Editor AI

Where Video editor › AI mode: On › ChatCost 20 credits, or 10 credits for a follow-up

In AI mode, you edit by saying what you want. The AI plans the change and applies it to your edit as a single step, and one Undo takes all of it back. It can cut and reorder clips, remove silences and filler words, write and style captions, clean up audio, add titles, change the format, apply a brand kit, record a voiceover, add music, and propose new footage for you to generate.

Ask for a change

Turn AI mode on in the editor’s header. The Chat tab opens on the left.
Type what you want in the box at the foot of the chat, for example “Cut the pauses, add captions and make it vertical.”
Press Enter (⇧+Enter for a new line). The chat shows what it’s doing, such as Working out the edit… or Listening to the speech…, with a timer.
When it’s done, the answer lists what changed and what it cost, marked Applied. The playhead jumps to the first change.
Not what you wanted? Press Undo on the answer, or ask for something else.

The request keeps running even if you close the window, and the chat picks it up again when you come back.

Suggestions

Above the box, the chat suggests next steps, such as Add captions, Cut the silences, Clean up the audio, Make it vertical, Add a title or Tighten it to 30 seconds. Press one to write it into the box; press again to send.

What you can ask for

Area	Examples
Cutting	“Cut the silences”, “Remove the filler words”, “Tighten it to 30 seconds”, “Move the product shot to the start”
Captions	“Add captions”, “Highlight the spoken word”, “Captions in Hinglish”, “Translate the captions to Spanish”
Sound	“Clean up the audio”, “Duck the music under the speech”, “Match the loudness for YouTube”, “Mute A2”
Voice and music	“Add a voiceover that says…”, “Read the captions aloud in a calm voice”, “Add calm music under it”, “Put Quiet lift from 0:10, quieter”
Picture	“Make this scene warmer” (with a frame attached), “Add a cross dissolve between every clip”, “Stabilise the second clip”, “Blur the face with a mask”
Text and brand	“Add a title at the start”, “Use my Acme brand kit”, “Put the logo from example.com top left for the whole video”
Format	“Make it vertical”, “Reframe for TikTok following the speaker”
New footage	“Generate a five-second shot of a city at dusk and put it at the start”
Export	“Export an Instagram version of the in-to-out range”

The AI knows the playhead, so “here” and “this clip” work.

Show it what you mean

The AI reads your edit’s clips, timing, titles and captions, but it doesn’t see the picture or hear the sound unless you show it. To point at something:

Attach frames. Press This frame on the frame strip under the preview, or click a still on the strip. Up to four frames per message.
Attach media. Press + beside the box, then Import from this device or Choose from your media (up to four files). Then say what to do: “Put it top right, small.”
Name a website. The AI can read a site’s logo and pictures, for example “Put the logo from example.com in the corner.”
In manual mode, right-click a clip and choose Ask AI about this frame.

Proposals you start yourself

Some things are proposed as a card and only start when you press the button:

New video or pictures: the card shows the description, length, format, where it will go, and the price. Press Generate. You can keep editing; it’s added to the edit when ready.
Auto reframe: the card shows the format, what it will follow, and the price. Press Reframe…
Export: press Export… to open the export dialog with the AI’s settings.

Free commands

These exact phrases run the tool directly, with no AI request and no credits:

remove silence, cut pauses, remove filler words, find scene cuts, detect scenes, clean up audio, studio sound, remove background noise, denoise, enhance voice, level loudness, normalize audio, remove hum, reduce reverb, remove echo.

The Cut the silences and Clean up the audio suggestions are free commands too.

Cost

The price of your next message is shown beside the send button:

You see	Meaning
Uses 20 credits · 1,240 left	The first message of a conversation.
Follow-up · 10 credits · 1,240 left	A message within 30 minutes of your last charged one.
Needs …	Not enough credits. The credits dialog opens if you send.

One request’s price includes the captions it listens for, the voiceover it records and the music it composes. A video or picture it proposes is charged at its own price when you press Generate. A reply that changes nothing, a request that fails, and the free commands cost nothing.

What it can’t do

Bring in a file on its own: attach it with +.
Remove a background that isn’t a green or blue screen (use the Ultra Key effect for those).
Separate one voice from other voices, or a voice from music.

When you ask for something it can’t do, it says so in its answer.

Limits

One request at a time, from 3 to 100,000 characters.
Up to four attached files and four frames per message.

Troubleshooting

You see	What to do
Add footage to this edit first, or ask for a voiceover, music or a generated shot.	Add media to the edit, or ask for something that doesn’t need footage.
The connection dropped while waiting for the answer. Your request is not lost.	Press Check again.
This request was still being made.	Press Check for its answer.
…has changed since this answer was made, so it is not in the edit.	You changed the edit while the AI worked. Press Ask again; the credits for that answer were given back.
Stopped at “…”: … What came before it was kept.	Part of the answer couldn’t be applied. Undo it, or ask for the rest again.
Apply to the edit as it is now	The edit is open in another window. Apply the answer here if you want it.
Not sent: …	The request was refused before it started. Nothing was charged.