AI can help you trade, but what I most want to know is: Why was this position opened? Can the strategy be backtested? And if the judgment was wrong, can we trace back the reason? After reviewing Travix_fi’s project documentation, I believe this direction is worth paying attention to. Travix positions itself as an Agentic Omnibroker—a blockchain-based trading platform that integrates multi-asset trading, event analysis, and AI-driven strategy execution. According to the team, it has completed a seed round backed by Amber Group and is currently advancing toward mainnet launch. The asset selection is particularly interesting, with a strong focus on compute and East Asian markets. Assets such as H100 chips, memory, and electricity, along with semiconductor stocks in Japan and South Korea, are all part of its product roadmap. The goal is to allow users to manage cross-asset positions under a unified margin system, using 24/7 perpetual contracts to express views or hedge risks. If you follow the AI industry, this approach makes perfect sense. Chip supply and compute costs impact different companies across the value chain; bringing these assets together enables coherent trading around a single thesis. But having assets isn’t enough—you also need to understand how events affect them. Travix’s Financial World Model attempts to track changes in the probability of events occurring, combining forward-looking data such as prediction markets to analyze how these shifts propagate to specific assets. As designed, the output includes probability changes, data sources, confidence levels in asset impact, and key risk factors—allowing users to audit the reasoning behind their judgments. The subsequent Trusted Agent layer converts this information into testable strategies, then executes trades according to clearly defined rules, preserving the rationale for entry, conditions for strategy failure, and full execution logs. This is what I value most. It’s easy for AI to generate a bullish rationale—but turning that rationale into executable rules, validating them with historical data, and reviewing them post-trade is what allows you to seriously assess whether it actually works. The order book is already live; AI trading and causal-chain trading remain in beta, with mainnet timing yet to be announced. For now, it’s best to monitor product progress closely and pay particular attention to how well strategy backtesting and execution records are implemented. If you’re interested in AI × trading or compute assets, follow @Travix_fi.
