TrustDAI - Trusted Decentralized AI
Own your AI data—control, sell, and track who accesses it with decentralized storage and smart contracts. Take back your digital identity
About
The problem it solves ## The Problem It Solves We're used to AI and online services learning about us—Spotify knows your music taste, TikTok curates content for you, and Amazon predicts your next purchase. But do you actually **own** that data? Your AI memories? Right now, your behavior data is locked inside these platforms. If TikTok disappeared tomorrow, your AI memories would be lost. Worse, companies collect, sell, and profit from your data without your consent, and you never truly know who has access to it. ### What This Changes With our system, **you own your AI memories**. Instead of platforms harvesting it for free, you **store it securely on decentralized storage** and **control access through smart contracts**. - 🔍 **Transparency** – Know exactly **who** accesses your data and **why** through an open, verifiable system. - 💰 **Monetization** – Sell your insights directly to companies **on your terms**, cutting out the middlemen. - 🔄 **Portability** – Take your preferences, AI history, and behavioral insights with you across platforms. - 🔒 **Security** – Your data is encrypted, protected with decentralized keys, and cannot be exploited without your consent. - 📊 **AI Knowledge Ownership** – Keep the data that AIs generate about you and add it to your profile, building your preferences over time. You decide which preferences can be shared and with whom. ### How It Works Marketers, AI services, and businesses already pay for user data—but now, **they buy it from you**. - 🤖 An **LLM** can instantly tailor responses based on your **self-owned** knowledge base, making AI interactions smoother and more relevant. - 📑 An **insurance agent** can autofill a quote using only **what they need**, purchasing access to specific data points you authorize. - 🎵 A **music app** can instantly learn your preferences without tracking you—because **you provide the data yourself** instead of being passively monitored. User Interaction and Data Flow ## User Interaction and Data Flow ### 1️⃣ Account Creation Users (or an AI agent) connect their **Web3 wallet** to the **dApp**, creating their account. - The user enters optional **basic information** (e.g., age, hobbies, travel, food). - Only a **name** is required—other data is added over time. ### 2️⃣ AI Behavior Profile Growth As users interact with **LLMs, music services, social media, shopping, or government sites**, their **profile updates automatically**. - Data is organized into **JSON categories** for controlled sharing. ### 3️⃣ Secure Storage & Encryption - Data is stored on **EthStorage** but is **unencrypted** by default. - **Lit Protocol** encrypts each category, ensuring privacy. ### 4️⃣ Granting Third-Party Access 1. 🔔 A third party **requests data** via the web app. 2. ✅ The user **approves or denies** the request. 3. 🔓 Approved data is **decrypted** with **Lit Protocol** and shared. 4. 🔄 The third party **uses the data** in their application. ### 5️⃣ Profile Updates After Interactions - New insights from the interaction **update the user’s profile**. - The updated data is **re-encrypted** and stored securely. - Users **can always view and manage their profile**. ### 🔐 Full Control & Transparency Users always know **who accessed their data**, **what was shared**, and **how their profile evolves**—ensuring **true ownership** of AI-driven personal data. The project architecture and development process ## Project Architecture and Development Process ### 🏗️ Overview This project enables **user-controlled AI data ownership**, using **decentralized storage, encryption, and smart contracts**. Instead of platforms passively collecting user data, users store their own AI-generated insights and **control access through Web3 mechanisms**. ### ⚙️ Core Functionality 1. **User Profile & AI Behavior Data** - Connects via **Web3 wallet** to create an account. - Profile grows dynamically as users interact with **AI, music, social media, and government services**. 2. **Secure Storage & Encryption** - **EthStorage** stores user data **immutably**. - **Lit Protocol** encrypts profile data, ensuring privacy. 3. **Smart Contract Access Control** - Users grant **fine-grained permissions** for data access. - **Blockchain logs all interactions**, ensuring transparency. 4. **Permission-Based Data Sharing** - Third parties request access via the **web app**. - Users **approve, deny, or sell access** based on need. - Decrypted data is shared via **Lit Protocol**. 5. **Profile Updates After Interactions** - AI-generated insights update the **user’s profile**. - Data is **re-encrypted** and stored securely. - Users always have access to their **full data history**. ### 🏛️ Architectural Components - **Frontend (React + Web3)** - User authentication via **MetaMask, etc**. - Data management dashboard for approvals & permissions. - **Backend (Node.js + Smart Contracts)** - **Solidity smart contracts** control access & payments. - Integration with **EthStorage & Lit Protocol**. - **Blockchain Layer (Ethereum + Smart Contracts)** - Stores **permissions and access history** for transparency. ### 🔨 Development Process 1. **Smart Contract Development** – Solidity contracts for data access and monetization. 