Climate Simulation Lab

Subconscious AI

Climate Simulation Lab

Simulates climate interventions with synthetic populations to reveal what drives real, equitable behavior change at scale

New York, NY, USA

USA

For-profit, including B-Corp or similar models

Climate programs are being launched without any reliable way to predict whether people will actually adopt them. Governments and NGOs urgently need to reduce emissions and protect vulnerable communities—but too often, their efforts fail to drive real behavior change. Why? Because current tools—surveys, A/B tests, and pilots—are slow, expensive, and poorly suited to uncovering why people act the way they do. Among the 3.4 billion people living in urban areas—many in climate-vulnerable regions UN-Habitat—too many are excluded from solutions meant for them. Globally, more than $1.3 trillion was spent on climate-related programs in 2021–2022 Climate Policy Initiative, yet the World Bank notes that many adaptation efforts lack systematic data on behavioral aspects and resilience outcomes, limiting their effectiveness. This is a massive blind spot. When programs don’t account for the real drivers of human behavior—trust, habit, identity, tradeoffs—they miss the mark. From clean energy adoption to climate-resilient housing, policies fail not because they’re wrong—but because they don’t resonate. Closing this behavioral prediction gap is essential to reaching global climate goals.

We developed a behavioral simulation platform that helps governments, NGOs, and urban planners test climate interventions—before they are deployed in the real world. We use Causal AI, synthetic populations, and large language models (LLMs) to simulate how real communities will respond to climate policies like retrofit subsidies, energy pricing, or heat alerts. Think of it as a behavioral wind tunnel: users can test different policy designs, messaging strategies, or delivery mechanisms, and instantly see what drives action—across demographics, geographies, and time. Our synthetic populations are built using over 800 million real-world datapoints and validated by replicating 350+ published behavioral science studies. We’ve achieved up to 93% correlation with human decision-making in domains like public health, transportation, and energy adoption. Currently, we are working with the U.S. Department of Health and Human Services, a global health NGO, and academic researchers to simulate behavior around climate-related public health interventions. Our platform compresses months of policy testing into minutes, enabling smarter, faster, and more inclusive design. Every model is auditable, bias-tested, and fine-tuned to account for the realities of underrepresented communities most impacted by climate change.

Subconscious.ai serves governments, NGOs, and urban planners responsible for designing climate programs that reach the communities most vulnerable to climate risk—especially renters, low-income families, frontline workers, and transit-reliant populations in urban areas. These communities are often left out of traditional climate research due to survey fatigue, digital exclusion, or mistrust of institutions. This leads to programs that are misaligned with real behavior, poorly adopted, or unintentionally inequitable. Our platform helps decision-makers simulate how different segments of these populations will respond to specific policies, incentives, or messaging—before any resources are spent. For example, we’ve modeled how different language or framing impacts the effectiveness of heat alerts in multi-lingual neighborhoods, or how retrofit subsidies fail to reach renters under current designs. We currently support U.S. federal agencies, a global health nonprofit, and civic researchers working to improve climate-related public health messaging and housing equity. Altogether, our tools are being used to inform interventions that could impact millions. Long term, we aim to equip community organizations directly—giving them the power to simulate their own ideas and co-create policies that reflect lived experience. We're already piloting this model in partnership with public health departments and academic research labs