Today, we analyze Yzi Labs leading a $4 million AI project - Gata. In simple terms, this project did two things:

  • Turn ChatGPT chat history, your chosen AI response preferences, and idle computer computing power into a 'mining tool' that directly rewards you with points (which can be exchanged for tokens in the future);

  • Generate top-tier AI training data using a Web3 crowdsourcing model and sell it to companies like OpenAI and Stability AI for profit.

Currently, Gata has surged to the top of the BNB Chain ecosystem rankings and is also conducting a 100,000 points airdrop activity. Let's see how to participate:

1. Project Overview

Gata (formerly Aggregata) is a decentralized project dedicated to building cutting-edge artificial intelligence data, with the core goal of promoting the development of safe and human-centered 'super intelligence' through decentralized infrastructure.

图像

  • Project Financing - Completed a $4 million seed round financing, led by Yzi Labs under Binance, with follow-on investments from more than ten well-known institutions including IDG Blockchain, Maelstrom Fund, Manifold Trading, Gate Ventures, etc.

  • Platform Ranking - Ranked first on the BNB Chain's DappBay platform, becoming a leading project in Greenfield dApps.

图像

2. How to Participate

There are currently several ways to participate:

1️⃣ Start DataAgent (DVA):

  • Enter the Gata official website, (https://app.gata.xyz?invite_code=pg41zvgk)

    , connect wallet

图像

  • Follow the official Twitter and join DC

图像

  • Start DVA - Click Start DVA, the system will automatically assign tasks, assessing the quality of internet image-title pairs (scoring range -1 to 1). These data are used to filter high-quality visual-language model training sets.

图像

  • Accumulate Gata Intelligence points: After completing tasks, users earn points rewards based on data quality scores. Note that points distribution relies on a consensus mechanism - users can only receive rewards if the majority of nodes agree on the task results (the current Beta phase success rate is about 10%).

图像

2️⃣ Contribute Human Preference Data

  • Enter the chat from the left side, contribute 'human preference data' by selecting preferred responses, and accumulate Human Preference points.

图像

  • Enter from the left side Earnings to receive points

图像

3️⃣ GPT-to-Earn

  • After installing the Chrome extension, users can earn points by sharing ChatGPT conversation data. The system allocates rewards based on conversation quality (such as the depth of interaction with GPT-4 or GPT-4o).

图像

3. Participating Activities (BNBchain DappBay)

👉 Event Introduction: The BNBChain DappBay event is now open for participation, with a total reward of 100,000 Gata Intelligence points. The event features two leaderboards for running Gata DVA and inviting friends, with the top 50 on each leaderboard becoming winners (rewards can be won multiple times).

⚠️ Note: The event ends on May 12, 2025

Event declaration page: https://dappbay.bnbchain.org/campaign/384-join-gata-giveaway-to-share-100-000-intelligence-points

图像

👉 How to Participate

  1. Log in or register at Gata.xyz

    。

  2. Start running DataAgent.

  3. Invite friends.

4. Official Document Analysis

1️⃣ Data Production Product Matrix

1. GPT-to-Earn:

  • Value Logic: Data generated by ChatGPT users is the gold standard for 'human-AI interaction', but in traditional models, users cannot share data dividends. Gata captures conversation data through plugins, feeds back to AI companies for model optimization, while also distributing points to users.

  • Technical Details: Data contributions must be manually selected through an extension to ensure user privacy and data compliance.

2. All-in-One Chat:

  • Aligning super intelligence: By comparing the responses of three major models and recording user preferences, Gata has built a human preference dataset, which is the core resource for achieving 'super alignment'.

  • Commercialization Path: In the future, datasets can be licensed to AI companies to train models that better align with human values.

3. DataAgent: DVA:

  • Addressing Industry Pain Points: The current internet image-title datasets (such as LAION) face two major issues - public data exhaustion and title noise. DVA filters high-quality data through distributed scoring while exploring synthetic title generation (such as generating accurate descriptions based on GPT-4V).

  • Technical transparency: Task inputs and outputs are displayed in real-time, ensuring user oversight of the data processing flow.

2️⃣ Points System and Economic Model

  • Points distribution mechanism: Dynamically adjust weights based on data contribution types (such as ChatGPT conversations, human preference selections, DVA tasks), reflecting the market value of different datasets.

  • Tokenization design: Gata Points can be exchanged for project tokens in the future, and holders will share dataset licensing revenue and product development profits.

3️⃣ Business Model

1. Data licensing and access fees - Charging AI companies for the use of datasets, with tiered pricing based on scope (research/commercial/exclusive). Currently, we have accumulated 5.3 million pieces of ChatGPT user data, 106,000 pieces of GPT-4o reasoning data, and a vast amount of human preference data.

2. Jointly develop AI products - Based on exclusive datasets (such as upcoming unpublished data), Gata plans to co-develop vertical AI applications with partners, profiting through subscription or revenue sharing. For example, a 'alignment optimization model' trained on human preference data may become its first commercial product.