Troubleshooting

Frequently Asked Questions

What Score Does a Draft Need to Pass Automatically?

The default approval threshold is 85 out of 100. Drafts at or above 85 with no critical flags are approved without any action from your team. Drafts between 70 and 84 are flagged for your review; the redlines are advisory. Drafts between 50 and 69 are sent back for revision. Drafts below 50 require significant rework and are rejected.

These thresholds can be adjusted per client in the dashboard. Check the client's brand configuration if the threshold seems wrong for a particular account.

What Are the Four Dimensions Brand Guard Scores Against?

Every draft is scored across:

• Voice (up to 30 points): tone, register, vocabulary, and sentence rhythm match for that client's established voice.
• Visual (up to 25 points): consistency of any attached images or visual elements against the client's visual standards. Text-only drafts receive full marks on this dimension by default.
• Messaging (up to 25 points): the topic sits on one of the client's content pillars, the positioning is consistent with what the client publicly defends, and the call to action fits the client's funnel stage.
• Governance (up to 20 points): required disclaimers are present, claims are defensible, and no prohibited phrases appear.

Can I Override a Block?

Yes. When Brand Guard blocks a draft, the calling agent's card (in Marketing Content, Marketing Social, or Marketing Email) includes a "Publish now" option. Clicking it overrides the block and sends the content forward. The override is logged with your name and the draft reference so it can be reviewed later.

Does Brand Guard Keep a Record of Every Decision?

Yes. Every audit is recorded: the client, the score, the verdict, the flags, and any redlines. You can retrieve past decisions by asking Brand Guard in the brand-guard channel. The app surface also shows audit history per client.

What Counts as a Prohibited Phrase?

Prohibited phrases are configured per client in the dashboard. The platform also maintains a set of default phrases that apply to all clients: over-claims like "guaranteed ROI," generic filler phrases that signal AI-generated text, and any phrases the client's brand rules specifically forbid. Brand Guard checks for these before running the full score.

Why Did Brand Guard Flag Something as an Ai-Text Pattern?

Brand Guard looks for phrases and sentence structures that appear very frequently in AI-generated content, regardless of whether the draft was written by a person or an agent. These include generic openers, certain cliche phrases, and perfectly uniform list structures. Getting flagged here does not mean the content is low quality; it means the draft has a pattern that tends to read as impersonal or generic. The redline will show the specific phrase and a concrete alternative.

Common Issues

Drafts Keep Getting Blocked for the Same Issue

If the same client's drafts are repeatedly blocked for the same reason:

1. Check whether the per-client brand rules in the dashboard match what that client actually expects. If the rules are stricter than intended, an operator can adjust the threshold or prohibited-phrase list.
2. Review the redlines from recent blocks. If the pattern is consistent (for example, a recurring CTA format that triggers a flag), consider adjusting the brief template the drafting agent uses so the issue is caught at the source.
3. If the same issue appears across multiple clients, ask Brand Guard in the brand-guard channel to help identify the pattern and what rule is driving it.

Brand Guard Applied the Wrong Client's Rules

Each client's rules are applied based on the brand identifier the drafting agent passes through the publish flow. If Brand Guard appears to have scored a draft against the wrong client:

1. Confirm that the drafting agent had the correct client selected when the draft was created. The client context is set at the start of the drafting session.
2. If the brand identifier looks correct but the rules seem off, check the client's brand configuration in the dashboard to confirm the prohibited phrases, voice rules, and thresholds are set up as expected.
3. Contact support@agenteous.com if the mismatch persists after checking both.

Brand Guard Is Not Responding in the Brand-Guard Channel

1. Check that you are posting in the correct Slack channel for Brand Guard.
2. Confirm that the mention format is correct: @Agenteous followed by your question.
3. If no response arrives within five minutes, the interaction router may need attention. Check the Agenteous dashboard for any service alerts.

Auto-Approval Is Not Working for a Client That Meets the Threshold

Auto-approval applies when the score meets or exceeds the threshold AND no high or critical flags were raised. If a draft that scored above 85 was not auto-approved:

1. Look at the flags in the block notice. A high-severity flag (such as a missing required disclaimer) will prevent auto-approval even when the total score is above the threshold.
2. Check whether the client's brand configuration has a higher threshold set than the platform default.
3. If neither explains it, ask Brand Guard in the brand-guard channel to explain why that specific audit did not auto-approve; it will retrieve the recorded decision and walk through the reasoning.