Web3 & AI

SOLUTIONS

Products

Services

Web3 & AI

SOLUTIONS

Services

Products

Industries

Become Our Client

About Us

Resources

Web3 & AI

SOLUTIONS

Services

Products

Industries

  • Limited Seats: TokenMinds' RWA Tokenization Summit @ Consensus Hong Kong 2026

JOIN NOW

  • Limited Seats: TokenMinds' RWA Tokenization Summit @ Consensus Hong Kong 2026

    JOIN NOW

JOIN NOW

  • Limited Seats: TokenMinds' RWA Tokenization Summit @ Consensus Hong Kong 2026

Download

Agentic Payments: AI-Routed Checkout Process with Cryptographic Approval

Executive Summary

We built an AI-powered e-commerce platform for Project K. 

What makes it special is the integration of agentic payments on a private & permissioned blockchain.

Key features we built:

  • Agent-led checkout

AI agents prepare the payment, then request the user’s cryptographic approval.

  • Private blockchain rules

Payments run on a private network; smart contracts enforce the rules. Wallets are managed by the platform for a smooth experience.

  • Bank gateway settlement

After approval, the payment is executed through the bank’s Digi-Pay gateway (via Apigee).

Confidential Project

Industry: E-commerce and AI

Blockchain: Private Permissioned Blockchain

Client’s Objective

The client aims to launch an AI-powered e-commerce platform centered on agentic payments. Specifically, to:

  • Enable agent-led payments that prepare transactions and request cryptographic approval.

  • Run payments on a private blockchain with smart contracts for controlled, confidential processing.

  • Provide custodial wallets so the platform can manage keys. This also enables automatic updates to balances and loyalty points.

Client’s Challenges

  1. Unclear payment approval

Users can’t verify how a payment is approved. Approval needs to be explicit and cryptographic.

  1. Unsafe processing path

Payments aren’t protected by a private, rule-enforced ledger with custodial key management. This raises execution risk.

  1. Disjointed checkout steps

Product selection, wallet/loyalty updates, and payment aren’t coordinated. An agent-led handoff is needed end to end.

TokenMinds Solutions

Agentic payment flow

Front-man agent leads checkout, Bank-man agent prepares the payment, and the user gives cryptographic approval.

Private Tokenized Payments

Payments run on a private Hyperledger Fabric network. Chaincode enforces the rules, and users transact through custodial wallets.

AI Recommendations

Product-man uses the Data Lakehouse and Zary API to surface relevant products during shopping.

Digi-Pay Execution

After approval, settlement is executed through the bank’s Digi-Pay wallet gateway (via Apigee).

1. Integrated Agentic Payment Flow

We integrated a chat-led checkout. AI agents prepare the payment on the private ledger and request cryptographic approval in the same conversation.

Front-man agent guides checkout → Bank-man agent drafts transaction (sender/receiver/amount) → Cryptographic approval prompt → Payment executes on private ledger → Balances & loyalty update

What we built

  • Orchestration between Front-man agent (UX) and Bank-man agent (payment prep)

  • Transaction draft generator for the private ledger

  • In-chat approval prompt and signature capture

Why it matters

  • Approval is explicit and auditable.

  • Checkout stays in one conversation.

  • Post-payment updates are immediate.

2. Implement Private Tokenized Payments

We have implemented a controlled private blockchain. Within the system, smart contracts govern the issuance, transfer, and balance of tokens.

Wallet action → Chaincode checks rules → Transaction commits on private ledger → Balances update → Loyalty points reflect activity

What we built

  • Hyperledger Fabric network with permissioned access

  • Chaincode for issuance, transfer, and balance logic

  • Custodial wallet service integrated into the platform

Why it matters

  • Transactions are confidential and controlled.

  • Rules are consistent and tamper-evident.

  • Keys are managed for a seamless user experience.

3. Enable AI Recommendations

We built an AI layer for relevant product suggestions. It reads trusted signals and serves picks in real time.

