Mantic Emerges Out of Stealth Mode to Provide a New Kind of Foresight

Mantic, a platform that predicts global events with superhuman accuracy, led by Ben Day and Toby Shevlane, announced its launch with a world-class team of AI technical staff, including experts from Google DeepMind, Citadel, the universities of Cambridge and Oxford, and other AI startups. Mantic has raised $4 million in its pre-seed round led by Episode 1 with backing from DRW and various top AI researchers at Anthropic and Google DeepMind. Mantic is advancing the frontier of AI forecasting accuracy, winning the top prize in the Q1 2025 Metaculus AI Benchmark Tournament and setting a new state of the art when backtested on 348 questions from Q2 2025.

Mantic develops AI systems to predict events across geopolitics, business, policy, technology, and culture—domains where data alone falls short. These areas demand flexible reasoning and research, which is why human superforecasters still outperform automated methods. Mantic aims to change this by delivering automated predictions with unprecedented accuracy and scale.

Good decision-making depends on accurate forecasting, yet human prediction remains limited. Superforecasters have shown that skilled judgment can outperform domain experts and even intelligence analysts, but their insights rarely shape high-level strategy. Prediction markets offer another approach, though they too fall short of their ambitions. Unlike weather forecasting, which has achieved steady, measurable progress and deep economic integration, judgmental forecasting has yet to realize its transformative potential. Mantic aims to close this gap by solving the challenge of judgmental forecasting at scale.

Human forecasters take months to test new techniques, but AI can be backtested instantly by restricting knowledge to past information and replaying world events. This collapses evaluation latency from months to milliseconds, enabling rapid iteration and reinforcement learning at scale. With a dataset of over 10,000 high-quality forecasting questions from 2024–2025, AI systems can accumulate far more forecasting experience than any human ever could.

Human forecasting capacity is scarce, and prediction markets face scale and relevance limitations. AI forecasting overcomes these constraints by delivering forecasts rapidly, at scale, and tailored to client needs. Unlike humans, AI can continuously refresh predictions and surface hidden information streams, as shown in the China photovoltaic case—where daily updates quickly anticipated a major shift that human consensus only recognized later.

Forecasting abundance unlocks entirely new possibilities: moving beyond isolated predictions to dynamic maps of unfolding events, radar-like scans that highlight what matters, and deep explorations of complex scenarios. At an industrial scale, forecasting can reveal trajectories too vast for any individual to comprehend—building a far richer picture of the future than ever before.