View Live Project

AutoTCG

AutoTCG turns a pile of scanned trading cards into priced, published inventory without manual, soul-crushing work.

TCGInventory AutomationSaaSComputer VisionAI VisionScanner Workflow
AutoTCG
May 2026

The Problem

Processing trading card singles at volume is a grind. Every card needs exact identification — right set, right printing, right finish, right language, right condition. Then you price it, convert the currency, create a SKU, and push it to your storefront. Do that a thousand times and mistakes pile up. Manual handling is slow, expensive, inconsistent, and impossible to audit. But going fully automatic is risky when you can't tell if that's a first edition holo or a played uncommon.

Our Solution

Split the work: automate what's obvious, surface what's not. Sentinel lives on the scanner machine, watches for new images, and uploads batches with structured manifests. The cloud pipeline picks them up and runs each card through identification, pricing, and publishing stages in real time. High-confidence cards flow straight to Square. Low-confidence, high-value, or damaged cards get routed to an operator review queue — no guessing, no silent mistakes.

Card shops drown in singles. Scan a stack of 500 cards and you're still hours from having them listed: identifying exact printings, checking finish and language, pulling prices, converting currencies, building SKUs, uploading to Square. It's tedious, error-prone, and doesn't scale.

AutoTCG fixes the pipe. A lightweight Windows tray app (Sentinel) watches your scanner folder, grabs card images, and ships them to the cloud. From there, a real-time pipeline handles the rest — side detection, AI-powered identification, canonical matching, pricing with FX conversion, and Square catalog sync. Cards the system is confident about go straight to inventory. Anything ambiguous, damaged, or high-value lands in a review queue for a human to call.

The operator's job shrinks from "process every card" to "handle the exceptions." Everything else is automated, auditable, and fast.

Tech Stack

RustNext.jsReactTypeScriptAI IntegrationsVector EmbeddingsTauriConvexCloudflare R2Zod

Key Outcomes

Scanner-to-storefront pipeline that actually works at batch scale
Real-time operator dashboard: batches, card details, review queue, issues, settings, device management, admin usage
No drift between desktop and cloud; Reliable upload contract between Sentinel and Convex using shared Zod schemas
Pluggable providers for card identity, pricing, FX, and commerce (swap sources without rewiring)
Full Square integration: OAuth setup, catalog targeting, upload retry, removal controls
Usage and cost tracking for billing and margin visibility
An operator workflow that's 99% accurate without manual import, while flagging the remainders for review

Gallery

AutoTCG screenshot 1
AutoTCG screenshot 2

Want similar results for your business?

Schedule a Consultation