An AI report should inform the author before it edits the author.
Analysis should not silently become revision
B-Maker can analyze a manuscript across dimensions such as audience fit, engagement and pacing, structure and logic, genre expectations, argument and evidence, publishing position, story consistency and AI-authorship signals. Those results are reports, not automatic edits.
Audience analysis answers a different question
A manuscript may be internally strong and still be unclear about who it is for. Audience analysis can produce hypotheses about primary and secondary readers, what may resonate, risks, expectations and positioning. B-Maker labels these as AI hypotheses rather than a market forecast.
Reports need provenance
Saved reports keep the analyzed scope, source-word count, model settings, reasoning effort, token usage and creation time. A report can become outdated when the manuscript changes, which is important: advice based on an old snapshot should not masquerade as current truth.
Useful findings can return to the project
You can create a working note from an AI report, preserving the result as project context instead of copying it into a random external document. When you actually want to rewrite text, the separate AI revision flow can create a new version for review.
AI-authorship signals need a warning label
Stylistic signals are not proof of authorship. B-Maker therefore treats AI-authorship estimation as a heuristic with limitations and conclusion, not as a detector that can establish who wrote a text.
Let AI make claims before it makes changes.
Reports stay inspectable; revisions stay reversible; the canonical manuscript remains under the author's control.