London-based AI startup Mantic has raised $25 million in seed funding to advance its artificial intelligence technology for forecasting future events. The funding follows the company’s performance in the Summer 2026 Metaculus Cup, where its AI system outperformed human participants in forecasting political, economic and cultural developments.
The funding round was led by Radical Ventures, with participation from M12, Microsoft’s venture fund, Thinking Machines Lab, Balderton Capital and other investors. The valuation at which the round was raised was not disclosed.
Mantic was founded in 2024 by Toby Shevlane and Ben Day. Shevlane previously worked as a research scientist at Google DeepMind, while Day has a background in machine learning research and previously served as Head of Research at Foresight Data Machines. Mantic’s stated objective is to develop AI systems capable of making accurate medium-term predictions about events that require contextual understanding and judgement.
AI Forecasting Against Human Forecasters
Mantic’s technology focuses on predicting developments across areas such as geopolitics, business, economics, technology and culture. The company says its system is designed to make predictions typically ranging from one week to one year into the future, using publicly available information and reasoning rather than relying solely on conventional numerical forecasting models.
The company’s recent performance in the Metaculus Cup has provided an important benchmark for the technology. According to the results reported by Reuters, Mantic’s system assigned probabilities to political, economic and cultural events more accurately than the human competitors in the Summer 2026 tournament. Reuters reported that the performance of Mantic and other AI entrants represented the first time technology dominated the competition.
Mantic’s own account of its performance shows that the company has been competing against both human forecasters and other AI systems across multiple Metaculus tournaments. It says its system ranked fourth among 539 human forecasters in the Summer 2025 Metaculus Cup, which it describes as its best AI result at that point.
The startup’s approach is to take existing frontier AI models developed by other companies and specialise them for forecasting. According to CEO Toby Shevlane, Mantic tests its system against historical forecasting data, evaluates its performance and then improves the system based on the results.
One area the company believes gives its system an advantage is its ability to avoid some forms of human herd behaviour. During the 2026 competition, Mantic made predictions that differed from the prevailing consensus on certain questions and subsequently proved more accurate on those outcomes.
From Forecasting Competitions to Business Applications
Mantic is positioning the technology beyond forecasting tournaments. The company says its system can provide forward-looking information for corporate strategy, risk management and government decision-making.
For businesses, potential applications include tracking competitor product launches, changes in customer behaviour, industry consolidation, new market entrants and emerging technologies. Its risk-related applications include monitoring regulatory developments, litigation outcomes, trade restrictions, supply-chain disruptions and cybersecurity scenarios.
The company has also attracted interest from businesses and government agencies, with some organisations already integrating its AI, according to Radical Ventures partner Aaron Rosenberg, who has joined Mantic’s board. Mantic has not publicly identified those customers. Rosenberg also said hedge funds and trading firms have shown particular interest in the startup’s forecasting capabilities.
Mantic previously raised $4 million in pre-seed funding, led by Episode 1, with participation from the trading firm DRW and angel investors including researchers associated with Google DeepMind and Anthropic, according to the company’s website. The latest $25 million round represents a significant increase in the capital available to develop and commercialise its forecasting technology.
With the new funding, Mantic is looking to further develop AI systems that can analyse uncertain future events and produce forecasts at scale. Its latest performance in the Metaculus Cup has provided an external benchmark for that approach, while investor participation from Radical Ventures, Microsoft’s M12, Thinking Machines Lab and Balderton Capital gives the company additional backing as it develops its forecasting platform.
