Flip AI

Inspiration

The second-hand clothing market is exploding, with platforms like Vinted hosting over 80 million items. However, for resellers trying to make a profit, the process is a "manual nightmare".

We realized that most people find deals by "doom scrolling" endlessly swiping through random listings, hoping to get lucky. It is inefficient, slow, and totally based on being in the right place at the right time. We wanted to change the game from luck to math. We built FlipAI to stop the scrolling and start the profiting. We treat sneakers and vintage hoodies like stocks, creating a tool that looks less like a shopping app and more like a Bloomberg Terminal for flippers.

What it does

FlipAI is an "Automated Arbitrage Discovery" tool. Instead of showing you clothes you might like to wear, it shows you clothes that will make you money. Here is the step-by-step process of what the app does for the user:

1. Scans the Market: It continuously monitors Vinted for new listings (e.g., a "Vintage Nike Hoodie" listed for €15).
2. Price Checking: It instantly cross-references that item against "Sold" history on other marketplaces like eBay or Depop to see what people are actually willing to pay (e.g., the same hoodie selling for £95).
3. Calculates Profit: It does not just show the price difference; it calculates the "Profit Margin" using a probability engine. It subtracts the cost from the market value to show you the clear profit (e.g., €80 Profit).
4. Instant Alerts: Since good deals vanish in seconds, the app sends a push notification the moment a profitable flip is identified.

How we built it

We used a "Vibecoding" tech stack to automate the entire reselling workflow. The core logic is built in Next.js, serving as the glue that connects our different AI and data services.

Here is the detailed breakdown of the stack ():

The Engine (Oxylabs): We used Oxylabs to handle the heavy lifting of data collection. It scrapes Vinted for active listings and scrapes eBay for "Sold" history. This provides the raw data we need to find price gaps.

The Brain (Supercorp AI): Raw data can be messy. We utilized Supercorp AI to compare images and descriptions. This ensures that the €15 shoe on Vinted is actually the exact same model and condition as the €100 shoe on eBay, filtering out fakes and wrong matches.

The Finance (StockInvest): To predict future value, we integrated StockInvest. This pulls macro-economic data (like currency strength and inflation) to predict which luxury brands or items will hold their value over time.

The App (Natively): Speed is everything. We used Natively to convert our web application into a high-performance mobile app. This allows for real-time push notifications so users never miss a deal.

Challenges we ran into

Data Matching: It is very difficult to tell if two items are identical just by text. A "Blue Nike Jacket" could be a hundred different things. We had to rely heavily on AI image matching to ensure we weren't telling users to buy the wrong item.

Latency: In the flipping world, seconds matter. If our scraper was too slow, the item would be sold before the user got the alert. Optimizing the Python backend to process the Oxylabs data instantly was a major hurdle.

Accomplishments that we're proud of

The Probability Engine: We successfully replaced "gut feeling" with a mathematical formula:

Profit = (Market Value - Cost Basis) * AI Confidence.

Value Proposition: We built a product with the easiest sales pitch in the world: "Pay us money to make more money." One flip a month pays for the app.

What we learned

Reselling is Data Science: We learned that the second-hand market behaves exactly like the stock market, just with physical assets.

The "Manual" Gap: We discovered that while there are tools for inventory management (like Vendoo), there was a massive gap in "Active Discovery." No one else was automating the finding part of the process.

What's next for Flip AI

Phase 1 : Transition the current limited prototype into a fully functional system.

Phase 2 (Scale): We will expand beyond Vinted and eBay to support multi-marketplace analytics.

Phase 3 (Auto-Buy): We plan to introduce "One-Click Buy" and logistics integration, so the user doesn't even have to leave the app to secure the item.

Phase 4 (B2B): We will eventually open an API to sell our trending data to major retailers, giving them insight into what is hot in the secondary market.

Built With

• claude-code
• gemini-cli
• google-antigravity
• google-gemini
• google-stitch
• natively
• node.js
• oxylaps-api-scraper
• python
• superinterface-ai-chatbot
• typescript