Agenteous·
Marketing Agents

Marketing Brand Guard

Marketing Brand Guard checks a draft against a brand's rules and gives it a score out of 100. It reads the draft for voice, visual consistency, messaging, and governance, then returns one of four decisions: Auto-Approve, Review, Revise, or Reject. When a draft falls short, it also returns redlines that name the passage, say why it fails, and propose a replacement.

Brand Guard is the checker, not the author or the publisher. It never writes or sends content. The agents that create a draft call it, and you decide what to do with the result.

What It Does

Brand Guard works in two passes. First it runs fixed rules that do not depend on judgment: the brand's prohibited phrases, its required disclaimers, external links and named sources the draft cannot back up, performance claims with no evidence behind them, and a short list of stock phrases that read as machine-written. A serious hit on any of these rejects the draft outright, however well it scores otherwise. Then it reads the draft against the brand's voice and positioning and scores it on four areas: Voice out of 30, Visual out of 25, Messaging out of 25, and Governance out of 20. A text-only draft gets full Visual marks unless an attached image breaks the brand's standards.

The total decides the outcome. By default, 85 or higher is Auto-Approve, 70 to 84 is Review, 50 to 69 is Revise, and below 50 is Reject. A serious flag can pull a high score down to Review, and a critical flag forces Reject. Each brand can set its own cutoffs and its own rules, so one client's rules never affect another's.

Brand Guard also suggests content topics for a brand. You ask for suggestions in Slack, and each topic arrives with Greenlight and Dismiss buttons.

When It Runs

Brand checking has three modes, set in the Operator Console: Enforce, Warn, and Off. Warn is the shipped default.

  • Enforce: the check runs when a publishing agent asks for it, and a draft that fails is held as blocked. What you can do next depends on the agent; see that agent's Troubleshooting page.
  • Warn: the check runs and the flags are shown, but no agent holds the draft back. A draft the check rejects goes through and the rejection is recorded as warned. The one exception is a check that could not run at all, which still holds the draft. Meeting Assistant keeps such a follow-up draft and counts it as needing edits.
  • Off: no check runs when you approve or publish, and approved emails send and landing pages publish as normal. On-demand audits and topic suggestions still work.

When the mode is Warn or Enforce, the checks run when you approve or publish. The agents that call Brand Guard this way include Marketing Content, Marketing Social, Marketing Email, Marketing Web, Marketing Landing Pages, Marketing PPC for ad copy, Marketing Designer, Marketing Video, Marketing Podcast, Sales Assistant, Meeting Assistant, and Marketing Competitor Intel. Analyst agents such as Marketing SEO and Marketing Analytics do not publish, so they are never checked.

Key Capabilities

CapabilityWhat it gives you
Brand consistency scoreA score out of 100 across Voice, Visual, Messaging, and Governance, with a note on each area
Four decisionsAuto-Approve, Review, Revise, or Reject, based on cutoffs you can change
Hard failuresProhibited phrases, missing required disclaimers, unsupported external citations, and unsupported performance claims reject a draft regardless of score
RedlinesThe passage, the problem, and a proposed replacement for each fix
Per-brand rulesEach brand keeps its own prohibited phrases, required disclaimers, voice rules, and cutoffs
On-demand auditPaste a draft into the Brand Auditor screen and get a verdict without waiting for a publish step
Audit historyThe Brand Auditor screen lists recent audits that were not auto-approved, with a score, a decision, and the flags behind it
Topic suggestionsSlack cards of proposed topics you can Greenlight or Dismiss

The Approval Rule

Brand Guard decides nothing about publishing. It supplies a verdict and the producing agent acts on it.

Brand Guard does on its own:

  • Score a draft whenever a publishing agent calls it, in Warn or Enforce mode
  • Run an on-demand audit when you start one
  • Record every audit with its score, decision, flags, and redlines
  • Post a Blocked At Publish notice in Slack when it stops a scheduled blog or social post (see Notifications)
  • Propose topics when you ask for them

You decide:

  • Whether to edit a draft and send it back for another check
  • Whether to Greenlight or Dismiss a suggested topic. Greenlighting moves a topic into the brief and draft stage; it does not publish anything
  • Whether to record an override on an audit that was not auto-approved. Recording an override notes your reason; it does not republish the draft
  • Which mode brand checking runs in, and each brand's rules and cutoffs. Those are set in the Operator Console, by people who have permission to edit governance rules

Who It's For

  • Account managers: see why a draft was held and what to change before client sign-off.
  • Content and social teams: get specific redlines instead of open-ended editorial notes.
  • Agenteous owners: keep brand standards consistent across clients, and choose how strictly they are enforced.