Human review standards

A reviewer must have an active Tabeerly reviewer profile and an explicit tradition or school scope. The queue shows only work within that scope. Tabeerly does not currently publish named reviewer profiles, and it does not use anonymous reviewer counts or ratings as marketing proof.

Pending work reveals only queue metadata. A reviewer must claim a job before the dream, context, AI draft, and source trail can be decrypted. One reviewer can hold the assignment, access is logged without dream text, and another reviewer cannot open or submit that review unless it is reassigned.

The assigned reviewer sees the complete draft and its citations. They record one of three decisions: approve, amend, or override. Amendments and overrides create a separate reviewed version while the original AI draft remains preserved; every decision requires a written rationale. The decision, timestamp, reviewer scope, and note remain visible in the completed interpretation record.

A completed review is not a fatwa, prediction, guarantee, diagnosis, or claim to know divine intent or the unseen. It does not mean every scholar in a tradition would agree. The current workflow uses one assigned reviewer per user interpretation; the separate source-claim verification process requires its own provenance and review record.

Priority purchase changes queue order only. Tabeerly does not promise a completion time until measured staffing and service levels support one. Formal public rules for conflicts of interest, reviewer compensation, credential verification, reassignment, and appeals are still governance requirements; they must be published before the program presents named reviewers as institutional authorities.

A user can report a source or review concern through the contact channel. A disputed source claim must leave retrieval while it is assessed. Material corrections should preserve the original decision trail, document the reason and date, and identify affected outputs. The public correction ledger is not yet live and is a prerequisite for scaling indexed authority content.