It interviews you. It never writes for you.

Your expertise, in your actual words.

Most AI writing tools produce a competent version of something anyone could have written. Still My Voice does the opposite. It asks you questions, follows up on what you actually said, finds the evidence, tells you where the argument is weak — and then stops, because the sentences are yours to write.

The rule

It does everything except put words in your mouth.

It does: interview you and follow up on what you actually said; outline; find and organise your evidence and name the claims that have none; surface your own words back to you with attribution; recommend what to lead with and what to cut; research, flagging every outside claim for verification; and push back when something doesn't hold up.

It does not: draft your sentences, write a paragraph in your voice, invent anything, or attribute a quote your source doesn't support.

Dictation cleanup is the single exception, and it's diff-gated: punctuation, filler and transcription mangling only. Blunt stays blunt. You approve it word by word, and the acceptance is logged.

A playbook is data, not code.

Audience, interview strategy, evidence standard, output formats and sensitivity guardrails. Adding a playbook is an afternoon of writing questions, not a release.