No LLM
Every alert comes from a fact you confirmed, not a guess. No hallucinations.
Most writing tools now push AI that drafts your prose. Trame takes the opposite bet: your words stay yours. Trame doesn't write your story — it remembers it.
You write a scene, naturally.
Trame reads it locally and proposes what matters: characters, places, facts.
You confirm what's true with one click. That builds your Nexus — your story's memory.
As you keep writing, Cortex re-reads each scene and flags when it contradicts something you confirmed: a dead character acting, a blind character seeing, a prisoner walking free.
Every alert comes from a fact you confirmed, not a guess. No hallucinations.
Your manuscript never leaves Trame. No cloud, no model trained on your book.
If Trame flags something, it can tell you exactly why and when you established the constraint.
“Trame caught it. Without any AI.”
Cortex and Nexus run locally — no cloud upload, no model trained on your work. If we ever add an optional AI audit, it will only run when you explicitly ask for it, one scene at a time.
| Feature | Free | Plus |
|---|---|---|
| Editor + chapters & scenes | ✓ | ✓ |
| Nexus (sheets, attributes, relations) | ✓ | ✓ |
| “Add to Nexus” suggestions from your manuscript | ✓ | ✓ |
| Cortex coherence alerts (dead-active, constraints, contradictions) | — | ✓ |
| Confirmable narrative facts (Remember / Dismiss) | Limited | ✓ |
| Full active story memory | — | ✓ |
| No AI, 100% local | ✓ | ✓ |
Free remembers your characters. Plus catches your continuity errors before your readers do.