"Studio One"NLHRD_S1_120
url: "resale-j.example.com/i/991"
matched:mouth ringdiffuser fit
Vision intelligence
Counterfeit detection.
As a data feed.
Klaimed builds vision models that match products the way a human expert would — and explain each match, feature by feature. Delivered as structured detection data to brand protection platforms and marketplaces that act on it.
GETapi.klaimed.ai / v1 / detections.streamlive
"Chrono 38"ESWTC_CH_038
url: "marketplace-i.example.com/p/3318"
matched:sub-dial layoutpusher shape
"Trail Mid 92"MXSNK_TM_092
url: "social-h.example.com/p/Da8821"
matched:upper mesh weavelugs
"Salon Edition Mini"AUHRD_SE_091
url: "marketplace-g.example.com/p/8821"
matched:vent geometrycable joint
"Field Watch 38"BRWTC_FW_044
url: "marketplace-f.example.com/listing/4421"
matched:lug profilecrown geometry
What Klaimed delivers
Unique capabilities. One data feed.
Vision intelligence that finds the product, explains the match, and ships into your platform — not a tool your team logs into, a data layer your system runs on.
Image search
Find the product, not the pixels.
Our vision models build a representation of each product from its structural attributes — and retrieve similar products, regardless of angle, colorway, or context. Not perceptual hashing. Not pixel matching.
Explainability
Every match, reasoned
attribute by attribute.
Every match returns the visual features that align and diverge — feature by feature, ready to use as reporting evidence. Explainability not just a similarity score.
Integration
Data feed straight into your stack.
Structured detection records — URL, confidence, attribute reasoning, evidence image, product data — returned through one API, ready to ingest. Built for brand protection platforms, IPMS vendors, and marketplaces.
How the data flows
From reference catalog
to detection records, in four steps.
One API. Your catalog in, structured detections out. Everything in between — visual matching, attribute reasoning, signal scoring — runs on our infrastructure.
01.
Reference catalog.
Send image URIs and product metadata for the items you want to monitor. The catalog is indexed visually and structurally — each reference becomes a queryable target.
WTC_RW_218
SNK_CR_218
HRD_AP_001
BAG_AT_055
02. Vision
Monitoring surface.
Stream in your own scraped listings — image URI plus listing metadata — or hand off collection to us. Either way, the input contract is the same: visual content and associated signals.
03.
We match, reason, and score.
Vision models retrieve visually similar listings. Attribute-level reasoning explains every match. Price, seller, and listing signals score the likelihood of counterfeit or grey-market activity.
04.
Receive structured records.
For each reference, an array of matched listings — URL, similarity score, matched attributes, counterfeit flag, signal breakdown. All in JSON. Ready to ingest.
How klaimed detects
More than vision. One signal, many sources.
Visual, lexical, text-anomaly, and behavioral signals in a single detection workflow. Blend multiple relevance signals through one API call, then let built-in scoring rank what matters.
Visual
Image search
Visual embeddings capture product identity, surfacing matches — even when listings share no text in common.
Lexical
Trademark search
Lexical indexes match registered marks, product names, and known variants directly on listing text.
Text
Anomaly search
Full-text indexing surfacing the misspellings, lookalike marks, and obfuscation patterns counterfeiters rely on.
Behavioral
Behavioral signals
Price, seller, and listing patterns are indexed alongside content — flagging the anomalies that visual and lexical signals alone can't catch.
We sell data,
not software.
Klaimed plugs into brand-protection platforms, IPMS vendors, and marketplace trust & safety stacks — and makes each one measurably more valuable to its customers.
- More detections per customer.Catch the design counterfeits keyword and pixel methods miss — stronger retention and upsell on every account.
- Faster analyst review.Visual evidence up front — margin expansion at the same headcount.
- A new buyer category.Reach design-led brands in fashion, furniture, footwear, electronics — markets keyword tools can't serve.
"The visual similarity is strong on its own. What makes it truly unique for brand protection is being able to see exactly why two images match — feature by feature."
Delivered as data
Deploy in seconds.
Same JSON schema across every delivery method. Your engineers integrate once and we ship the data to whichever endpoint your platform consumes.
detection.schema.json
1{2 "detection_id": "det_9F4A21B8",3 "reference": "heritage-watch-RW-218",4 "url": "marketplace-a.example.com/listing/2218",5 "surface": "marketplace",6 "market": "DE",7 "angle": "front",8 "confidence": 0.96,9 "matched_attributes": [10 { "name": "dial_guilloche", "reasoning": "same concentric pattern" },