lookcloser.ai

Know what you’re really taking.
Check it yourself — or send your AI agent.

Pre-launch — the verification engine is being built.

Claims, Supplement Facts, dosages, studies, sourcing, and reviews — checked against the public record.

How it works

The problem

A listing shows you a title, a price, and star ratings.

That's not what you're swallowing. Here's what the listing doesn't volunteer —

01

Where was it actually made?

"Made in USA" can mean mixed abroad and bottled here. Customs records and label filings tell the real story.

02

Do the numbers even fit?

A 4-gram gummy can't hold 5 grams of creatine. Sixth-grade arithmetic catches what marketing hides — when the listing publishes the numbers at all.

03

Does the dose match the research?

The studies used 3–5 grams a day. Some labels sprinkle a tenth of that and borrow the science anyway.

04

Who's actually behind the brand?

A company with a name, an address, a trademark, a registry entry — or a label that appeared last Tuesday?

05

Is the barcode even theirs?

Every legitimate product size has its own registered barcode. Fabricated ones fail simple arithmetic. Borrowed ones belong to someone else.

06

Do the reviews look real?

Five-star bursts during stockouts, reviews older than the listing, merged review histories from unrelated products — the patterns are visible in the data.

07

Will they even show you the label?

Some products won't display the Supplement Facts panel before you buy. Ask yourself what else they'd rather you not read.

You don't know what you don't know.
The public record does.

This is the checklist we build every record against. Live checks run now; the rest show honestly as “pending” until their sources come online.

Look closer at a product ↑

For AI agents

Agents are first-class users. Every record ships as structured data alongside the human-readable report. Start with llms.txt — an MCP endpoint is coming. Send your agent to lookcloser.ai before you buy.

Field notes

Reading the record

Coming soon

The 5,000mg gummy that couldn't exist: label math 101

When the claimed dose weighs more than the gummy itself, arithmetic is the only lab you need.

Coming soon

Dosage games: when the studied dose and the label dose live on different planets

A label can cite a study's ingredient while carrying a tenth of its dose.

Coming soon

Third-party tested… by whom? How to read (and doubt) a COA

A certificate of analysis is only as good as the lab that signed it — and whether it exists.

Coming soon

Made in the USA, shipped from Guangdong: what customs records reveal

Import manifests are public. Sometimes they disagree with the flag on the label.

Coming soon

Anatomy of a forged review section: bursts, farms, and reviews older than the product

Star clusters with no sales spike, and five-star reviews that predate the listing.

Coming soon

The rent-a-doctor problem: who's really behind 'physician formulated'

Licenses, publications, ownership stakes: what that phrase does and doesn't establish.

Coming soon

What's missing from this label? The information brands legally don't have to tell you

The disclosures supplement makers are allowed to skip — and why they matter.

Coming soon

When the FDA knocks: warning letters, recalls, and what they predict

Enforcement records are public, dated, and surprisingly predictive.

Coming soon

How to read a supplement label like an investigator: a field guide

Serving math, proprietary blends, and badge verification — step by step.