Sell Applesauce turns a photo of something you want to sell into everything you need to list it. Point your phone at an item and get an identification, real market prices from multiple sources, live shipping rates, sales velocity data, and ready-to-post listing copy for eBay and Facebook Marketplace. One photo in, everything else out.
Selling something online means 15 minutes of research per item: searching eBay comps, checking Amazon retail, writing keyword-optimized titles, estimating shipping, crafting descriptions. Multiply that by the 30 items you picked up at a garage sale and you've lost your whole evening.
Sell Applesauce compresses that into seconds per item. One photo replaces the entire research loop.
Upload a photo and the system runs everything in parallel:
After pricing loads, the decision panel calculates your expected profit and ROI. Enter what the item costs at the store and see the math: sell price minus platform fees minus shipping minus cost equals your net. Color-coded signals tell you whether to buy, consider, or skip. When you're standing in Goodwill with 30 minutes to shop, you need a fast answer.
Built specifically for the thrift store floor. A streamlined interface focused on the buy/pass decision. Snap, see the price, enter your cost, get the signal. Designed for speed over detail.
Every analysis gets a shareable link. Send it to a buyer, post it in a reselling group, or save it for later. The share page shows the full analysis: photo, identification, price research with histogram, sales velocity, and shipping. No account needed to view.
Every analysis saves to your inventory automatically. Track items through their lifecycle — listed, sold, or passed — and log costs and sale prices. The dashboard aggregates it all: total invested, revenue, ROI, average margin, and inventory breakdown by status.
Post directly to eBay without leaving the app. Connect your eBay account via OAuth and list items using the AI-generated title, description, category, and suggested price. Facebook Marketplace posting works similarly.
The vision model is surprisingly good at the hard cases that barcode scanners can't touch:
If a human can identify it from a photo, the model usually can too. Multiple photos from different angles improve accuracy for items where brand markings or condition details matter.
The platform is actively expanding. Areas we're building toward:
The stack is deliberately simple. Python FastAPI serving Jinja2 templates. No JavaScript framework — plain HTML with vanilla JS. SQLite for storage. Custom CSS with a dual-theme design system (warm and dark modes).
AI identification uses OpenAI's vision models (GPT-4.1-mini, GPT-4.1, and GPT-5.1) in weighted rotation to balance speed, cost, and accuracy. Images are resized client-side to 1024px before upload, and perceptual hashing (dHash) ensures re-uploads return cached results instead of burning another API call. The whole analysis typically completes in 5-7 seconds.
Sell Applesauce is built by a solo developer at applesauce.chat. I'm building this because I think resellers deserve better tools — and I want your help shaping what comes next. Every message, suggestion, and bug report goes straight to me. Reach out at hello@applesauce.chat or use the contact form.