Search projects, hackathons, sponsors and tech

    ← All projects

    Vision Pay

    Real Time Vision Payment Assistant using A2A, AP2 mandates, Hedera Agent Kit and PYUSD

    Demo

    About

    VisionPay AI-Powered Autonomous Payment Assistant Like Ray-Ban Meta glasses meets Apple Pay - AI watches your world and handles payments intelligently. VisionPay Blockchain Protocol 🎯 The Vision Imagine wearing smart glasses that let you pay for things just by looking at them. Point your camera at a coffee shop logo, and your AI assistant handles the payment automatically based on rules you set. No phone, no wallet, no buttons. VisionPay brings this future to today. ✨ How It Works 1. Set Payment Rules "Auto-pay coffee shops under $10" "Auto-pay transit fares under $5" "Always ask for restaurants over $20" 2. Point Your Camera The AI continuously watches through your camera (or wearable device), detecting vendor logos and signs in real-time. 3. Agent Decides When a vendor is detected, the autonomous agent checks your payment rules: Match found? → Pays automatically with HBAR or PYUSD No match? → Asks for your approval No button clicks needed. Just point and go. 🚀 Technology Stack Real-Time AI Vision Gemini 2.5 Live Flash - Multimodal streaming (video + audio) Continuous vendor detection Voice interaction ("I see Starbucks, paying now...") Autonomous Blockchain Agent Hedera Agent Kit - Makes payment decisions autonomously Risk assessment and learning Executes transactions based on your rules Secure Authorization AP2 Protocol (Agent Payments Protocol) Cryptographically-signed delegated mandates Audit trail for every transaction Multi-Chain Payments HBAR (Hedera) - Fast, low-cost transactions PYUSD (Ethereum/Polygon) - Stablecoin flexibility 💡 Use Cases 🚇 Contactless Transit Walk through subway turnstile, AI detects SFMTA logo, auto-pays $2.50 fare. ☕ Coffee Shop Auto-Pay Point camera at Starbucks, agent checks mandate ("coffee under $10"), pays instantly. 🍽️ Restaurant Safety Check Point at expensive restaurant menu, AI sees $45 price, asks for approval first. 🛒 Grocery Shopping Scan items as you shop, agent tracks total against daily spending limit. 🎫 Event Entry Look at event QR code, AI verifies and pays ticket fee automatically. 🎨 User Experience Mobile-First Design Full-screen camera view (like Instagram/TikTok) Bottom sheet payments (swipeable event history) Touch-optimized (large buttons, active states) Back camera ready (perfect for phones) Desktop Experience Split-screen layout (camera + payment sidebar) Real-time status (agent insights, audio visualization) Protocol badges (Gemini, Hedera, AP2) Autonomous Intelligence Risk levels (low/medium/high color-coded) Confidence scores (AI explains decisions) Learning history (improves over time) Daily limits ($100 safety check)

    How it's made

    🏗️ Architecture Frontend Next.js 15 with App Router React 19 with Hooks TailwindCSS 4 for responsive design TypeScript for type safety AI & Vision @google/genai for Gemini Live streaming Real-time video frame processing Audio input/output for voice interaction Tool calling for vendor detection Blockchain @hashgraph/sdk for Hedera transactions ethers.js for Ethereum/PYUSD hedera-agent-kit for autonomous decisions Protocols AP2 Delegated Mandates for authorization Agent decision engine with risk assessment Transaction history for learning 🔐 Security Delegated authorization via AP2 mandates (cryptographically signed) Daily spending limits (configurable, default $100) Execution count limits per mandate Risk assessment on every transaction User confirmation for high-risk/large amounts Audit trail for all payments