2. **Storage & Encryption** – Deploying **EthStorage** and **Lit Protocol** Product Integrations ## Product Integrations This project integrates **key Web3 and AI services** to ensure **secure data storage, encrypted access control, and seamless user interactions**. Below are the core APIs and services used and their roles: ### 🔑 **Authentication & Wallet Integration** - **MetaMask / WalletConnect** – Enables users to **authenticate and sign transactions** using their Web3 wallets. - **Ethereum Blockchain** – Hosts **smart contracts** for permission control, monetization, and audit logging. ### 🗄️ **Decentralized Storage & Encryption** - **EthStorage** – Provides **immutable, decentralized storage** for user AI behavior profiles. - **Lit Protocol** – Encrypts stored data and manages **decryption keys** for controlled access. ### 🤖 **AI & Personalization** - **ElizaOS** – Manages **AI interactions**, processes user behavior data, and refines **AI behavior profiles**. - Supports **plugin-based integrations** with services like **WhatsApp, Discord, X**, and other communication platforms. ### 🔗 **Smart Contracts & Access Control** - **Solidity Smart Contracts** – Governs **data access, payments, and sharing approvals** on Ethereum. ### 💰 **Monetization & Payments** - **Ethereum / Layer 2 Solutions** – Supports **fast, low-cost transactions** for buying/selling AI data access. ### 📡 **Third-Party Access & Permissions** - **REST API** – Provides a **developer-friendly interface** for third parties requesting data. By combining **ElizaOS, blockchain, decentralized storage, and smart contract access control**, this project **gives users full ownership over their AI-driven data** while enabling **secure, monetized sharing**. 🚀 Key differentiators and uniqueness of the project ## Comparison with Similar Projects ### 🔹 **Your Project** - **Focus:** AI behavior data ownership & monetization. - **Data Ownership:** Users fully control and sell their AI-generated data. - **Storage:** **EthStorage** for decentralized storage. - **Encryption:** **Lit Protocol** for secure, permission-based access. - **Monetization:** Smart contract-based AI data sales. - **AI Integration:** **ElizaOS with plugin support** (WhatsApp, Discord, etc.). - **Transparency:** Users track **who accessed their data**. --- ### 🔹 **Ocean Protocol** - **Focus:** Open data exchange for businesses & researchers. - **Data Ownership:** Data providers share datasets but **mostly for open use**. - **Storage:** Uses **IPFS & blockchain**. - **Encryption:** Optional, but **not a focus**. - **Monetization:** Uses **data tokens** for dataset trading. --- ### 🔹 **AIOZ Network** - **Focus:** Decentralized AI dataset & model hosting. - **Data Ownership:** Users sell **datasets & AI models** in a network. - **Storage:** Uses **AIOZ blockchain & decentralized nodes**. - **Encryption:** Users **manage access manually**. - **Monetization:** **AI developers license & sell models**. - **AI Integration:** Focuses on **AI model hosting** --- ### 🔹 **DECORAIT** - **Focus:** AI training **opt-in/opt-out registry** for content creators. - **Data Ownership:** Creators control **how AI models use their work**. - **Storage:** **Blockchain-based metadata registry**. - **Encryption:** No encryption, just **tracking**. - **Monetization:** **Royalties for AI training contributions**. - **AI Integration:** **Manages content rights**, not personal AI behavior. --- ### 🚀 **Key Differentiators** ✅ **AI-Specific Data Ownership** – Unlike **AIOZ**, your project **manages AI-generated personal behavior data**. ✅ **End-to-End** encryption Trade-offs and shortcuts while building ## Trade-offs and Shortcuts While Building ### ⏳ **Access Permissions for Agents** We made a trade-off by **simplifying the access control system** for AI agents. Instead of a granular permission structure, we opted for **basic approval mechanisms**, allowing users to grant or deny access to entire block of data. Future improvements include **fine-grained permissions per agent** ### 🔍 **Identity System Limitations** A more robust **identity verification system** using **verifiable hashes** was planned but not implemented due to time constraints. This would have ensured **cryptographic proof of identity**, making AI behavior data more **trustworthy and portable**. Future versions will explore **DID (Decentralized Identity) integration**. ### 🧪 **AI Testing & Group Interaction** We couldn’t **fully test AI behavior** in group scenarios. While AI agents function well individually, **multi-agent interactions (e.g., travel assistant working with real estate assistant)** remain **untested**. Further testing will improve **collaborative AI decision-making**. ### 🛒 **Marketplace for Data Monetization** The initial vision included a **marketplace** where users could **sell or license their AI-generated data** to businesses or AI systems. We didn’t have time to implement this, but a **future version** could include **smart contract-based transactions** for direct user-to-business AI data exchanges. ### 🔄 **AI Writing Back User Behavior Data** Currently, AI agents can **read user preferences** but **do not write back new behaviors**. Ideally, AI should update a **user's stored preferences** dynamically based on real interactions, ensuring **profiles evolve over time**. Future iterations will allow AI agents to **write back behavior updates securely**, while maintaining **user oversight and edit rights**. Additional Features Our project was started at the Buildathon, but we had some ideas bouncing around on how to do storage with IPFS, but ethstorage and Lit solved a lot of the security methodology we were looking for so data remained private and sharable.