WalkiePuppie

Video Demo

About this project

Walkie Puppie is an AI-powered AR puppy companion that motivates users to stay active through walking. Built as an iOS application using Unity, AR Foundation, ARKit, Swift, and Xcode, users can place a virtual puppy in augmented reality, track walks, chat with an AI companion, receive personalized fitness insights, and create blockchain-backed photo memories. MongoDB Atlas stores user profiles, NFT records, photo gallery data, metadata references, and explorer links. Its flexible document model allowed us to rapidly build and iterate on user-facing features. Snowflake powers both analytics and AI insights. Every walk session is stored using Snowflake SQL, including distance, duration, mood, timestamps, and activity history. We use Snowflake Cortex AI to analyze this data and generate personalized insights. Users can ask questions such as “How did I do this week?” or “What patterns do you see in my walks?” and receive AI-generated feedback based on their actual walking history. Backboard provides persistent memory for the puppy. Instead of acting like a standard chatbot, the puppy remembers previous conversations, user preferences, and activity habits, creating a more engaging and personalized companion experience. DigitalOcean Gradient AI powers the conversational intelligence of Walkie Puppie. Using Mistral 3 14B, the puppy generates responses, incorporates memory context from Backboard, and provides personalized encouragement based on user activity. For photo authenticity, Walkie Puppie uses Solana Devnet, Metaplex, Pinata, and IPFS. Users do not need a crypto wallet. Photos are uploaded to IPFS, metadata is generated, and NFTs are minted by a project-owned Solana wallet. The NFT is stored in the platform wallet and linked to the user account, creating a completely walletless experience while preserving blockchain verifiability.