Nigel, your Web3 DeFi AI Companion on Telegram
Snow is a Telegram-based AI agent built to onboard DeFi newbies and help them make smart and safe investments in the Web3 world. Nigel—the personality behind Snow—runs on ElizaOS and leverages real-time market sentiment data from Chainlink to guide users through the AVAX C-Chain (starting with Fuji testnet).
- Conversational AI: Nigel uses ElizaOS to assist users in a friendly and approachable way, directly via Telegram.
- DeFi Coach: Nigel suggests optimal liquidity pools based on market conditions, APY, and risk.
- Market-Aware: Integrates with Chainlink's Fear and Greed Index to assess sentiment and adjust strategies accordingly.
- Smart Contracts: Custom LP contracts on Fuji testnet simulate real DeFi investment environments.
- Autonomous Strategy: Nigel calculates the best pool and reallocates funds when market conditions change.
- Cross-Chain Ready: Architecture designed to scale to other chains using Chainlink CCIP and ICM.
Nigel connects various decentralized components to provide intelligent, real-time investment advice:
-
Fetch Market Sentiment:
- A Chainlink oracle fetches the Fear and Greed Index from off-chain data.
- The index value is stored on-chain in a dedicated smart contract.
-
Analyze Pools:
- Nigel considers a list of LP pools with:
- APY (Annual Percentage Yield)
- Risk Rating (1 for stable-stable, <1 for stable-alt)
- Nigel dynamically adjusts preferences:
- Greed (High Index): Boost preference for stable-stable pools.
- Fear (Low Index): Boost preference for altcoin exposure.
- Rising Index: Increase allocation to alts (FOMO behavior).
- Falling Index: Shift toward stables (capital preservation).
- Nigel considers a list of LP pools with:
-
Decision Making:
- Nigel computes a weighted score per pool:
Score = APY * Risk Ratio * Sentiment Adjustment - Funds are allocated to the pool with the highest score.
- If market sentiment shifts, Nigel triggers rebalancing.
- Nigel computes a weighted score per pool:
A central feature of Snow is its use of the Fear and Greed Index, provided via Chainlink's decentralized oracle network:
- Nigel reads the current sentiment from an on-chain FearAndGreed contract, which gets updated via Chainlink feeds.
- The contract stores:
- The latest index value (0–100)
- A timestamp
- Historical trend (to track rising/falling sentiment)
This sentiment directly influences investment logic:
- Low index (fear): Favor riskier assets with potential upside.
- High index (greed): Prefer safer, stable options.
- Trending Up: Nigel assumes FOMO and tilts toward alts.
- Trending Down: Nigel hedges in stables to preserve capital.
In the 2-day hackathon, we focused on building a working MVP with:
- Telegram bot with ElizaOS integration (Olaf).
- On-chain Chainlink-powered Fear and Greed index (on Fuji).
- Two fake LP contracts with dummy APY and types (stable/stable vs stable/alt).
- Nigel’s decision-making logic that:
- Reads the current index.
- Calculates adjusted scores for each LP.
- Invests in the best pool.
- Rebalances when the index changes.
- Full Cross-Chain Deployment: Expand from Fuji to Ethereum Sepolia via Chainlink CCIP and other Avax L1 like Dexalot with ICM.
- Real LP Integration: Pull live APYs and risks from DeFi protocols.
- On-Chain Governance: Let users vote on Nigel’s behavior models.
- Portfolio Tracking: Telegram-based dashboard for holdings and performance.
- Auto-Staking: Let Nigel auto-stake/reinvest yields.
- AI: ElizaOS
- Blockchain: Avalanche Fuji, Chainlink Oracles, Custom Solidity Contracts
- Telegram Bot: Node.js (Typescript)
- Chainlink CCIP & ICM: For cross-chain interactions (coming soon)
Built with love and caffeine by 2Guys1Laptop at Avalanche Hackaton London 2025.

