Souk Lens hero
PayPal AI Hackathon · IndieVision · Listing · Studio · Voice + PayPal Checkout

From one market-stall photo to a ready-to-sell store in under a minute

Snap a stall → segmentation isolates each item → the listing model prices and writes the full listing → the image model cleans the shot. Pay with real PayPal buttons. Voice, text, links and store import work too — every result is a saved, sellable product.

0 saved productsEG · EGPDemo / Live ready
Impact in numbersRun the demo to measure live
Souk Lens time (measured)
—
No run yet
Manual time (market estimate)
15–20 min
Shoot + write + price by hand — estimate, not measured
Speedup
—
Based on time actually measured on this device
Full listings saved
0
Images + prices + variants + specs + SEO

Three AI models in one pipeline: segmentation isolates, the listing model prices & writes, the image model cleans the shot — every run is labeled Live or Demo honestly.

market stall
Illustrative demo story

An informal-market seller — from messy stall to a selling store

The seller journey in a minute
An illustrative scenario built on real in-app capabilities — not a paid testimonial. You can run every step yourself from the tabs.

Snaps the stall with his phone

0:00

Segmentation isolates every item

0:10

Full listing, local price

0:25

Sells: fixed, auction, live or campaign

0:60

“I can't write ads or shoot well. I snapped the stall, it did the rest — and my first piece sold at auction.” — the seller's voice in this demo scenario; the point is any seller can take the same journey without reading a single technical word.

Seller — create & sell
Buyer — compare & buy
Analyze a market photo
Upload or paste a link — segmentation isolates, the listing model writes the full listing, the image model cleans the shot.
How does it work? Step by step▾
  1. 1Paste an image URL or upload from your device — the preview shows what will be analyzed.
  2. 2Name what you're looking for (optional) for sharper detection.
  3. 3Press Analyze: segmentation isolates each item, the listing model writes a full listing with price and variants per item, the image model cleans the hero shot.
  4. 4Tap any box on the segmentation map to open its listing, and drag the before/after slider to compare market vs studio.
  5. 5Press Save all to keep the products in My products with full details and images.

Direct image link (https) — or upload from your device. Photo tip: get close to the items and cut empty surroundings (bare table/floor) — detection is far sharper on tight photos.

“All items” generates a listing for everything segmentation isolates. “Specific” focuses only on what you name.

Real multi-item stall photos — pick “All items” for batch generation, or name specific ones.

Runs segmentation, then the listing model, then the image model in order on the same photo.

input
AI models in this tabsegmentationSegment items out of the photolistingRecognize + local price + full listingstudio imageClean studio shot

Result

Demo previewsegmentation + listing + studio image + voice (ids via env)

Every product is auto-enriched: variant photos + 360° views + GLB 3D support — expand the full section under any card or open the buyer page to see it as customers do.

How does it work? Step by step▾
  1. 1The proof bar above shows what each model did, how long it took, and whether it was live or demo.
  2. 2Segmentation map: every colored box is a real item in the photo — tap it to open its full listing.
  3. 3The before/after slider compares the original market shot with the generated studio shot.
  4. 4Product cards: save to My products, add to cart, or open full details — and pay with real PayPal buttons.
  5. 5The raw JSON button reveals the models' original response for technical review.

Nothing here yet

Snap it, say it, paste a link, or import a store — every result becomes a full listing saved with all details and images.