Guide

Rules, presets, and contexts

How the three layers work together, example stacks for different channels, and how to avoid rule conflicts that make AI output rigid or contradictory.

Published

Last updated

Pro-mode tooling stacks three layers: presets define voice, rules define hard constraints, and contexts add project-specific background. Used well, they reduce editing time dramatically. Used poorly, they fight each other and produce wooden text.

Presets

Broad style guidance applied by default. Think: “How should we sound in most videos?” Keep presets short enough to read in one pass—if longer than a paragraph, split into preset + rules.

Rules

Specific instructions: legal disclaimers, banned phrases (“guaranteed results”), required segments (“include sponsor block”), formatting constraints (“always three bullet takeaways”). Rules are not optional suggestions—only add them when violation is unacceptable.

  • Good rule: “When discussing health, include ‘not medical advice’ once per script.”

  • Weak rule: “Be engaging”—that belongs in the preset.

Contexts

Optional background for a project or campaign: audience level, product constraints, competitor names to avoid, or event timing. Contexts prevent you from pasting the same brief into every prompt.

Stacking without conflicts

  1. List all rules and read them for contradictions (“be brief” vs “always include five examples”).

  2. Resolve conflicts by priority: legal and safety beat style.

  3. Remove the lowest-value rule first when output feels rigid.

  4. Test with one section before generating long scripts.

Maintenance

Quarterly audit: delete rules nobody needed in 90 days. Presets evolve with your channel—update examples when your voice matures.

If generations feel wooden, remove rules before adding more—constraint overload is the silent killer of voice.


Name contexts clearly (“Q2 launch — no pricing mention”) so collaborators pick the right scope.