FutureSearch's forecasters are now available to the public

Frontier intelligence now available as your team of AI researchers

the story

FutureSearch cofounders Dan Schwarz and Lawrence Phillips have been on a quest to predict the future for years. While at Waymo in 2019, Dan built what became Google's internal prediction market, still running there today. He then served as CTO of Metaculus, the pre-eminent public forecasting platform, where he met Lawrence. Lawrence had joined Metaculus as Head of AI in 2021, after his PhD in theoretical physics and serving in senior ML roles in biotech.

At Metaculus, Dan, Lawrence, and the teams they managed worked to predict the future of AI. When GPT-4 came out in 2023, they saw the opportunity to greatly advance the frontier of accuracy, and left to co-found FutureSearch. For two and a half years, they published research, provided intelligence to frontier labs, and built a public track record of accuracy.

Now, in April 2026, they are bringing their best techniques to the world. Everyone can now get their own team of AI researchers that work together to provide accurate information about the present and future.

track record

dec 2023 First to use an AI forecaster to beat humans in a prediction market tournament.
jul 2024 First to correctly break down OpenAI's revenue when public estimates were inaccurate.
2024–2025 Sold intelligence to Google, Anthropic, OpenAI, xAI.
apr 2025 Co-authored the core timeline forecasts for the AI 2027 report.
may 2025 Published Bench to the Future and Deep Research Bench, benchmarks and leaderboards for AI forecasters and research agents.
apr 2026 Published paper demonstrating SOTA AI forecaster and made it available to the public.

the product

FutureSearch is your team of researchers. Use them to forecast and research at scale, or analyze entire datasets.

Batched forecaster: $0.15–0.50 / question
High-effort forecaster: $1.60 / question

In addition to forecasters, your team can work as researchers, scorers, classifiers, or matchers, coordinating to solve any large-scale analysis or research problem.

what early users of the public product are saying

"FutureSearch makes me more accurate and that's essential for my work."

— ceo

"The API runs beautifully and the data quality we are getting is exceptional for our quantitative screening."

— founder

"I've been actively using FutureSearch and I'm genuinely impressed. The depth of research it provides, especially around industries and startup ideas, is very strong."

— founder

"Very impressed! It worked right out of the box. I've tried other platforms and it's been hit or miss, but FutureSearch just worked on my first shot."

— software engineer

team

12 people, 7 PhDs, mostly worked together on forecasting prior to FutureSearch.

backed by

FutureSearch raised a $5.7M seed round in Nov 2024, led by Metaplanet, who co-led Anthropic's first round and were early DeepMind backers. Other AI specialist investors include Juniper, Macroscopic, Ought, and TwinPath.