ManticAI’s Top-10 Finish Built on Shoulders of Giants like OpenAI and Google

While British startup ManticAI is earning praise for its AI’s top-ten finish in a forecasting competition, its success also highlights the foundational role of tech giants like OpenAI, Google, and DeepSeek. ManticAI’s innovative system works by strategically orchestrating large language models developed by these industry leaders.
The startup’s key innovation is not in building a massive new model from scratch, but in creating a sophisticated multi-agent system that leverages the best of what’s already available. ManticAI’s platform acts like a “general contractor,” breaking down a forecasting problem and assigning sub-tasks to a “roster” of AI models based on their specific strengths.
For example, a task requiring deep and broad knowledge might be assigned to a Google model, while one that requires nuanced text generation might go to an OpenAI model. This approach allows the startup to stand on the shoulders of giants, combining the power of multi-billion dollar research efforts into a single, cohesive forecasting engine.
This model of innovation is becoming increasingly common in the AI ecosystem. As foundational models become more powerful and accessible via APIs, startups can focus on building value at the application layer—creating novel workflows and systems that solve specific business problems, like ManticAI has done with forecasting.
The eighth-place finish in the Metaculus Cup is therefore a win-win. It’s a victory for ManticAI’s clever architecture and a validation of the incredible power of the underlying models from OpenAI, Google, and others. It demonstrates a healthy and dynamic AI industry where innovation can happen at multiple levels.