Climate systems fail at the ecosystem level, yet most restoration decisions are made using fragmented data. ProgramEarth applies AI to environmental monitoring systems designed with local and Indigenous stewards, ensuring models reflect how land, water, and species actually interact.

Earlier in my career at GitHub, I worked with CTOs to determine when AI delivered highest on developer productivity. The most effective AI systems were those trained on large, diverse datasets of real-world behavior. ProgramEarth applies the same principle to nature by training AI on ecosystem behavior defined and validated by long-term local stewardship.

Climate restoration moves faster when we shorten the distance between what the land is telling us and how we respond. ProgramEarth uses AI to synthesize environmental sensor data alongside community knowledge, the result helps restoration teams identify risks, patterns, and remediation pathways sooner.

In the Klamath River Basin, ProgramEarth worked with the Hoopa Valley Tribal members to model how Chinook salmon restoration affects wildfire risk, erosion, flooding, and soil stability. The analysis was used by Tribal and regional councilmembers to show that keystone species recovery could reduce California’s long-term wildfire, flood, and erosion response costs by more than $10 billion annually, reframing restoration as a high-ROI climate resilience strategy.

In Bluff, Utah, ProgramEarth partnered with the Ute Mountain Ute Tribe to guide bioremediation through invasive Tamarisk removal, biochar application, and cottonwood reforestation. AI-guided monitoring showed improved soil permeability and aquifer recharge within a single growing season, and this data was used by the Utah Division of Water Resources to evaluate land-based water replenishment as one of the strategies to implement.

With the Texas Tribal Buffalo Project, ProgramEarth is modeling how Apache-led buffalo grazing restores native grasses and wildflowers across the San Antonio plains. Early vegetation recovery and soil moisture indicators are being used to assess improvements in infiltration and surface water retention, informing long-term regional water resilience planning.

This talk demonstrates how AI-driven ecosystem modeling, informed by steward knowledge, helps governments, funders, and insurers prioritize restoration interventions that reduce risk, avoid disaster costs, and deliver durable resilience.