is Ai2’s family of open foundation models for Earth observation. The post states it was trained on terabytes of satellite imagery, radar, and other satellite data, and is used to detect vessels, map land cover change, estimate wildfire risk, and identify deforestation. The post states that analysis that once took months can now be completed in hours.
OlmoEarth is openly available for partners to adapt. The post cites Global Mangrove Watch as an early access user, stating the organisation found that using OlmoEarth achieved 97% accuracy while significantly reducing processing time — enabling what the post describes as near-real-time monitoring of mangroves for the first time. Mangroves, the post notes, play a role in climate change mitigation.
The Skylight platform is described in the post as powered by the same technology as OlmoEarth, and a separate Ai2 post published the same week describes the addition of custom embedding exports to OlmoEarth Studio.