Runs

Now in private testing with creators

Your editing style, applied to new footage.

Runs is a personal AI video editor trained on videos you’ve already edited. Most AI editors learn what works for everyone — this one learns how you work, then shapes new footage the way you would. An editable first version, not a template.

Opens a short application — tell us what you make. We review each one personally and bring on new creators in small groups.

For creators who edit their own Reels · Powered by AY-CUT

How I edit my videos

A real cut from @ayukoene, with the numbers measured from it.

5.6 cuts / 10s · median shot 1.60s

What it learns

It reads the decisions you stopped noticing

reads it now

Your pacing

Cuts as fast — or as slow — as you do.

reads it now

Your captions

Your wording, your spot on the screen.

reads it now

Your sound

Levelled the way you deliver. Music-free.

reads it now

Your structure

Opens fast, lands the way you would.

learning next

Your moments

Learns which takes you'd keep.

learning next

Your corrections

Every fix teaches it more of you.

The idea

Every creator has an editorial signature

It shows up in decisions you no longer notice making. One video looks like a template. A body of work reveals the editor.

How it works

From your old edits to your next cut

Start here, finish anywhere. Generate an editable first version shaped by your previous work, then keep refining it in the editing tools you already use — Edits, CapCut, Premiere or by hand.

  1. 01

    Add your edits

    Five videos that feel like you. It finds what repeats across them.

  2. 02

    Meet your signature

    Plain findings, each with the number behind it. Keep or correct.

  3. 03

    Add new footage

    A short brief — what it's about, what matters, how long.

  4. 04

    Get your first cut

    Editable from the first frame. Every change teaches it more.

An editable first cut, shaped by your profile — never a locked render.

A creator's body of work

This is the kind of work it learns from

Published reels from @ayukoene. A body of work like this is where AY-CUT reads a signature — not the output, the input.

Not a preset

A persistent editor

Presets remember settings. Templates remember structures. This remembers decisions — so the more you use it, the less you explain.

What exists today

The engine already works

This is the first phase — the mechanics are real and already running. What it hasn't learned yet is you.

  • Reads finished edits into a profile of how you cut.
  • Applies that profile to new footage.
  • Gives two creators visibly different cuts from the same clips.
  • Returns a real, editable timeline — never a locked render.
  • Re-measures its own output, so a signature is evidence, not a vibe.

The experiment

Can a creator recognise their own edit?

The best reaction isn’t “it saved me three hours.” It’s “it chose the exact moment I would have.” That’s the question the first creators in are helping answer.

Your data

Your work trains your editor. No one else’s.

Your footage is private by default, and your signature is never used to train a shared model without your say-so. You control:

  • Which videos become examples.
  • Which findings you accept.
  • Which edits update your profile.
  • What is stored, and what is deleted.

You developed the taste. Let your editor learn it.

Opens a short application — tell us what you make. We review each one personally and bring on new creators in small groups.

Private testing is open to creators with an existing body of edited work.