Analyze a COA — TitrateLab

Analyzer paused Public analysis is temporarily off while we finish counsel review of the reputational-claims disclaimer. Check back soon.

Anonymous · No account

Drop a file. That's it.

How it works

1
Pre-filter classifierTesseract pass + Haiku 4.5 vision check that the image is actually a peptide COA. Non-COAs get rejected before we pay for full OCR.
2
Primary OCRGemini 2.5 Flash extracts peptide, dose, purity %, method, test date, batch ID, manufacturer.
3
Dual-OCR cross-checkA second OCR engine reads the same image. Disagreement on the purity number flags the document for your review.
4
Lab signature verifyFor Janoshik tests, we fetch the issuing lab's archive and compare the fields. Match = green check; mismatch = red flag.
5
Corpus comparisonIf we have ≥20 published tests for this vendor, we show how this batch compares to the vendor's 12-month mean.
6
Verdict assembledDiscrete badges, no aggregated "confidence" label. You read the checks and decide.

Before you upload

Redact PII. Names, emails, addresses, order numbers — anything that identifies you. We OCR the whole image; our automated PII pass is a backstop, not a substitute.
It's an analysis, not a chemistry test. We summarize what the document says + how it compares. We do not independently verify the underlying assay.
If we don't have it, we save it. New COAs feed the corpus that powers vendor pages and trust grades. You're contributing to the same dataset that helps everyone.
Vendor disagreement? Vendors who think a number is wrong can write to [email protected] for review.