PricePilot predicts what your handmade or resale item will actually sell for, from your own sales data, with an open-weight model on your laptop, then writes the listing. No account, no API bill, and your numbers never leave your machine.
Describe the item the way you'd list it. An open tabular model predicts the sell price with an honest range, shows which levers raise it, and drafts the listing.
Category, materials, size, make-time, condition, photos, season. The same facts you'd put in a listing anyway.
An open-weight tabular foundation model predicts the sell price from past sold listings in one CPU pass, with a 10–90% range and a measured error, not a guess.
A local open-weight Gemma 3 model turns the item and that exact price into a title, description, and tags, grounded on the number, never inventing one.
Your sales data stays yours. Your margins and what sells are a seller's edge. PricePilot runs on your laptop, so nothing is uploaded to a competitor's cloud or logged to a server you don't own.
$0 per listing. Open weights mean no per-call API fee. A seller pricing 300 items a month pays nothing. A closed API turns pricing into a recurring bill.
Tune it to your niche. Swap in your own exported sales CSV and the model learns your corner of the market (vintage synths, hand-poured candles) where a general API is weak.
No vendor lock-in. Point the listing writer at any open-weight model Ollama runs, or retrain the predictor. You own the stack, not someone else's pricing page.