Inspiration
Independent makers often answer the same product questions across listings and messages before a buyer is ready to purchase. Important details—such as materials, care instructions, and delivery information—can be difficult to keep consistent.
TradeWeave brings product information, an AI shopping assistant, and checkout into one shopping experience.
What it does
The prototype includes a maker workspace for managing products, options, prices, stock, images, product facts, and shop policies. Buyers can browse the storefront, review product details, ask questions, and add items to a cart. The hosted demo is buyer-facing; the public merchant workspace is disabled.
The AI shopping assistant receives seller-provided product facts and shop policies as context. When information is missing, it is instructed to say so rather than invent an answer. Buyers can also send a question to the maker and view the maker’s reply.
The storefront offers English, Simplified Chinese, Japanese, Spanish, and French interface options. Product content is currently maintained in English and Chinese; a fallback notice appears when content is unavailable in the selected language.
Buyers can try checkout through PayPal Sandbox. The app records order item and price details, and the maker workspace keeps payment and fulfillment statuses separate.
How I built it
TradeWeave uses JavaScript, Node.js, HTML, and CSS. It uses SQLite for local development and Turso (libSQL) for hosted persistence. The PayPal JavaScript SDK and Orders API support the Sandbox checkout flow. The AI assistant uses an OpenAI-compatible chat-completions interface; development testing has used OpenRouter’s free model router.
Challenges and what I learned
One challenge was keeping payment status, inventory reservations, and order information consistent when checkout is interrupted or retried. I added saved checkout snapshots, recovery paths, and checks against duplicate payment processing.
Another challenge was handling missing product information. I tested scenarios involving details that were not supplied, such as shipping prices and guaranteed arrival dates, using simulated provider responses. These checks help ensure the assistant is instructed to avoid inventing details. AI responses can still vary and should be checked against the seller’s facts.
Current scope
TradeWeave is a prototype using a fictional sample shop. PayPal checkout runs in the Sandbox environment, so no real money moves. The hosted demo is buyer-facing, and the public merchant workspace is disabled. Shipping and tax are not calculated; real shipping, refunds, and production payments are not implemented.
Log in or sign up for Devpost to join the conversation.