product categorization

Your catalog has thousands of products in the wrong category. You just can't find them.

WhiteBox classifies every product through multiple AI models. When they agree, auto-tag. When they disagree on a category, flag it for review. Fix your inventory at scale.

the real problems in product data

The real problems in product data

01
SKU mismatches

Supplier sends "BLU-WIDGET-2024" but your catalog has "Blue Widget (2024 Edition)" -- are these the same product?

One model says yes, another says no. WhiteBox catches the disagreement before you create a duplicate listing.

02
Wrong categories at scale

"Bamboo cutting board with built-in knife sharpener" -- is this Kitchen, Tools, or Kitchen Accessories?

One model says Kitchen, one says Tools, one says Accessories. WhiteBox shows you the split and lets you decide once, then auto-apply to similar products.

03
Duplicate detection

"Apple AirPods Pro 2nd Gen" vs "AirPods Pro (2nd Generation) - Apple" vs "APPLE AIRPODS PRO V2"

Three different listings, same product. WhiteBox identifies when models agree these are duplicates vs when the descriptions are ambiguous enough to need a human check.

how it works

Multi-model consensus in action

whitebox categorization
auto-categorized
whitebox classify "organic bamboo cutting board, 18x12 inch, with juice groove"
options: ["kitchen", "dining", "tools", "outdoor"]
01gpt-4o-minikitchenlogp -0.08
02claude-3.5kitchenlogp -0.05
03llama-3.3kitchenlogp -0.11
04deepseek-v3kitchenlogp -0.09
verdict
kitchen · confidence 97%
SHIP
whitebox categorization
escalated
whitebox classify "rose-scented exfoliating dish brush, bamboo handle, plant-based"
options: ["personal_care", "household_cleaning", "kitchen", "bath"]
01gpt-4o-minihousehold_cleaninglogp -0.62
02claude-3.5kitchenlogp -0.88
03llama-3.3personal_carelogp -0.74
04deepseek-v3household_cleaninglogp -0.81
verdict
no consensus · confidence 38%
ESCALATE
routed to human review · queue: product-ops

Every run, every log-prob, every disagreement -- recorded. Replay any decision from its ID.

use cases

Anywhere products need categories, you need consensus

01
Marketplace listings

Auto-categorize seller-uploaded products across your taxonomy. Catch miscategorized items before they go live.

02
Catalog migration

Migrating from one PIM to another? WhiteBox maps products to your new taxonomy with confidence scoring.

03
Duplicate detection

Find duplicate listings hiding under different names, SKUs, or descriptions across your catalog.

04
SKU matching

Match supplier SKUs to your internal catalog. Flag uncertain matches before they create inventory errors.

05
Product tagging

Auto-tag products with attributes (color, size, material, occasion) from unstructured descriptions.

06
Category cleanup

Scan existing catalogs for miscategorized products. Fix thousands of wrong categories in bulk.

bulk processing

Categorize 100 products at once

POST /api/v1/decide/bulk
batch API
{
  "items": [
    {"input": "organic bamboo cutting board 18x12"},
    {"input": "stainless steel chef knife 8 inch"},
    {"input": "rose scented dish brush bamboo"},
    ...up to 100 items
  ],
  "options": ["kitchen", "dining", "tools", "bath", "personal_care"],
  "prompt": "Categorize this product into one department",
  "webhook_url": "https://yourstore.com/webhooks/categorized"
}

Submit 100 products. Get results via webhook when done. $1 per batch.

by the numbers

What multi-model categorization looks like at scale

97%
auto-categorized
products where models agree and ship automatically
3%
escalated
edge cases flagged for human review
$0.01
per product
cost per categorization
100
per batch
bulk API limit
comparison

Why multi-model beats single-model for product data

Feature Single model WhiteBox
Edge cases Silently miscategorized Flagged for review
Ambiguous products Random category assigned Disagreement surfaced
Duplicates Missed Caught by consensus
Audit trail No record Every decision logged
Bulk processing Build your own 100/batch, webhook callback
Human review Not built in Queue with full model breakdown
playground

Try it. Paste a product description, see the category.

kitchen dining tools bath personal_care household outdoor
whitebox sandbox · simulated client-side
[--:--:--] waiting · press categorize to dispatch
models
4
median latency
0.8s
cost / product
$0.01
audit retention
forever
pricing

$0.01 per categorization

20 free to start. No credit card.

That's 1,000 products categorized for $10.

free tier
20 decisions
per product
$0.01
subscriptions
none
get a key
get started

Stop shipping products into the wrong category.

20 free categorizations. Then $0.01 each. The audit trail starts the moment you install.

get a key API docs