Inspiration
Small furniture businesses need to present their products clearly and respond to customer questions, but creating product descriptions and managing inquiries can take time. I wanted to build a tool that brings these tasks together while helping sellers avoid misleading product claims.
What it does
FurniSmart AI is an AI-assisted furniture marketplace for sellers and buyers.
Sellers can create and manage furniture listings.
The AI assistant helps draft product descriptions using seller-provided information.
Buyers can browse furniture listings, save favorites, and send inquiries.
Sellers can review customer inquiries and manage conversations.
Administrative features support seller and listing approval.
A key design principle is that the AI should not invent product features, prices, stock availability, delivery terms, warranties, or other business promises. Sellers remain responsible for reviewing AI-generated content before using it.
How we built it
I built the application as a local web app using Python, Flask, and SQLite. The project includes separate buyer, seller, and administrative workflows. I also integrated a local AI model to help generate furniture descriptions, with checks intended to prevent unsupported claims.
Challenges we ran into
One of the biggest challenges has been making AI-generated descriptions useful while preventing the model from adding details the seller never provided. The safety checks have flagged unsupported claims during testing, so improving the reliability of this feature remains an important part of the work.
Building this project has taught me about connecting web application features, testing complete user journeys, handling AI output carefully, and designing tools that keep people in control of important business decisions.
Accomplishments that we're proud of
Built a working foundation for a furniture marketplace.
Implemented seller workflows for creating and managing furniture listings.
Developed buyer workflows for browsing products, saving favorites, and sending inquiries.
Added administrative workflows for managing sellers and approving listings.
Integrated a local AI-assisted product-description feature.
Tested key buyer journeys, including opening a listing, saving a favorite, submitting an inquiry, and viewing the conversation.
Identified an important AI reliability challenge and continued testing safeguards against unsupported claims.
What we learned
Building FurniSmart AI taught us how to develop a web application with Python and Flask, organize data with SQLite, and connect features across buyer, seller, and administrator workflows.
We learned that integrating AI into a real business tool involves more than generating text. AI outputs must be checked carefully so they do not invent product details or make unsupported promises about price, delivery, warranties, or availability.
Testing the application also showed us the importance of checking complete user journeys, protecting existing data, and keeping sellers in control of the content they publish.
What's next for FurniSmart AI
My next priorities are to improve the AI description workflow, test the main buyer and seller journeys, and make the project easier to demonstrate and use.
Log in or sign up for Devpost to join the conversation.