BlockAnalytica
decentralizing zachXBT with a swaof AI Agents
About
The problem it solves Our AI-powered agent streamlines on-chain forensic research by leveraging a swarm of specialized AI agents that collaborate to deliver the most accurate answers, backed by on-chain proof and rigorous methodology. What People Can Use It For: • On-Chain Investigations – Identify token holdings, linked wallets, and multi-chain addresses (e.g., Solana, Ethereum) effortlessly. • Due Diligence & Risk Assessment – Spot potential scams, wash trading, or sybil attacks before interacting with a wallet. • Portfolio Insights – Track wallet movements, side wallets, and token activity without manually searching multiple explorers. • Automated Alerts & Monitoring – Get notified of suspicious transactions, new holdings, or changes in wallet behavior. How It Makes Tasks Easier & Safer: • No More Manual Searches – Skip navigating Etherscan, Arkham, or Dune; just query in natural language. • Cross-Chain Intelligence – Automatically link wallet identities across chains to uncover hidden connections. • Contextualized & Actionable Insights – Get pre-defined prompts to extract only relevant data, reducing noise. • Decentralized & Trustless – Unlike centralized analytics platforms, results are generated by an open network of AI agents, making the process more transparent and verifiable. This tool democratizes access to on-chain intelligence, making investigations, research, and security assessments faster and more efficient. User Interaction and Data Flow Users may interact with it from a web interface or a smart contract (AVS) or a telegram bot or on twitter by tagging the agent on X. Right now only web interface and smart contract are supported. web: https://blockanalytica.vercel.app/chat avs: https://holesky.etherscan.io/address/0x6116dfc1e59cf0ebfa03dbd1da70c9225227f075#writeProxyContract The project architecture and development process The project is an agentic backend built with LangChain where a Supervisor Agent coordinates the use of one or more specialized agents. While the current architecture runs all agents within the same runtime, the long-term vision is to transition to a swarm of self-deployed AI agents, each solving highly specific tasks. In this model, the Supervisor Agent dynamically activates the appropriate solvers as needed. We currently suppport several scenario: find side wallets, analyse copy trading activity, retrieve historical ens domains... Product Integrations We use ORA for LLM inference, Hedera to demonstrate on-chain capabilities of the agent, and Coinbase’s AgentKit to showcase trading strategies derived from its insights. The agent is deployed as an EigenLayer AVS, with execution data, including on-chain proofs, pushed to EigenDA. Key differentiators and uniqueness of the project This project is highly inspired by Arkham Intelligence, BubbleMaps, Dune, Stalkchain, and similar tools, all focused on enhancing on-chain data readability. Our AI Agent takes this further by making blockchain intelligence more accessible through natural language interaction. However, our ultimate goal is to create a seamless, non-chatbot experience, enabling users to explore on-chain data intuitively—much like navigating links on Arkham or Etherscan, but with AI assistance anticipating their next moves. The key difference is that our platform will always present data in a user-friendly, visually engaging way, potentially with beautifully designed components for seamless exploration. Trade-offs and shortcuts while building The project’s codebase is highly entangled, with both backend and frontend components. The backend is not fully optimized, leading to high response times. Moving forward, the goal is to stream intermediate steps, keeping users informed of progress in real time and enhancing the visual and interactive experience. Additional Features The project has been fully built for this hackathon.