01 / STRUCTURE
Shape the timeline
Turn a script, a JSON brief and rights-described local assets into a deterministic timeline.
A production engine for considered stories
Make the story.
Keep the thread.
NarrativeCut helps solo creators and small teams turn scripts and rights-described local assets into structured documentary and explainer productions.
Start a conversation ↗SCRIPT → TIMELINE → FINISHED CUT · SEE A CLAUDE PLAN ↓
THE CUT
01 / SCRIPT
02 / ASSETS
03 / TIMELINE
04 / QUALITY
Story, assembled with careNC—001
From research to a reviewable cut.
One connected production workflow
02 / SUBTITLES
Keep words in sync
Generate subtitles as part of the same production flow.
03 / RIGHTS CONTEXT
Track asset provenance
Carry rights descriptions for local assets into a clear license report.
04 / QUALITY
Check before delivery
Optional local FFmpeg rendering and automated QC require FFmpeg, narration and decodable media.
Built for small teams
More attention on the story. Less friction between the steps.
NarrativeCut is designed for independent creators and small production teams working on long-form documentary and explainer content. It connects script, local media, timeline, subtitles, rights notes and optional rendering in one structured workflow.
Claude scene planning, on a real run.
Unedited output · claude-opus-5-5
Inside were more than 3,000 letters written by factory workers between 1920 and 1950. Most had never been read by anyone outside the families who donated them.
Asset search query
handwritten factory worker letters 1920s 1940s archive boxes
On-screen text
3,000+ letters, 1920–1950
Rationale
The count and date range are specific claims best shown with the physical letters themselves.
She noticed that dozens mentioned the same strike in the winter of 1934.
Asset search query
1934 letter mentioning factory strike close-up handwriting
On-screen text
Winter 1934
Rationale
The pattern across letters is the central discovery and should be shown through the letters that mention the strike.
Newspapers from that year barely covered the strike.
Asset search query
1934 local newspaper front page archive microfilm
Rationale
Contrasting the scant press coverage with the letters needs period newspaper pages as evidence.
Editorial risks flagged by Claude (2 of 7)
- The claim that newspapers barely covered the strike needs verification through a review of 1934 local press.
- Archival images of factory gates and strikes must be from the actual site or clearly labelled as illustrative to avoid implying they show this strike.
Input: a fictional eight-beat demo script. Claude groups beats into scenes and proposes visuals as schema-validated JSON; NarrativeCut checks every beat is covered once, in order, and keeps timing deterministic. Full scene-plan.json ↗
About / Current status
Bootstrapped. Open source. Early-stage.
NarrativeCut is a bootstrapped early-stage project founded in 2026 by solo founder İlker Cem Tanrıverdi.
Project founding date: September 2026. Not yet incorporated. Current status: early-stage development.
Open-source core available on GitHub ↗ · Public releases ↗
The current core produces a deterministic timeline, subtitles and an asset-license report from scripts and rights-described local assets, with optional FFmpeg render and automated QC. Rights metadata records supplied descriptions; it does not verify legal ownership. An optional Claude-powered planning step is implemented: it groups script beats into editorial scenes and proposes visual roles, asset search queries and fact-checking risks as validated JSON. A saved plan now guides deterministic asset selection; on the demo it changes the chosen asset on 4 of 8 beats.