superagent_playarts
Transforming personal data into valuable, community-owned AI assets with dynamic NFTs.
Demo
About
The problem it solves Current NFT and AI models lack flexible monetization and effective royalty-sharing methods. Traditional static NFTs don't adequately leverage AI capabilities or interactive enhancements. Superagent PlayArts solves these problems by combining dynamic NFTs with Intelligent Personal Agents (IPAs), enabling royalty-sharing assets that continuously evolve and provide users with ongoing value and monetization User Interaction and Data Flow ### 1. Creating Your Personal Agent (Chat Interface) - Open chat and send an image of your character, or directly import from your existing NFT. - Chatbot confirms receipt and begins training your AI agent with one-shot training. - Receive a chat notification once training is complete. ### 2. Wallet and NFT Analysis (Chat Interface) - Type your wallet address directly into the chat. - The chatbot analyzes your wallet activities and NFTs. - It responds with a clear, conversational summary and insights. ### 3. Generating Images & Tweeting via Chat - Describe the type of image or outfit you want directly in chat. - The chatbot uses Anime Segmentation and Inpainting with a Magic Brush to remix and style your NFT while keeping the original style intact. - Confirm if you want the chatbot to automatically post the image to Twitter. ### 4. Monetizing Your Agent (Chat Interaction) - Chatbot provides an easy-to-use link to monetize your trained agent. - Set pricing or usage rights directly via conversational prompts. - Receive automated royalty updates within the chat interface. ### 5. Continuous Engagement - Chat with your agent regularly; the chatbot learns and improves from each interaction. - Get suggestions and real-time feedback through friendly chat conversations. ### 6. subscription-based tweeting using IPA-trained agents - Collaborative media thinking (similar to LLM's thinking mechanism) to optimize outfits through multiple model coordination. ### Example Chat Flow: 1. You: "Import this NFT and train an AI agent from it." 2. Bot: "Got it! Training your personal AI agent now. I'll notify you shortly." 3. Bot: "Your agent is ready! What would you like to create?" 4. You: "Remix my NFT with a futuristic outfit using the Magic Brush." 5. Bot: "Done! Should I tweet this for you?" This simplified chat-based approach makes using advanced AI features, NFTs, and monetization straightforward and enjoyable for all users. The project architecture and development process # Agent-Based Web3 Service Integration Project This project offers an all-in-one solution that unifies: - **One-Shot (LoRA) Training**: Build a personalized agent with only one (or a few) images. - **Web3 Wallet/NFT Analysis**: Analyze wallets and NFT data. - **Twitter Auto-Posting**: Generate context-aware images and automatically post them on Twitter. - **Revenue Sharing & Agent Monetization**: Turn your trained agents into valuable assets. - **True AI Remix NFT Evolving**: The chatbot uses Anime Segmentation and Inpainting with a Magic Brush to remix and style your NFT while keeping the original style intact. ## Key Features ### One-Shot LoRA Training - Train a LoRA model with just a single or few images. - `TrainerService` handles background removal (SAM), compositing, auto-caption, and training. - Once training completes, the agent is registered in the DB with a unique `agent_key`. ### Web3 Wallet / NFT Analysis - `NFTService` and `WalletService` to parse wallet addresses, NFT data, and transactions. - Summarize findings via LLM-based reporting. ### Agent API (Monetization & Revenue Split) - After training, the agent is ready for image generation via `/agent/<agent_key>/inference`. - The agent owner can share or sell usage rights, enabling potential revenue streams. ### Twitter Integration - Create an AI-driven tweet: LLM to interpret user messages or NFT/wallet context → agent-generated image → auto-upload to Twitter. ### Revenue Sharing Model - Deploy agent as an NFT, or integrate a smart contract to handle royalty splits. - Earn from API usage fees or secondary sales. Product Integrations ### Product Integrations All services are powered by our proprietary On-Premise APIs and a scalable internal API infrastructure. except twitter api - **Playarts Main API** [https://api-ai-playarts.ai/swagger](https://api-ai-playarts.ai/swagger) Provides core services and scalable internal APIs. We currently offer five API extension services: - **Authentication Service** [https://api-auth-alpha.playarts.ai/swagger/index.html](https://api-auth-alpha.playarts.ai/swagger/index.html) - **Community Service (Likes, Comments, etc.)** [https://api-community-alpha.playarts.ai/swagger/index.html](https://api-community-alpha.playarts.ai/swagger/index.html) - **Real-time Service & Push Notifications (WebSocket-based)** [https://api-push-alpha.playarts.ai/swagger/index.html](https://api-push-alpha.playarts.ai/swagger/index.html) - **AI Agent Service** [https://api-ai-agent.playarts.ai](https://api-ai-agent.playarts.ai) Central to this Agent Buildathon, the LLM Engine handles all decision-making, analysis, and processing of requests, enabling advanced automated AI-driven functionalities. Additionally, two more internal extension services are continuously developed and expanded. Key differentiators and uniqueness of the project **Key differentiators and uniqueness of the project** - **One-shot NFT-based AI Training** Quickly creates personalized AI agents directly from existing NFTs or single images, enabling effortless personalization. - **Dynamic NFT Outfit Remix** Utilizes advanced **Anime Segmentation** and **Magic Brush Inpainting** to remix NFT outfits while preserving original artistic style and integrity. - **Integrated Revenue Sharing** Implements automated, transparent royalty distribution through blockchain smart contracts, ensuring fair and easy monetization. - **Chat-based Interaction** Offers a simplified, conversational chatbot interface, enhancing usability and engagement compared to traditional, complex platforms. Trade-offs and shortcuts while building **Trade-offs and shortcuts while building** - **Rapid Prototype Focus** We prioritized quickly building key features like one-shot AI training and NFT remixing, opting for simpler integrations initially to speed up development. - **Simplified User Interface** Chose a basic chat interface over advanced UI/UX design, enabling faster iteration and user testing. - **Limited Scalability** Currently, the system uses basic GPU management; future improvements include adopting advanced container orchestration (e.g., Kubernetes) for better scalability. - **Revenue Sharing Mechanism** Implemented a simple, preliminary revenue-sharing model; planning more robust smart contract integrations and security enhancements in future updates. Additional Features **Additional Features** All features submitted to GitHub are newly developed specifically for this BUIDLathon, including: - **One-shot NFT-based AI training** - **Advanced outfit changes using Anime Segmentation and Magic Brush Inpainting (True AI Remix feature)** - **Subscription-based tweeting using IPA-trained agents** - Collaborative media thinking (similar to LLM's thinking mechanism) to optimize outfits through multiple model coordination. - Integrated collaboration between LLM and CLIP models for characterizing images effectively before automatic tweeting. None of these features existed in the project prior to this BUIDLathon.