A US consumer product intelligence startup helps brands surface verified metadata for their catalogs, certifications, sustainability claims, and labor policies, information that lives in supplier documents but rarely makes it onto product listings. Our challenge was to help them turn scattered supplier documentation into trustworthy, catalog-scale product claims.
Brands selling cosmetics, personal care, and pet care products are sitting on a metadata problem. The certifications, sustainability claims, and labor practices modern consumers care about exist somewhere, buried in supplier BOMs, certification authority websites, and WhatsApp threads with factories.
Gathering it manually is slow, inconsistent, and impossible to scale across a full product catalog. For a platform built to surface exactly this kind of trust signal, that gap was the core product problem.
Programmatic Research Pipeline Built on AWS Step Functions and Lambda, the pipeline automatically researches each catalog item the moment it's uploaded, pulling public reviews, web research, and supplier documents in parallel, at catalog scale, instead of someone chasing paperwork by hand.
Hybrid Product Matching Each product is matched by barcode first, the most exact method, falling back to a keyword and similarity search when documents lack structured data, so a formatting gap never leaves a product unmatched.
Multi-Model Tag Extraction Two Claude models on Amazon Bedrock split the work, one for complex extraction, one for high-volume tasks at lower cost, and every tag comes with a confidence score, a source citation, and a type: fact, experience, or claim.
Multilingual Processing The pipeline handled German-language reviews and Dutch-language supplier documents without modification, producing accurate English tags from both.
The PoC proved brands no longer have to choose between speed and trust: verified certifications, sustainability claims, and labor practices can be extracted automatically, at catalog scale, with a source citation behind every tag. The client now has a validated foundation to bring this as a core capability of their product intelligence platform.
Tell us more about yourself and what you're got in mind.
Prefer email? hello@apexlab.io
Or even a quick call?
We don’t care about fancy CVs or long interview processes. Provide some basic information, let’s have a quick chat, and then show us what you love doing the most: write some code and create a product.
