Build Your Crypto AI Agent with Bankr and Zerion

Crypto AI agents that can earn online are no longer theoretical. With Bankr and Zerion API, you can set one up in minutes.

This guide walks through the process: what a crypto AI agent actually is, why Bankr and Zerion API fit together, and how to get your agent running.

Crypto AI agent basics

At the core, crypto AI agent is an LLM-powered program with its own wallet. 

Unlike a simple chat bot, it can check balances, execute trades, and respond to onchain events. And instead of rigid algorithms, all decisions and tool use is done through a LLM like ChatGPT, Claude, or Gemini. 

High-level overview

Building one takes three steps:

1. Set up OpenClaw. OpenClaw is an open-source AI agent framework (formerly known as Clawd and Moltbot). It runs locally and accepts plug-in skills that extend its capabilities.
2. Give it crypto skills. Install the Bankr skill for trading and wallet management, and the Zerion skill for rich onchain data such as portfolio balances, DeFi positions, transaction history, PnL, and more.
3. Fund it. Send a small amount of crypto to the agent's dedicated wallet. Bankr sponsors gas on supported chains, so the agent can start transacting with minimal upfront cost. For a self-sustaining setup, you can also launch a token via Bankr and trading fees will flow back to your agent's wallet automatically.

The result: an agent that can reason about any wallet's portfolio, execute swaps across multiple chains, and even fund its own LLM inference.

Why Bankr: self-sustaining agent infrastructure

Bankr provides the financial rails for autonomous agents. 

Every agent gets a cross-chain wallet that works across Base, Ethereum, Polygon, Unichain, and Solana. Gas fees are sponsored on supported chains, so your agent can start transacting immediately.

Bankr gives your agent real DeFi capabilities: spot swaps, limit orders, DCA strategies, stop-losses, leveraged trading, NFT management, and even Polymarket trades. Everything is done via natural language.

Bankr agent can be expanded with different skills from its own skills marketplace. 

Why Zerion API: onchain data for your AI agent

Trading is only half the picture. An agent also needs to understand what's happening onchain. That's where Zerion API comes in.

"Agents start with a wallet but that’s just the foundation. Bankr turns that wallet into an execution layer and with Zerion as a skill agents gain full financial awareness across portfolios DeFi positions and PnL. They can see decide and act without touching RPCs or indexers." — Igor Yuzovitskiy, DevRel at Bankr

Zerion API provides normalized, interpreted wallet data across all major EVM chains and Solana through a single integration:

• Portfolio balances and token prices across 40+ chains
• DeFi positions from over 8,000 protocols, including lending, staking, liquidity pools, etc
• Transaction history decoded into human-readable actions
• PnL tracking per wallet and per token
• NFT holdings with metadata and floor prices

Raw blockchain data is hard for LLMs to work with. Different formats per chain, no labels, no USD values. Zerion API handles that normalization. Your agent receives clean JSON with token symbols, dollar amounts, and protocol names, which means fewer wasted tokens and faster reasoning.

Zerion API already powers onchain data at scale for AI agents like AskGina and HeyElsa.

Payments via x402

Zerion API also supports x402 payments, and Bankr x402 Cloud lets agents deploy paid endpoints, handle payments, and make services discoverable.

An agent with a Bankr wallet on Base can call any Zerion API endpoint, pay $0.01 USDC per request, and receive structured data back. No signup, no API key, no human onboarding.

This is particularly useful for autonomous agents. Instead of requiring a human to register, set up billing, and manage usage, the agent pays per request from its own wallet. Combined with Bankr's self-funding model, your agent can earn trading fees and spend them on Zerion data, and do even more onchian, creating a flywheel and moving towards autonomy.

While all this might sound complex, it can all be done by just asking your agent (nicely). 

Steps to create your AI agent

Here are the detailed instructions for how to set up your own AI agent for crypto:

1. Install OpenClaw. Follow the setup at openclaw.ai to get the agent framework running on a VPS or locally (it’s safer to run it on a separate machine)
2. Install the Bankr skill. Tell your OpenClaw agent: install the bankr skill from https://github.com/BankrBot/skills This gives your agent wallet management, trading, and DeFi capabilities across Base, Ethereum, Polygon, Unichain, and Solana.
3. Install the Zerion skill from Bankr marketplace. Tell your OpenClaw agent: install the zerion skill from https://github.com/BankrBot/skills/tree/main/zerion This enables your agent to query portfolio data, DeFi positions, transaction history, and token prices via Zerion API.
4. Fund and test. Ask the agent for its Bankr address on Base. Then send a small amount of USDC and try some commands:
   • "What's your portfolio worth?"
   • "Buy $1 of ETH on Base"
   • "What's the PnL for wallet 0x...?"
5. Optional: launch a token and register your agent. You can just ask your agent something like "deploy a token called X with a ticker Y”. And it will launch a token. Then you can also create a profile for your agent in the Bankr agent directory. 

Your turn

You now have everything you need to build a crypto AI agent that reads onchain data, executes trades, and can potentially even fund itself.

Create free Zerion API keys to start pulling wallet data for your agent. Explore the Zerion’s documentation about building with A for MCP, x402, and CLI integration options.