Everything in v0.1, tagged honestly.

MP4, MOV, MKV, AVI, WebM and M4V, from a file or your Workspace library.

Paste a script, or describe an idea and let the co-pilot draft one.

Note: Drafting needs your Anthropic key; pasting a script does not.

Record or upload a voiceover and the script's words align to your voice.

Note: Needs the optional transcription pack.

Group videos with their images, audio, music and sound effects. Import once, reuse everywhere.

WhisperX (large-v3) with alignment. CUDA when you have it, CPU when you don't.

Note: English only today. Optional install; the first run downloads about 3.5 GB of models.

The transcript is split into 3 to 8 balanced, titled scenes to start from.

Pauses are detected as cut proposals.

Note: Available to the co-pilot; no Story button yet.

Click a word to seek. Drag across words to cut or restore. The gap closes on its own.

Drag scenes to reorder. An arc strip colours each scene by role: Hook, Problem, Example, Payoff, Wrap.

Picture, B-roll, Graphics, Music and Captions lanes with a waveform, in sync with the script.

The whole video as a readable script.mdx you can copy.

Note: View-only for now; editing the source is planned.

Drag B-roll, graphics, music or a sound effect onto a word. Suggestions are ranked against the phrase you selected.

Your footage, a full-frame animation over your audio, or a character cameo landing on a chosen word.

Stage, Phone screen (up to three phones), Title card and Sign-off; push-in, pull-out and pan; seven backdrops.

A co-star and up to three more performers. Face-offs, fist bumps, high fives, think and say bubbles.

Pick an expression, pose and move, try takes and read critic notes. Or describe the shot in words.

Note: Describing a shot in words needs your Anthropic key.

Lower third, callout, evidence card, chapter card and a simple chart.

Posters drawn by the same engine, saved as PNG.

Add new drawn characters without touching the engine.

Note: Developer-oriented: packs are trusted local code.

Plan scenes, cut filler words, suggest B-roll and set a shot, grounded in your selection.

Note: Optional. Uses your own Anthropic API key, billed by Anthropic.

Finds ums, false starts and repeats in the transcript. Accept them all as one undo step.

Note: Speech recognisers drop some fillers, so results vary by recording.

A read-only planning mode, per-video notes and recall of what you accepted or rejected before.

Note: In the backend; no toggle in the Ask bar yet.

A few words at a time with the active word highlighted, identical in preview and export.

Alternative caption looks.

Note: In the export pipeline; no control in the UI yet.

YouTube 16:9, 4K, Shorts and Reels 9:16, Square 1:1, with per-scene progress, cancel and retry.

NVIDIA NVENC when present, libx264 on the CPU otherwise. Audio normalised to -14 LUFS.

Unchanged animated scenes are reused on re-export.

Keep the speaker centred when cropping to 9:16.

Note: Defaults to a centre crop; tracking needs optional dependencies.

Work across machines or with a team.

Note: Not built. Each installation is standalone by design today.