Loading page...
The four end-to-end flows behind SuperCatch — how listings go live, how cards get identified, how orders ship, and how channels stay in sync.
Jump directly to the sections that matter for this page.
How a card travels from an import or scan to a live, sold listing.
Cards enter through CSV or marketplace import, a photo scan, or manual creation.
The recognition pipeline matches each card to a catalog identity with a confidence score.
The seller reviews flagged matches and fixes missing data — approve, edit, or dismiss — before anything becomes a listing.
Approved items become DRAFT listings with catalog links, images, and pricing; the seller reviews and submits.
Submitted listings sit PENDING while an admin verifies card details and images, then approves or rejects.
Approval moves images to permanent storage, generates SEO, and activates the listing; it syndicates to connected sales channels.
A buyer checks out; quantity decrements everywhere and the order enters fulfillment.
What happens between uploading card photos and confirming the identified card.
Front and back photos are compressed, validated (JPEG/PNG/WebP, 500×500 minimum), and archived as originals before any processing.
Cheap matches run first: an exact image-hash cache hit answers without any AI spend, and graded cert numbers verify against grading company databases.
A validated cache or cert hit at the 0.85 confidence bar answers immediately — the vision model only runs when the cheap paths miss.
The cached or cert-verified identity is confirmed with no vision model call.
One scan credit is reserved, then the vision model (Webanto local LLM primary, GPT-4o-mini fallback) reads player, year, brand, set, and card number off the images.
Catalog candidates are matched in tiers — strict identity, fuzzy create-gate, same-set siblings, then DINOv3 image rescue — and ranked with a confidence score.
The model's raw confidence decides the route: 0.95 and above can auto-approve; below 0.7 always needs a human.
High-confidence matches are confirmed; at 0.95 and above manual review is skipped entirely.
Low-confidence results land in the review queue flagged for the seller instead of auto-confirming.
The seller confirms or corrects the suggested match before anything publishes.
Confirmed cards continue into the listing lifecycle as draft or review listings headed for approval.
The path an order takes from checkout to seller payout.
The buyer completes the three-step checkout (shipping, review, payment); PayPal captures payment and the order is created for the seller.
Proceeds sit in the seller's pending balance, and a postage reserve from the buyer's shipping payment is set aside for the label.
The seller buys a USPS label (Shippo or ShipStation optional), packs the cards, and the order is marked shipped with tracking.
Carrier scans update the shared order timeline; a confirmed delivery automatically promotes the order to delivered.
Delivery confirmation opens the window to report an order problem.
Support reviews the case; the disputed amount stays held and pending payouts pause until the dispute resolves.
After the window clears, proceeds minus fees and any outstanding debt are paid out to the seller via PayPal.
How a connected platform stays aligned with your SuperCatch inventory.
OAuth or stored credentials connect the channel from the seller dashboard, and the product cache pre-warms.
In the import wizard the seller picks products and maps platform fields to SuperCatch fields before anything imports.
A background job pulls the platform catalog; images are processed, AI-matched, and unmatched data is flagged for review.
Scheduled runs pull only records changed since the last successful sync, with quick retries on failure.
Real-time order and product events flow between SuperCatch and the platform where the platform supports them.
Explore the adjacent workflows that usually sit next to this topic.
