Jarvis
Jarvis is an autonomous AI agent that understands a user's risk level and optimizes yield by managing liquidity pool positions across ZKSync boosted LP Pools and restaking ETH on EigenLayer with P2P.
About
The problem it solves Interacting with DeFi protocols is complex and time-consuming (e.g. swapping, bridging, needing to monitor yield rates constantly). Jarvis abstracts this for the end user by directly interacting with the smart contracts of DeFi protocols and monitoring APR rates--all while keeping your unique risk profile in mind. To start, we're targeting the ZKIgnite Program, where 300M ZK tokens (~30M USD) are allocated to boost DeFi positions on ZKSync for the next 6 months. LP pools have some of the highest boosted APR in the program (50-80% APR). New opportunities are introduced every two weeks, and the total APR (boosted APR + native APR) changes every few days. Jarvis will keep track of all new opportunities, calculate the total APR, create new positions for LP pools with the highest APR, and remove liquidity from positions with APRs that have decreased over time. User Interaction and Data Flow 1. Verifiable Risk Profile: We securely verify your CreditKarma FICO credit score via ZkTLS 2. Review your yield strategy split determined by your Verifiable Risk Profile and set it off! 3. Jarvis will autonomously manage LP positions across ZKSync boosted LP Pools and restaked ETH, rebalancing when necessary. Sit back, relax, get yield. The project architecture and development process - ZkTLS for secure sharing of data (via Reclaim Protocol's web sdk) - Verifiable inference with Gaia - Hashing data to ensure integrity of centralized DB - P2P combined with custom smart contract to allow for fractional contributions to EigenLayer restaking node - 12 custom actions added to Coinbase Developer Platform's Agentkit for Jarvis to autonomously manage your money Product Integrations https://whimsical.com/verifiable-ai-agent-data-architecture-TJQ2VNUxYtV17pCoVQjckF Key differentiators and uniqueness of the project We're building an autonomous DeFi Agent: - People don’t have much time to to optimize yield - Chat interface AI agents solve complexity not time - Strategies are based off verifiable behaviors not flawed self-reported risk Trade-offs and shortcuts while building Right now, everything is running locally and we don't have any user management. We want to do the following in the future: - Deploy agent to the cloud - User management - Improve verified risk profile (ZkTLS for investment & spending accounts) - Future Integrations: Unichain & other EVM chains (Base, Arbitrum etc.), LST’s Additional Features N/A