Cassi is a Superstrategy engine that combines frontier AI forecasting with collective intelligence to help organizations make probabilistically optimized decisions. The platform benchmarks human judgment against AI predictions, claiming 50% greater accuracy than humans alone and ranking 1 on the ForecastBench leaderboard. Cassi tracks and prices every resolvable claim in AI scenario reports through its Forecast Room feature.
Funding
Funding not disclosed
Founders
Product
Problem
Organizations struggle to make strategic decisions under uncertainty, relying on human judgment that is often biased, overconfident, or influenced by group dynamics. Traditional forecasting methods fail to systematically integrate diverse perspectives with the predictive power of advanced AI, leaving decision-makers without reliable probabilistic guidance for complex, fast-moving scenarios.
Solution
Cassi provides a Superstrategy engine that pairs collective intelligence—making a team's judgment legible and stripped of bias—with frontier AI forecasting to produce decisions that are more accurate than either approach alone. The platform claims to forecast 50% more accurately than humans and ranks #1 on the ForecastBench leaderboard among AI systems. Cassi's Forecast Room puts numbers on AI scenario reports by pricing and tracking every resolvable claim, giving organizations a clear, quantified view of future possibilities. The system is designed to tell decision-makers what they need to know, not what they want to hear, countering the "Cassandra curse" where accurate predictions go unheeded.
Target Audience
Primary customers are organizations and decision-makers who need accurate, probabilistic forecasts for strategic planning, including corporate strategy teams, risk management functions, and government or military intelligence units facing high-stakes, uncertain futures.
Features
- Combines collective intelligence with frontier AI forecasting to deliver probabilistically optimized decisions
- Ranks #1 on the ForecastBench leaderboard with a score of 68.8, outperforming elite human superforecasters (69.2) and other AI systems like Grok 4.20 (68.1)
- Forecast Room feature that prices and tracks every resolvable claim in AI scenario reports
- Benchmarks human judgment against AI predictions to identify and remove cognitive biases
- Participates in 23 forecasting tournaments, demonstrating continuous validation and refinement of its predictive models
- Publishes scientific essays and research on forecasting methodology, including benchmarking approaches