AI Integration Overview

The AI Checkout interface enables you to build AI Agents (Chatbots, Copilots, or Autonomous Agents) that can orchestrate crypto payments.

Architecture & Responsibility

To integrate AI Checkout, you are building an Agent Runtime that acts as the bridge between the User's Wallet and the GStable API.

Unlike a standard checkout where we host the UI, you are responsible for the "Body" and "Brain" of the Agent, while we provide the "Payment Logic" and "Settlement Rail".

Responsibility Matrix

Feature	Who Builds It?	Description
User Interface	You (Developer)	The chat window, wallet connection button, and transaction confirmation cards.
Agent Runtime	You (Developer)	The backend or client-side logic that calls our API, parses responses, and constructs EIP-712 messages.
Wallet Connection	You (Developer)	Integration with libraries like wagmi, ethers.js, or viem to request signatures from the user.
Payment Logic	GStable SaaS	Validating products, calculating prices, handling cross-chain routing, and generating transaction calldata.
Settlement	GStable SaaS	Monitoring on-chain events and settling funds to your merchant account.

Interaction Diagram

The following diagram illustrates how your Agent Runtime coordinates between the User and the GStable API.

Integration Core Concepts

1. The "AI View"

Our API returns a special aiView object designed for LLMs. It contains:

• Semantics: What the product is (without HTML formatting).
• Workflow: What the Agent should do next (e.g., "ask user for address" or "request signature").
• AuthorizationSpec: How to construct the EIP-712 message for the user to sign.

2. EIP-712 Signatures

We use EIP-712 for all state-changing actions. This ensures that the Agent cannot spend user funds or create orders without explicit, human-readable consent from the user's wallet.

You will need to implement a signing handler in your runtime using standard Web3 libraries.

Next Steps

Ready to build? Go to the Integration Workflow guide for a code-first tutorial.