-
-
home page
-
AI dashboard
-
AI Research results
-
Features
-
Integrated with Paypal and payment options
About PayPilot AI
Inspiration
Online shopping often overwhelms users with too many choices, confusing product comparisons, and the challenge of staying within a budget. We wanted to make shopping smarter and simpler by building an AI-powered personal shopping assistant that understands what users need instead of making them search through endless product listings.
This idea inspired us to build PayPilot AI — Your Intelligent Shopping & Payment Copilot.
What it does
PayPilot AI transforms a simple shopping goal into a personalized product bundle. Users describe what they need and enter their budget. The AI analyzes their intent, selects relevant products from the catalog, checks the total cost, and explains why the bundle is suitable.
Key features include:
- AI-powered recommendations: Uses Google Gemini to understand user requests and select relevant products.
- Smart budget management: Keeps recommendations within the user's specified budget.
- Personalized bundles: Combines useful products based on the user's goal.
- PayPal integration: Integrates PayPal Sandbox Orders for checkout preparation and payment capture.
- Order tracking: Stores order activity and transaction status using SQLite.
How we built it
We developed PayPilot AI using Python and Flask for the backend and HTML, CSS, and JavaScript for the frontend.
Google Gemini 3.5 Flash powers the recommendation engine, interpreting user prompts and selecting suitable products from our catalog. We implemented budget validation to ensure the selected bundle does not exceed the user's limit.
PayPal Sandbox Orders API is integrated for checkout, while SQLite stores order details and transaction activity.
Challenges we faced
One of our biggest challenges was ensuring that different user prompts generated different and relevant recommendations. Initially, the application repeatedly returned the same products. We improved the product catalog, refined the Gemini prompt, and added category-aware fallback logic.
We also faced challenges with Gemini model availability and API errors, which we addressed by testing available models and switching to a working model.
Another challenge was PayPal Sandbox payment testing with an Indian billing address due to regional restrictions. We retained the checkout integration and documented this limitation rather than claiming an unverified successful payment.
What we learned
Through this project, we gained practical experience in:
- Integrating generative AI into a real-world application.
- Building API-driven applications using Flask.
- Designing AI-based product recommendation workflows.
- Implementing budget validation and product selection logic.
- Integrating payment APIs and managing order records with SQLite.
- Debugging API availability and handling real-world integration limitations.
What's next
We aim to expand PayPilot AI with a larger product catalog, more advanced product comparisons, multilingual shopping assistance, merchant insights, and improved checkout experiences.
Our vision is simple: One goal, one budget, one intelligent shopping copilot.
