Brand-Voice Drafting
Draft release notes, posts, and blogs against your brand-voice file, then reshape each for its channel.
What It Does
The agent reads your brand-voice file before it writes, drafts the artifact, checks whether you have said it before, and adapts the approved piece into each channel's native shape. A person approves before anything is scheduled.Without it: One-off chats do not compound. Every prompt starts from zero, and the draft sounds like everyone else's.
- Names the archive piece a passage repeats, rather than scoring similarity
- Claims with no source in the release note are left as marked gaps
- No brand-voice file, no draft, so it stops rather than writing something generic
- Drafts land unpublished in the CMS with the voice rules each choice leans on
Illustrative example
A blog draft on the audit-log rewrite comes back with three paragraphs flagged as near-repeats of the March post 'Every action, logged', and a bracketed gap where the draft asserted a 90-day retention window no release note states. The writer fills the gap, cuts a paragraph, and approves. Only then does the agent cut the LinkedIn and changelog versions, each rewritten to its own shape and left sitting unpublished in the CMS.
Brand-voice file
Voice rules, required
Content archive
What you said before
CMS
Drafts sit here unpublished
Brand-voice drafting
SuggestStarts when
A release ships, or the calendar calls for a piece
How far it goes
it proposes, and a person approves before anything commits
Never
PUBLISH: enforced below the model, not asked of it
Drafts
Left unpublished in CMS
Repetition check
Flagged against the archive
Channel versions
Cut after approval
More Marketing Agents
Content calendar planning
Set the quarter's themes and objectives, then put next week's posts on the calendar.
Launch orchestration
Build the launch sequence, work the readiness checklist, and hold the go until a named human approves.
Performance reporting
Read the analytics you already have, write the weekly digest, and name what to stop.