How RepTell reasons

What a DealCheck uses

RepTell starts from the structured facts you enter: seller, claimed watch, claimed factory, price, and terms. It also uses the private listing photos, messages, and pasted text you attach to that case.

The server then selects a bounded set of curated seller, factory, and watch records that are allowed for RepTell. When policy allows, a finding may also rest on reviewed external evidence already attached to those records. That is selected evidence, not a live scrape or a dump of the research archive.

How a report is written

The published DealCheck is a public-safe report. It shows a disposition, findings, citations, limits, and questions to ask next. Citations name the public record or labeled inference that supports a finding. They do not expose prompts, internal taxonomies, source bundles, or your raw files.

Uncertainty stays visible. Unknown facts and missing evidence are part of the report. They are not hidden behind a single score.

Community views stay evidence

Forum consensus, seller reputation talk, and repeated buyer reports can inform a finding. They remain evidence. Informed people can disagree. RepTell does not silently convert a popular view into a fact about factory origin, seller intent, or this transaction.

Why a report may stop short

A report can return insufficient evidence or ask for more when usable watch photos are missing, the images cannot be assessed, seller or factory identity is too thin, or the selected records do not cover the claimed watch. That is a limit of the submitted evidence and the available curated knowledge.