joe-y

joe-y is an AI-powered personal finance agent that gives users a single conversational interface to manage both crypto and traditional money. Instead of bouncing between wallets, banking apps, DeFi dashboards, and block explorers, users can simply talk to joe-y and ask for what they need: check balances, review portfolio positions, inspect transaction history, see vault deposits and APY, swap tokens, move funds, and make transfers. joe-y is especially designed for the World ecosystem, so users can sign in with World App and benefit from identity-aware financial features such as boosted vault rewards for higher verification levels, including Orb-verified users.

On top of its onchain capabilities, joe-y can also access and manage a user’s bank account, making it possible to monitor balances, handle day-to-day financial operations, and execute transfers from the same chat experience.

The result is a unified financial copilot that combines AI, crypto, identity, and banking into one product.

We built joe-y as a full-stack monorepo with a modern web frontend, an agentic backend, and supporting smart contract infrastructure.

The frontend is built with Next.js, React, and Tailwind CSS, and provides the login flow, dashboard cards, chat interface, vault views, and agent profile modal.

For onboarding and identity, we integrated World App / MiniKit so users can authenticate with the World ecosystem and unlock verification-aware experiences.

The backend is built with Hono on Node.js and powers the agent workflow, API routes, and data access.

For the intelligence layer, we used Anthropic models in a tool-calling loop so the assistant can do more than answer text prompts: it can retrieve wallet data, reason over user context, and trigger finance actions.

We used Supabase to persist users, wallets, verification level, and chat history so the agent can maintain continuity across sessions.

For blockchain functionality, we used viem, @worldcoin/agentkit, and x402 to connect agent behavior with onchain infrastructure across networks like Worldchain and Base, and we included a contracts layer with Hardhat and OpenZeppelin for vault/token-related logic.

One of the more interesting parts of the build is how we unify multiple financial surfaces behind a single conversational layer: the user interacts with one AI agent, but under the hood that agent is coordinating identity, storage, wallet operations, vault logic, and banking access through TuleBank-style skills so the experience feels seamless instead of fragmented.