Product pillars

Guardrails that serve the story—not the model.

Every proposed Italus control gives the author a clearer way to define intent, evaluate output, and understand the cost and history of AI-assisted work.

01 · Canon

A single source of narrative truth.

Keep the people, places, chronology, lore, and promises of a novel available as active constraints—not scattered notes the model may overlook.

Character canon

Identity, motivation, relationships, knowledge, and changes over time.

World canon

Locations, terminology, systems, rules, and boundaries of the setting.

Timeline

Sequence events and flag contradictions before they compound.

Scene context

Bring only the relevant facts into a specific generation task.

02 · Prose rules

Tell the system how not to write.

Positive direction matters, but strong editing also requires limits. Italus is designed to let authors identify habits that flatten voice or make prose feel machine-produced.

Example controls

  • Limit or prohibit em dashes.
  • Flag repeated sentence openings.
  • Protect viewpoint and tense.
  • Restrict filter words and filler transitions.
  • Track banned phrases and genre clichés.
  • Set preferences for dialogue, exposition, and pacing.

These examples describe the intended product direction; final controls will be confirmed before application launch.

03 · Creative ledger

Provenance authors can understand.

The proposed ledger records significant generations, revisions, author edits, resubmissions, accepted versions, and token usage. It supports transparency without making unverified legal or publishing claims.

01

Trace revisions

Understand which direction produced each version and what the author changed afterward.

02

Mark acceptance

Distinguish raw machine output from author-approved manuscript prose.

03

Export a history

Prepare a clear project narrative for personal records or publishing conversations.

04 · Budget visibility

Creative freedom with cost awareness.

Track token usage by generation, chapter, and project so experimentation does not hide its cost. Future pricing will be published only after the application’s provider and billing model is confirmed.