Data Lakehouse signals → Product-man ranks products → Zary API pulls catalog items → Front-man agent shows recs in chat → User adds to cart

What we built

  • Product-man agent logic for ranking and selection

  • Zary API integration with the shopping catalog

  • Transaction-history analysis from the Data Lakehouse

Why it matters

  • Recommendations are timely and relevant.

  • Shoppers find items faster with less friction.

  • Discovery happens inside the existing chat flow.

4. Unify Checkout Execution

We connected the chat flow to payment rails. After approval, the platform executes payment via the bank’s Digi-Pay gateway.

Product picked → Cart summary in chat → Transaction drafted → Cryptographic approval → Digi-Pay settlement → Receipt + balances/loyalty update

What we built

  • Digi-Pay (Apigee) integration for settlement

  • Chat-to-payment handoff without leaving the flow

  • Post-payment records with wallet and loyalty updates

Why it matters

  • One continuous checkout reduces drop-off.

  • The settlement is bank-grade and auditable.

  • Post-purchase status is clear immediately.

 Key Improvements in Agentic Payment

Before

After

Opaque payment approval

Cryptographic approval inside the chat flow

Disjointed checkout steps

Agent-led, end-to-end conversational checkout

Public/loosely controlled rails

Private, permissioned ledger with smart-contract rules

Separate wallet & loyalty updates

Automatic balance and loyalty updates post-payment

Benefits of the Development

Secure Transactions

Payments run on a private, permissioned blockchain with smart-contract rules. Cryptographic approval makes authorization explicit.

Enhanced Shopping Experience

AI recommendations surface relevant products in real time. Signals come from the Data Lakehouse and the live catalog.

Seamless User Journey

Agents keep checkout inside one conversation from cart to approval. Fewer handoffs make the process faster and clearer.

Stronger Customer Loyalty

Balances and loyalty points update automatically after payment. Users see rewards reflected immediately.

Technology Stack

  • Git

  • GitHub

  • JAVA (Spring Boot)

  • TypeScript

  • Redis cache

  • PostgreSQL

  • Hyperledger Fabric

  • Golang

  • Node.js

  • LangChain

  • Docker & Kubernetes

  • MCP (Model Context Protocol)

  • ​​Chaincode (in Hyperledger Fabric)

  • Proprietary CoreAI Suite

    • Multi-AI-Agents Orchestration System

    • Product Recommender Module

  • Data Lakehouse Platform

  • Digi-Pay wallet gateway (a financial institute’s own wallet gateway)

  • Zary Shopping Platform API 

Development Process

  1. UI/UX Setup

Built a responsive Next.js interface with a chat flow for agent-led checkout.

  1. Agentic Checkout

Front-man agent guides the flow; Bank-man agent drafts the transaction; user gives cryptographic approval.

  1. Blockchain Integration

Deployed Hyperledger Fabric with chaincode for issuance, transfer, and balances; enabled custodial wallets.

  1. AI Recommendations

Implemented Product-man using Product Recommender Module from our Proprietary Core AI Suite, customer’s Data Lakehouse signals and Zary Shopping Platform API for relevant products.

  1. Payments Integration

Connected the bank’s Digi-Pay wallet gateway (via Apigee) for settlement.

  1. Backend Development

Built MCP services in JAVA. These MCP services handle product data and update balances and loyalty in PostgreSQL. Redis cache is used for session memory.

  1. Testing & Deployment

Ran end-to-end checks and deployed the integrated platform.

Anticipated Results

30% boost in user engagement

AI recommendations surface relevant products in chat. Interaction increases as shoppers see timely picks.

20% higher conversion rate

Agent-led checkout with cryptographic approval reduces drop-off. The path from cart to payment is clearer.

40% increase in operational efficiency

Private-ledger automation and custodial wallets cut manual steps. Teams spend less time reconciling payments and points.

Launch Agentic Payments with TokenMinds

Schedule a free call with us now