• 

  PayPal integration

• 

  Order Confirmation

• 

  Order parsing

PayFlow Copilot – AI-Powered Order to Payment Assistant

Inspiration

Small businesses often receive customer orders through informal messages. Converting these messages into structured orders and payment requests takes time and can lead to mistakes.

We built PayFlow Copilot to simplify this process using generative AI and PayPal.

What it does

PayFlow Copilot transforms natural-language order messages into structured, reviewable payment requests.

• Understands customer orders written in everyday language.
• Extracts product details, quantities, prices, and customer information.
• Calculates the total order amount.
• Allows sellers to review and confirm AI-generated details.
• Integrates with PayPal Sandbox to initiate payment checkout.

For example, a seller can enter:

"2 handmade candles at $18.50 each for Priya"

PayFlow Copilot interprets the order and calculates a total of $37.00.

How we built it

We developed a web application integrating Google's Gemini AI with PayPal Sandbox APIs.

The workflow follows four stages:

1. Seller enters a natural-language order.
2. Gemini AI extracts structured order information.
3. Seller reviews and confirms the order.
4. The application initiates PayPal Sandbox checkout.

We also used Playwright for browser-based testing and validation.

Challenges we ran into

One challenge was handling Gemini API free-tier rate limits while developing and testing the application.

Another challenge was integrating PayPal Sandbox checkout and managing payment approval flows.

We addressed these challenges through debugging, validation, and iterative testing.

Accomplishments

• Built an AI-assisted order interpretation workflow.
• Integrated structured order review and confirmation.
• Connected PayPal Sandbox order creation.
• Developed a demonstration of the order-to-payment workflow.

What we learned

We gained practical experience integrating generative AI with payment APIs, validating AI-generated financial information, and developing user-controlled payment workflows.

What's next

We plan to improve payment error handling, expand automated testing, add order history, and support more complex customer orders.

Built With

• javascript
• node.js
• playwright