RedCloud — Commerce Computer Vision | Punch
← ALL WORK / AI PROJECTS / REDCLOUD
RedCloud NASDAQ: RCT

Wave the phone. The shelf uploads itself.

How Punch gave RedCloud's field agents a camera that reads a store — recognizing every product on the shelf so a merchant gets onboarded in minutes, not an afternoon of typing SKUs.

COMPUTER VISION PRODUCT RECOGNITION ON-DEVICE INFERENCE MOBILE
RedCloudINVENTORY CAPTURE IN THE FIELD POINT · RECOGNIZE · DONE
PRODUCT WALKTHROUGH · ONE PASS DOWN THE AISLE CAPTURES THE CATALOG
THE WAND · WHAT HAPPENS AS YOU WALK
POINT CAMERA SWEEPS THE SHELF
→
DETECT ON-DEVICE PRODUCTS BOXED, LIVE, OFFLINE-OK
→
RECOGNIZE SKU ID + CONFIDENCE
→
MATCH CATALOG PRICE · PACK · BRAND
→
IN THE CART ✓ AGENT CONFIRMS QUANTITIES
FIG. 01 — ONE SWEEP DOWN THE AISLE, START TO CATALOG
THE EVIDENCE · A SHELF, RECOGNIZED
SKU·0.97 SKU·0.93 SKU·0.95 SKU·0.91 CHECK·0.61 SAMPLE — SPEC RC-01 · LIVE RECOGNITION OVERLAY
FIG. 02 — RECOGNIZED SKUS LOCK GREEN; LOW-CONFIDENCE ITEMS FLAG AMBER FOR A ONE-TAP CONFIRM (SAMPLE)
THE OLD WAY VS THE WAND
MANUAL SKU ENTRY~3 MIN / 10 ITEMS
WAVE TO CAPTURE~15 SEC / 10 ITEMS
SAMPLE — DIRECTIONAL, CONFIRM WITH FIELD DATA
WHY ON-DEVICE

Shops have spotty signal. Detection runs on the phone so the wand works whether or not there's a bar of service — syncing the catalog when it reconnects.

WHAT WAS SLOWING ONBOARDING

Every new merchant meant typing a thousand SKUs by hand.

RedCloud sends field agents into shops to bring merchants onto the platform. The catch: onboarding a store meant manually keying in every product on every shelf — slow, error-prone, and the single biggest brake on how many merchants one agent could sign in a day.

Punch replaced the keyboard with the camera. The agent walks the aisles, waves the phone like a wand, and the app recognizes products on the shelf and pulls them straight into the catalog.

“Commerce infrastructure that scaled all the way to the bell.”
12×FASTER SHELF CAPTURE VS TYPING (SAMPLE)
0SKUS TYPED BY HAND
offlineRECOGNITION RUNS ON THE PHONE
WHAT WE BUILT
PLATFORM✓AWS-scaled commerce platform✓Mobile point-of-sale✓Payments & wallet integration
ENGINEERING✓Multi-market architecture✓Offline-first sync✓Public API platform
SCALE✓IPO-grade infrastructure✓Emerging-markets rollout✓High-availability operations
SERVICES PROVEN Computer vision systems On-device / edge inference Mobile app engineering Catalog matching & data ops CLIENT RedCloud — merchant onboarding (went public) · redcloudtechnology.com ↗ CASE STUDYredcloud case study coverCase study PDF ↓IN THE NEWS GLOBENEWSWIRE · MAY 2026 FY25 20-F filed — 2026 revenue guidance of $120M reaffirmed. ↗ STOCKTITAN · APR 2026 Specialist AI agents trained on $6.9B of FMCG trade data. ↗ GLOBENEWSWIRE · MAY 2026 Saudi operations launch under $30M JV with Kayanat. ↗
Punch / Lab NotesISSUE 14
LATEST BUILD
THIS ISSUEInside the models we shipped this month.What we trained, what surprised us, and the eval that let it ship.
PUNCH · AI RESEARCH NEWSLETTERThe AI we’re building next — in your inbox.Research notes from the lab, the models we’re training, and the new projects we’re shipping. A few times a month — no spam, no fluff.★FREE FOR SUBSCRIBERS · THE AI EVAL PLAYBOOK — SEE INSIDE ›
RECENT ISSUESPer-subscriber avatarsYOLO eval at scaleOffline product recognition· JOINED BY 2,400+ ENGINEERS & FOUNDERS
Putting AI in the hands of a field team? Talk to the engineers who built this →
←PREVIOUS — ZWILTZWILTThe interview that interviews back. NEXT — RECOVERYTREKRECOVERYTREKA witness in every pocket.→