FutureSearch

FutureSearch is an AI forecasting company founded in 2023 and based in San Francisco, California. The company was co-founded by Dan Schwarz, who previously served as CTO of Metaculus and created Google's internal prediction market, and Lawrence Phillips, who led Metaculus' AI team. Their core mission is to build AI systems capable of tackling complex real-world analytical tasks that existing AI systems cannot handle well, with a particular emphasis on producing forecasts with legible, inspectable reasoning. The company's platform deploys teams of LLM agents across tabular data to perform operations including ranking, classifying, researching, merging, deduplicating, and forecasting — offering what they describe as "a researcher for every row." Agent types include Researchers, Forecasters, Scorers, Classifiers, and Matchers, available via a Python SDK and API. Enterprise clients include Amazon, OpenAI, Google, xAI, and Anthropic. FutureSearch has been deeply embedded in the EA and AI safety forecasting ecosystem. In April 2024, the team published "The Rationale-Shaped Hole At The Heart Of Forecasting" on the EA Forum, arguing that the forecasting ecosystem produces accurate predictions but fails to preserve the underlying reasoning and models. FutureSearch was explicitly designed to address this gap through structured, reason-backed AI forecasting. In 2024, FutureSearch became the first organization to demonstrate an autonomous AI forecaster that was a net winner of bets against real humans on geopolitical questions. They also received two grants from Open Philanthropy: $606,600 to develop a public benchmark evaluating LLMs' ability to forecast geopolitical events, and $356,000 to support research on automating the generation of forecasting questions. In April 2025, FutureSearch contributed to the AI Futures Project's "AI 2027" report, providing probabilistic forecasts on superintelligence timelines. The team, including CEO Dan Schwarz, CTO Lawrence Phillips, and Director of Forecasting Tom Liptay (MIT PhD in AI and cognitive science), argued for somewhat slower timelines than other forecasters due to R&D bottlenecks, commercial priorities at leading AI labs, and the potential for regulatory intervention. The company raised a seed round of approximately $6.2 million in December 2024 from investors including Metaplanet Holdings and Polaris Ventures.