Verity

What is it? Verity is a prediction market protocol where anyone can paste an X link, and AI instantly creates a tradeable market while every trade is monitored for manipulation in real-time and markets auto resolve through Chainlink Price Feeds or AI-powered verification.
What problem does it solve? Prediction markets have three connected problems. Creation is friction-heavy users must manually craft questions that are often ambiguous with no clear resolution source. Insider trading is rampant and undetected on Polymarket, someone turned $32K into $436K profit on the Maduro capture hours before the public announcement. And resolution is broken UMA takes minimum 2 hours, is vulnerable to whale governance attacks, and has confirmed wrong outcomes.
How it works : Three CRE workflows handle the full lifecycle. Workflow 1 (Market Creation): users paste an X post link, tweet text, or type a question manually. AI fetches the content, extracts the verifiable claim, categorizes it (crypto price, social, or event), generates precise resolution criteria, and risk-scores it auto-approving safe markets, queuing borderline ones for admin review, and rejecting spam instantly. Workflow 2 (Anomaly Detection): fires on every single trade via Log Trigger. It reads market state on-chain, fetches real-time context from News API, and runs AI manipulation analysis if the score exceeds 80, the market is paused automatically with zero human delay. Workflow 3 (Smart Resolution) : crypto price markets settle deterministically through Chainlink Price Feeds no AI, no opinion, just on-chain math. Social and event markets resolve through AI cross-checking multiple sources via News API, with a confidence threshold that escalates uncertain outcomes instead of guessing wrong. Seven CRE capabilities, Chainlink Price Feeds, News API, and three AI-powered workflows one protocol, three problems eliminated.

Verity runs on three CRE workflows orchestrating seven distinct CRE capabilities, smart contracts on Base Sepolia, and a Next.js frontend.
Workflow 1 Market Creation: HTTP-triggered. Accepts manual questions, pasted tweet text, or X post links. For links, Confidential HTTP fetches the tweet content. Groq AI then extracts the verifiable claim, auto-categorizes it (CRYPTO_PRICE / SOCIAL / EVENT), generates resolution criteria with data sources, and assigns a risk score. Score 0-30 auto-approves via EVM Write to createMarket(); 31-70 goes to admin queue; 71-100 auto-rejects. For crypto markets, AI also identifies the correct Chainlink Price Feed address and target value.
Workflow 2 Anomaly Detection: Log-triggered on every BetPlaced event. EVM Read pulls market state (pools, volume, bettor count). Confidential HTTP fetches context from News API and price feeds. Groq AI analyzes the bet against market context, external signals, and known manipulation patterns. Score above 80 triggers EVM Write to reportManipulation(), automatically pausing the market.
Workflow 3 Smart Resolution: Log-triggered on SettlementRequested. Branches by category : CRYPTO_PRICE reads Chainlink Price Feeds on-chain — purely deterministic, 100% confidence. SOCIAL and EVENT markets fetch data via Confidential HTTP from News API and relevant sources, then AI determines the outcome with confidence scoring. Confidence below 90% triggers escalation instead of auto-resolution. Final outcome is submitted via EVM Write.
All three workflows use runInNodeMode with BFT Consensus. Smart contracts are Solidity/Foundry with USDC transactions and ERC-1155 position tokens. Frontend is Next.js with wagmi/viem and TailwindCSS.