OmniBet

OmniBet is an AI-powered, omnichain prediction market platform that allows users to bet on real-world events from any blockchain, and interact with dynamic, short-form markets created in real time. Built on top of LayerZero’s lzCompose and leveraging Flow blockchain, AI agents, and NFT rewards, OmniBet brings together the best of prediction markets, cross-chain UX, and gamified mobile engagement.

🌐 Core Idea OmniBet aims to make prediction markets accessible, fast, and fun by solving three problems:

Chain fragmentation — users shouldn’t need to switch networks to access a market.

Manual market creation — markets should emerge in real time based on live events.

Boring UI/UX — betting should feel intuitive, social, and rewarding.

🔮 Key Features

1. Omnichain Market Participation Users can place bets on markets created on any chain from any other chain — powered by LayerZero’s lzCompose.

Liquidity is automatically rebalanced across chains, and parallel cross-chain messages ensure low-latency execution.

1. AI-Curated Real-Time NBA Markets An AI agent watches live NBA games using stream analysis and LLMs.

It creates short-timed micro-markets like: “Will LeBron score in the next 2 minutes?”

These are deployed exclusively on Flow, designed for mobile-first interaction.

1. Swipe-Based UX (Tinder Style) Markets are presented in a Tinder-style card interface.

Users swipe right for “Yes”, left for “No” — fast, intuitive, and accessible.

No form-filling or confusing options — it’s prediction betting made for the TikTok generation.

1. Tamper-Proof Market Resolution Market outcomes are resolved using vLayer WebProof.

It fetches real-time NBA scores from trusted sources and generates cryptographic proofs that resolve the bets on-chain, without human input.

1. NFT Rewards with Real-World Utility Winning users receive NBA Player NFTs minted on Flow.

These NFTs can be redeemed for:

Official merchandise

Fan community access

Future loyalty points or collectibles

OmniBet was built as a deeply composable, modular system that brings together omnichain infrastructure, AI-generated prediction markets, and gamified UX, all working in real time across multiple blockchain ecosystems.

We used a mix of on-chain smart contracts, cross-chain messaging protocols, AI agents, and mobile-first front-end design, with a few “hacky” workarounds to enable live-market resolution and instant UX.

⚙️ Tech Stack Breakdown ✅ Smart Contracts EVM Contracts (Solidity) for general prediction markets

Factory + Singleton-style contract

Handles creation, user bet tracking, escrow, and payout logic

Flow Contracts (Cadence) for NBA AI-generated swipe markets

Swipe interaction stored as on-chain vote

NFT mint logic for winners triggered on market resolution

✅ Omnichain Infrastructure (LayerZero + lzCompose) Used LayerZero’s lzCompose to:

Let users on any chain place bets on markets hosted anywhere

Route token staking, message payloads, and user actions across chains

Automatically rebalance liquidity using multi-chain token contracts

lzCompose’s parallel message execution gave us near-instant UX and simplified bridging logic

✅ AI Agent for NBA Market Creation Built an AI agent using:

OpenAI Whisper to transcribe NBA commentary from stream audio

LangChain + GPT-4 to extract intent and generate time-boxed market statements

Example output: “Market: Will LeBron score in next 2 mins?”

These were auto-deployed to Flow with an internal “Market Publisher” function

✅ vLayer WebProof for Market Resolution Used vLayer’s WebProof to:

Scrape trusted NBA APIs and sports sites

Generate zk-verifiable proofs of event outcome

Submit those proofs back to the Flow or EVM contract to resolve the market

This let us resolve bets trustlessly, without needing oracles or manual intervention

✅ Frontend React Native for a mobile-friendly, swipe-based UI (Tinder-style)

Each market is shown as a card: swipe right to bet “Yes”, left for “No”

WalletConnect + Wagmi + Flow SDK for wallet integration

Used useEffect with LayerZero hooks to show confirmation modals once cross-chain txs landed

🧪 Hacky / Clever Workarounds 🧠 We built a custom card queue component to render swipe markets that auto-fetched the next available AI market from Flow.

🔁 For liquidity rebalancing, we piggybacked on lzCompose callback flows to “simulate” token transfers without bridging — keeping user assets in their native chain.

📺 For NBA detection, we scraped Twitch and YouTube stream titles in real time and cross-referenced team names to improve market context — even when video parsing was slow.