Content rules
Plain-English brand voice rules every blog writer reads. Always write in second person. Use British English. Toggleable rules.
Executive summary
Three short paragraphs explaining the feature and value.
Content rules is a list of plain English brand voice rules that the writer pipeline reads before every blog draft, audit summary, or any other AI generated content piece in the workspace. Rules look like sentences: always write in second person, never start a sentence with we, use British English, avoid corporate jargon, keep paragraphs under five sentences. Each rule is independently toggleable.
Brand voice is the hardest thing for AI writers to capture from a profile alone. Even with a strong Brand profile, the model drifts toward generic patterns over time. Content rules act as the explicit guardrails that override drift. A brand using GetXEO for a year accumulates twenty to fifty rules tuned to its specific voice, each born from a moment where the writer pipeline produced something off brand.
The interface makes rule management practical. Rules can be added inline by clicking edit rule on any generated content, toggled on or off without deletion, grouped by category (voice, structure, terminology, formatting), and applied workspace-wide or scoped to specific blog categories. The writer pipeline reads the active rule set before every draft so changes propagate immediately to subsequent content.
Key highlights
Five capability points teams should know about quickly.
- Plain English brand voice rules read before every draft
- Always write in second person, use British English, etc
- Independently toggleable rules without deletion
- Override AI model drift toward generic patterns
- Twenty to fifty rules per workspace after one year
Top FAQs
Five common questions answered for fast practical clarity.
How do I add a new rule?
Two paths. The Content rules surface has an add rule affordance accepting plain English sentence input. Or from any generated content piece, click edit rule next to a sentence that violated brand voice; the system generates a candidate rule that would have prevented that output, which the user accepts, edits, or rejects. The second path is how most rules naturally accumulate over time.
Are rules workspace-wide or per blog category?
Both, configurable per rule. The default scope is workspace-wide. Each rule can be optionally scoped to specific blog categories (apply only to Custom blogs, only to technical content, only to founder letters). Scoped rules let the brand maintain different voice profiles for different content types. The writer pipeline reads applicable scoped plus workspace rules per draft.
Can I toggle rules off temporarily?
Yes. Each rule has an active toggle. Disabled rules stay in the system (with their history and the example that triggered them) but the writer pipeline ignores them until reactivated. Useful when experimenting with a different voice for a specific content piece without permanently removing a rule that mostly serves the brand well across other content.
How does the pipeline enforce rules?
Rules feed into the system prompt for the writer model before every draft. The model is instructed to apply each active rule as a constraint during generation. The model also runs a self-check pass against the rule set before returning the draft. Final human review catches anything the model self-check missed. Three-layer enforcement keeps voice drift bounded effectively.
Do rules conflict with each other?
Sometimes. When the rule set grows, contradictions can creep in. The Content rules surface includes a conflict detection check that flags rule pairs likely to conflict, with suggested resolutions. Users review conflicts and either remove one rule, merge two into a clearer single rule, or scope each to non-overlapping categories. Conflict checking runs automatically on each rule addition.