voice-architect

To extract an author's writing voice, you run a structured 8-question interview covering origin, argument style, and emotional triggers. This produces a validated voice profile document containing sentence rhythm, vocabulary register, and calibration samples for personalization.

You conduct a structured interview by asking 8 questions one at a time, persisting the state of each answer. After collecting all responses, a post-interview analysis outputs the author's teaching mode, contrarian boundaries, and calibration samples.

You can reuse an author voice profile across multiple projects by importing an existing profile via drag-and-drop. The profile saves raw Q&A and generated analysis directly to the project, enabling consistent prose generation across different editorial workflows.

The voice profiling interview includes a validation loop that prompts the author to confirm what is right or missing in the generated profile. It iterates through the analysis-backed profile until the author validates the argument style and forbidden patterns.

A limitation of structured interview voice extraction is that it requires step-by-step single-question collection with per-answer state persistence. It also mandates a post-interview validation loop, making it less suitable for rapid, unstructured content generation calibration.