Taurus
Scroll. Signal. Trade.
The problem Taurus solves
Prediction markets like Polymarket remain “destination platforms,” requiring users to leave their main information source (usually X) to search markets, connect wallets, and trade: friction that slows decisions, fragments liquidity, and limits real-time sentiment capture. Taurus solves this by embedding prediction markets directly into the social feed: users can view live odds, place trades while scrolling, and react instantly to breaking news. When a tweet references an event with an active market, Taurus injects a contextual trading widget beneath it, turning passive consumption into actionable insight. Beyond human users, Taurus also powers autonomous agents like OpenClaw, which can monitor narratives, detect market signals, and execute trades automatically, enabling both humans and agents to interact seamlessly within a live, programmable prediction layer on X.
Challenges we ran into
Building Taurus meant embedding financial infrastructure directly into a fast-moving social platform, which introduced serious architectural challenges. To avoid triggering abuse detection on X, we implemented low-frequency, event-driven scanning instead of constant DOM polling. Requests are batched only when meaningful changes occur, reducing noise while staying responsive. Because real-time market matching (especially with LLM logic) is computationally heavy, we designed a thin client architecture. The extension handles UI only, while matching and inference run on the backend.
Use of AI tools and agents
Yes, we used AI to build this (shoutout to Cursor, Claude, ChatGPT, and Gemini). These tools helped debug race conditions, refactor TypeScript, test wallet edge cases, and navigate Chrome quirks, speeding up development and reducing subtle bugs. In the product, 0G (Decentralized AI) handles inference, providing decentralized execution, on-chain anchoring, and transparent, verifiable AI: essential for trusted prediction markets. For market matching, lightweight tokenization handles most tweets, and 0G AI is used only when ambiguity arises, keeping latency low and performance predictable.
Tracks Applied (3)
Futurllama
Best DeFAI Application
0g Labs
Integrate the Uniswap API in your platform
Uniswap Foundation