Mappliance exists because the evidence, not the rule, is the hard part.
A non-EU supplier is asked for documents by an EU buyer who cannot say precisely what would satisfy the regulation. Both sides act in good faith, and the pack still fails. We built the layer that names the artifact, tests it element by element, and tells both sides the answer before a deadline does.
Three constraints, in this order.
Primary legal sources
Every requirement is read from the regulation, the implementing act and the annex — not from guidance about them, and not from a competitor's interpretation.
A human-verified base
Every requirement in the base was read from the legal text by a person and carries the article it comes from. Breaking each one into individually checkable elements is the work in progress. What a regulation names as sufficient is a legal judgment, and it is made before anything is automated.
AI bounded to that base
Automation explains, maps and drafts inside the verified base. It does not decide what the law requires, and it cannot invent a requirement that is not in the source.
Built by someone who used to send these requests.
Mappliance comes out of years spent on the buyer side of exactly this problem — due diligence at amfori, sustainability and legal at Atlas Copco — sending suppliers data requests and watching the wrong document come back. Not because anyone was careless, but because nobody had told them what sufficient looked like.
If your case does not fit what you see here, we would rather hear it than guess.
