Build the claims-and-consent packet before drafting

Name the product, audience, channel, offer, required message, prohibited claims, substantiation source, disclosure rule, and final approver. Decide whether the presentation uses a real customer, contracted creator, actor, employee, fictional spokesperson, or synthetic character; those identities are not interchangeable. Capture likeness, voice, script, product, location, music, and usage permissions with territory and expiration. A Higgsfield ChatGPT UGC workflow must never turn UGC-style framing into fake customer provenance. Separate verified product facts from creative hooks and label unresolved claims so no drafting stage fills them with invented experience, rankings, results, scarcity, endorsements, or before-and-after performance.

• Substantiated claim ledger
• Presenter role and consent scope
• Asset and music permissions
• Disclosure and approval owner

Define each named surface as an explicit manual boundary

For every planned step, write the input, expected editorial artifact, file format, human review, and next-system handoff. A language task might organize approved hooks, objections, shot beats, or caption variants; a video task might receive an authorized brief and reference packet. Verify the current product, account, region, terms, upload behavior, controls, and export path independently before using it. Remove confidential or unnecessary data and preserve only permitted context. Do not claim a plugin, API connection, automatic transfer, shared memory, compatible schema, model identity, or SEELE integration unless dated first-party evidence and an observed authorized setup support that exact statement.

Review the assembled ad for truth, provenance, and delivery

Compare the script, visuals, performance, captions, product depiction, price or offer text, audio, and call to action against the approved packet. Check that a synthetic or performed testimonial is not presented as an actual customer account. Inspect labels, packaging, demonstrations, comparative language, health or financial implications, and platform-specific disclosures. Log which system produced each artifact, the visible settings and date, revisions, rejected material, and final approver. Run technical checks for aspect ratio, safe zones, loudness, captions, and destination policy only after claims pass. This page reports a reviewable workflow structure, not campaign execution, approval, reach, conversion, return on spend, or commercial success.