AetherGuard
AI cannot act without YOU!
About
The problem it solves # AetherGuard — Privacy-First AI Portfolio Guardian ## Overview **AetherGuard** is a platform that helps users get **AI-powered trading suggestions** while keeping their **investment strategies completely private**. Instead of sending strategies to AI providers, AetherGuard only shares the **user’s investment goal**. All actual strategies and guardrails remain private, yet **cryptographically verifiable**. The system leverages: - **Merkle proofs** to verify strategy compliance - **Fileverse encrypted documents** for secure and decentralized storage This ensures **AI can assist users** without ever seeing their private strategy. --- ## The Problem Most AI trading assistants today require users to send: - Portfolio details - Investment strategies - Risk and trading rules to centralized systems, which creates major risks: | Problem | Description | Risk | |---------|------------|------| | **Strategy Leakage** | Trading logic is visible to AI providers | Intellectual property theft or misuse | | **Centralized Control** | Platforms store user data in central databases | Users lose ownership and privacy | | **No Verification** | Users cannot prove AI followed their strategy | Black-box AI reduces trust and transparency | Even if users want privacy, traditional tools **do not allow strategy confidentiality**. --- ## Real-World Example A trader manages **$100,000** crypto portfolio with these rules: | Guardrail | Limit | |------------|-------| | Max allocation per asset | 30% | | Minimum liquidity | 20% | | Buy only high-volume assets | Yes | **Without AetherGuard:** AI platforms require sending the rules → full strategy is visible. **With AetherGuard:** AI only sees the goal and portfolio state. It can propose trades but cannot access private rules. Example workflow: | Portfolio (Before) | AI Suggestion | Portfolio (After Simulation) | |------------------|---------------|-----------------------------| | 100% Cash | Buy 20% ETH | 20% ETH, 80% Cash | | 20% ETH, 80% Cash | Buy 10% BTC | 20% ETH, 10% BTC, 70% Cash | --- ## The AetherGuard Solution ### 1. Strategy Creation Users define: - **Investment Goal** → Sent to AI - **Private Guardrail Rules** → Stored in **Fileverse encrypted documents** Rules are converted into a **Merkle tree**, and the **root hash** is used for verification. --- ### 2. AI Market Suggestions The AI only receives: - **Current portfolio state** - **Investment goal** It analyzes market data and generates suggestions: - Example: “Buy 20% ETH with available cash” AI **never sees strategy rules**. --- ### 3. Guardrail Verification Before execution: - Each AI suggestion is checked against the **Merkle root** - Violations trigger **UI warnings** - Users can **approve or reject trades on-chain** --- ### 4. Permanent Audit Trail All strategy and decision data are stored in Fileverse: | Document | Purpose | |----------|---------| | Strategy Document | Stores private guardrails | | Master Document | Aggregates all strategies of a user | | Decision Log | Records AI proposals, approvals, and executed trades | This ensures **transparent, tamper-proof auditability**. --- ### 5. Collaborative Multi-Sig AetherGuard enables **team portfolio management**: - Teams create a **shared Fileverse document** - AI proposals are reviewed by multiple members - Execution can require **multi-signature approval** This is ideal for **DAOs, treasuries, or collaborative trading teams**. --- ## Why AetherGuard AetherGuard allows users to: - Maintain **private strategies** - Receive **AI market insights** - Verify AI suggestions **cryptographically** - Approve trades **on-chain** - Keep a **permanent decentralized audit trail** > It combines AI assistance with **full privacy, security, and user control**, making AI-powered trading safe and trustworthy.
How it's made
Challenges we ran into # Challenges We Faced Building AetherGuard involved solving several key challenges around **privacy, collaboration, and portfolio simulation**. --- ## 1. Verifying AI Suggestions Without Revealing Strategy We wanted the AI to suggest trades **without ever seeing the user’s private strategy**. Normally, AI systems require the full strategy to know if a trade is allowed, but sending it to the AI would break privacy. **Solution:** We used **Merkle Root Proofs**: - Each strategy rule is hashed and stored in a **Merkle tree**. - The **Merkle root** acts as a fingerprint of the strategy. - AI suggestions are checked against this root to verify compliance **without revealing the rules**. - We also show the verification process in the **UI** for transparency. --- ## 2. Collaborative Multi-Signature Trading For team-managed portfolios or DAOs, multiple people need to review and approve trades. **Challenge:** How to enable **secure team collaboration** while preserving privacy and avoiding a single point of failure. **Solution:** We built **Fileverse-based multi-sig boardrooms**: - A **team document** stores proposals. - Members can review suggestions. - Trades require **multi-signature approval** before execution. This ensures **no single person—or AI—can move funds independently**. --- ## 3. Portfolio Simulation & Visualization As students, we don’t have access to real mainnet funds or large testnet portfolios. **Challenge:** How to **demonstrate trading flow** and AI decision-making realistically without risking real assets. **Solution:** - We use a **simulated portfolio**, starting with **100% cash**. - When AI suggests a trade (e.g., buy 20% ETH), the portfolio updates **virtually**: 20% ETH, 80% cash. - AI can read this simulated portfolio to propose **strategically meaningful trades**. - Execution still requires **user approval**, ensuring safety. This allows us to **show the full AI workflow** in the demo even without real funds. --- ## Summary The three main challenges we overcame: 1. **Strategy privacy vs verification** — solved with Merkle proofs. 2. **Collaborative multi-sig trading** — solved with Fileverse team documents. 3. **Portfolio simulation for demo purposes** — solved with a safe cash-to-simulated-asset system. These solutions ensure that AetherGuard **maintains privacy, security, and transparency**, while still providing a realistic AI-assisted trading experience.