This library describes a creative workflow using source references, AI-assisted production, quality checks, and human business approval. It labels real client work separately from illustrative examples and does not infer campaign results from a polished asset.
The production and review roles.
| Stage | Responsibility |
|---|---|
| Source preparation | The business supplies or confirms current product facts, approved references, audience context, and constraints. |
| Creative preparation | The workflow develops a message, script, visuals, and edit appropriate to the brief. |
| Quality review | Creative agents and the assisted production process check issues such as overlaps, empty scenes, transitions, captions, and audio. |
| Business approval | A human reviewer confirms product accuracy, wording, offer, and intended use. |
| Distribution and learning | Publishing and spending decisions follow their own permissions; performance conclusions require relevant data. |
The presence of quality checks does not guarantee that no defect can remain. The final exported asset is the object of review. Product-specific promises require a current source and an accountable business reviewer.
What each evidence label means.
- Illustrative creative: a finished example showing a production approach, without a customer outcome claim.
- Real client work: an identified customer delivery, with the source of its reported outcome stated.
- Final approved: the accepted version after any required revisions.
- First-pass accepted: would require records showing the first submission was accepted without revision.
- Measured result: would require a defined metric, dates, sample, source, and appropriate interpretation.
How we count the NAIARA delivery.
Aitoma’s production team reports a first-month delivery of 19 videos, 11 carousels, 3 images, and 30 stories: 63 content pieces. A carousel counts as one content piece, however many slides it has. All 63 were finally approved as confirmed by Aitoma’s production team.
This establishes delivery and acceptance as reported. It does not establish a first-pass approval percentage, independent campaign performance, or the amount of manual work saved. Those questions need different records.
How timing and improvement claims should be supported.
Define the start and end points, distinguish hands-on effort from elapsed time, include revisions consistently, and state the sample size. A generation timer alone cannot establish website-to-human-approval time.
For a campaign comparison, preserve the metric definition, platform, dates, currency, spend, outcome counts, and comparison source. Unverified benchmarks and isolated results are not turned into universal improvement claims in this library.
How corrections are handled.
Keep the original source, identify the affected wording or asset, and update the dependent material. A changed product screen can require more than a caption edit. Keep approved versions traceable so an outdated export is not reused accidentally.
Our AI-search citation field note applies the same distinction to search visibility: a brand mention, a linked destination and current product facts are separate observations.
If you notice a factual issue in a resource, contact hello@aitoma.ai with the page and the correction source.