Datagraph Arena

Problem Statement

This project allows users to complete RLHF (Reinforcement Learning from Human Feedback) tasks and earn WLD tokens as rewards inside the World app.Datagraph World is a "train-to-earn" platform built as a Mini App inside World App, created to solve AI's most critical challenge: sourcing high-quality, human-verified training data. As AI models become more powerful, their performance is limited by the quality of the data they are trained on, which is often unreliable and vulnerable to bots. Our project bridges this gap by creating a marketplace between AI developers who need authentic data and the 23 million verified humans on World App.Our solution is a seamless interface where any World App user can become an AI trainer. After a one-time verification of their personhood using their World ID—our fundamental guarantee against bots—users are presented with simple but crucial tasks. Initially focused on Reinforcement Learning from Human Feedback (RLHF), these tasks might involve ranking AI-generated responses or choosing the most helpful chatbot answer. Upon successful completion, users are instantly rewarded with WLD tokens sent directly to their integrated wallet via the MiniKit SDK, with zero transaction fees on World Chain.For AI developers, Datagraph World offers a scalable and ethical way to acquire the authentic, sybil-resistant data needed to build safer and more aligned models. For users, it creates a new, accessible stream of digital income, fairly compensating them for their uniquely human intelligence. Ultimately, Datagraph World leverages the Worldcoin ecosystem to democratize the AI economy, ensuring humans are a core, rewarded component in the future of artificial intelligence.

Solution

Datagraph World is a Mini App built to directly connect the demand for high-quality AI training data with World App's vast network of verified humans. Our core principle was to leverage the Worldcoin ecosystem to create a seamless "train-to-earn" loop that is secure, efficient, and instantly scalable.Core Technologies Our tech stack was chosen for rapid development and seamless integration with the Worldcoin platform:Frontend: We used Next.js (React) with TypeScript and styled it with Tailwind CSS. This allowed us to build a responsive, type-safe, and modern user interface that feels native within the World App environment.Backend: A lightweight Node.js server with the Express.js framework handles our core logic. It's responsible for serving AI training tasks to users and validating their submissions.Database: For the hackathon, we used Supabase, a PostgreSQL-based backend-as-a-service. This allowed us to quickly set up a database to store and manage the data bounties and user task completions without managing server infrastructure.Core Integration: The Worldcoin MiniKit SDK is the heart of our project, connecting our application directly to the World App's native features.How It's All Pieced Together The user journey is powered by a straightforward but powerful technical flow:App Initialization: A user opens Datagraph World inside the World App. Our Next.js frontend is loaded in the Mini App webview.Proof of Personhood: To begin working, the user must verify they are a unique human. Our app triggers the verify Minikit SDK command. This prompts the user for World ID verification, and upon success, the World App returns a cryptographic proof to our app. This is the gatekeeper that ensures every data point comes from a real person.Task Delivery: Once verified, our frontend makes an authenticated API call to our Node.js backend to fetch an RLHF task (e.g., ranking two different AI-generated responses).Submission & Validation: The user completes the task and submits their feedback. The frontend sends this data back to our backend, which validates the input and stores it in our Supabase database, linking it to the user's proof of personhood.Instant Payout: Upon successful validation, our backend triggers the pay Minikit SDK command. This command instructs the World App to seamlessly transfer a predetermined amount of WLD tokens from our project's wallet to the user's integrated wallet. Thanks to World Chain, this transaction is free for the user.Leveraging the Worldcoin Ecosystem Using Worldcoin's partner technologies wasn't just an option; it was our core strategy. It provided immense benefits:Instant Sybil Resistance: The verify command is our silver bullet for data quality. AI labs are paying for human intelligence, and World ID guarantees they are getting it, eliminating bot-generated, low-quality data that plagues other platforms.Frictionless User Experience: We didn't have to build complex user authentication, wallet management, or payment systems. World App provides a trusted environment with an integrated wallet, and the MiniKit SDK gives us direct access to it. This allowed us to focus solely on the core task experience.Scalability & Distribution: Building on World App gives us immediate access to 23 million potential data trainers. This solves the "cold start" problem and makes our platform instantly valuable to AI labs seeking large-scale data sourcing.Zero-Cost Rewards: The ability to pay users with free transactions on World Chain is a huge advantage. It means every bit of value we provide as a reward goes directly to the user, with nothing lost to gas fees.A Notable Hack To ensure a snappy, near-instant user experience, we implemented a lightweight task pre-fetching mechanism. While a user is actively working on one RLHF task, our frontend makes an asynchronous API call to fetch and cache the next task in the background. The moment the user submits their current work and the payout is initiated, the next task is rendered instantly from the cache. This eliminates perceptible loading times between tasks, creating a fluid and engaging workflow that keeps users in the loop and maximizes their earning potential.