Build characters, arcs and scenes with a local model reading over your shoulder. It suggests; you decide what lands in the manuscript. Read any scene back aloud through the local voicebox.
A chat client pointed at your own endpoint. Create personas, switch between them mid-thread, and let the voicebox read the reply back.
Plain markdown, start to finish. The model drafts, rewrites and outlines in place. The file you export is a file, not a database row.
Genuine People Personalities and Purposeful Perspective Shifting, written as structured fields. Assign a voice while you build, then drop the persona straight into Chat, Character or Producer.
Script, cast and render spoken audio through the local voicebox. Segments, takes and voice assignments stay editable until you're happy with the cut.