Belgravia Hartford has released a technical progress update on gravitio.ai, its Prediction Intelligence Platform, reporting 5,874 recorded AI-agent predictions since April 2026 and more than 8,000 cloned AI agents now active on the system. The disclosure also covers early internal performance metrics in crypto and football prediction, an expansion of sports coverage, and continued development of a proprietary prediction-performance data layer.
What a Prediction Intelligence Platform Actually Does
A prediction intelligence platform is a system that deploys AI agents to generate, track, and evaluate forecasts — in this case, across financial and sports domains. The value proposition is not just the prediction itself but the underlying data infrastructure that scores whether those predictions prove accurate over time. That performance layer is what separates a speculative tool from one that institutions or sophisticated users might eventually trust for decision-making.
Gravitio.ai is building precisely that infrastructure. The platform's prediction-performance data layer is described as proprietary, meaning the methodology for measuring and attributing agent accuracy is developed internally — a potentially meaningful moat if the metrics prove credible.
The Numbers Behind the Update
Since April 2026, the platform has accumulated 5,874 recorded AI-agent predictions. The cloned-agent figure — more than 8,000 — indicates the system allows users or internal processes to replicate and deploy AI agents at scale, which is how prediction volume compounds quickly from a relatively short operational window.
Early internal metrics are available for two verticals: crypto markets and football. The update characterizes these as internal figures, which means external verification is not yet available. That distinction matters for anyone evaluating the platform's claims; internal benchmarks are a starting point, not an audited track record.
Expanded Coverage and the Data Layer
Alongside the prediction metrics, Belgravia Hartford's update points to expanded sports coverage, broadening the domains in which gravitio.ai operates beyond its initial verticals. The continued development of the prediction-performance data layer is the thread that runs through all of this: without a rigorous and transparent scoring system, raw prediction counts carry limited analytical weight.
The Toronto-based company's update positions gravitio.ai as a platform still in active technical development. The volume figures suggest meaningful early activity, but the platform's long-term credibility will depend on how that proprietary performance data holds up as the prediction record grows and external scrutiny increases.