DAIDalusGet Early Access

Private beta · 2026

A professional video editing workspace with an approval-driven AI copilot.

The AI editor for people who actually edit. AI proposes. You decide. Nothing ships you didn’t approve.

Get Early Access

Editor preview · real screenshots in private beta

DAIDalus principles

FIG 0.1

Hybrid surface

Transcript and timeline are equal partners. Edit through text when speed matters. Trim on the timeline when precision does. Both views read from the same composition.

FIG 0.2

Approval-driven

Every AI action is a proposal. You accept or reject each suggestion. Nothing the copilot does ships without your explicit sign-off.

FIG 0.3

Local-first

Your media never leaves your machine. No upload step, no cloud lock-in, no streaming compromise. Renders happen locally with FFmpeg.

The hybrid editor in motion

1.0   Transcript surface

Edit through the transcript when speed matters.

Delete a sentence and the footage is cut. Rearrange paragraphs and the composition reflows. Strip filler words in bulk. The timeline updates as you read.

1.1   Timeline surface

Edit on the timeline when precision matters.

Frame-accurate trims, ripple edits, cross-dissolves, multi-track audio. A real NLE under the hood. Both surfaces read from the same backend composition.

2.0   The copilot loop

Every AI action goes through you.

The copilot watches, proposes, and waits. You accept what fits, reject what doesn’t. Nothing ships you didn’t approve.

Copilot panel · real screenshots in private beta

Accept

One keystroke commits the proposed change to the composition.

Reject

Dismiss the proposal. The copilot never argues; it moves on.

Preview before commit

Scrub the suggested range. See exactly what gets cut before approving.

3.0   Under the hood

A real tool, not a wrapper.

  • Frame-accurate

    Real NLE engine under the hood. Trim to the frame, not the second.

  • Real transitions

    Cross-dissolve, cube, slide, dip-to-black. Rendered via FFmpeg.

  • Local preview

    Remotion-powered, instant scrubbing. No render queue.

  • 25 mutation types

    Hand-built composition engine. Optimistic concurrency, single revision per edit.

  • Multi-track audio

    Separate audio + video tracks. Caption tracks aligned to the transcript.

4.0   Private beta

Get early access.

Drop your email. We’ll be in touch when a seat opens.