No filming:An AI avatar or clone yourself
    Versely

    What the Editor Agent can do

    The Editor Agent sits beside the web video editor timeline. Six capability groups, five one-click automations, free edits now, credits wait for OK.

    Versely Team••8 min read

    The Editor Agent is the chat that sits beside the timeline on the web video editor. You type what you want in plain words. A free edit lands on the cut immediately. Anything that spends credits is shown as a plan and waits for your OK before it runs. Five one-click automations cover the jobs you will repeat. That is the whole contract.

    Open the video editor. On a wide screen the agent panel is already open next to the timeline. Drop a clip from your library, or go to the Media tab, paste a TikTok, Instagram or X post link, wait for the free preview card, then hit Import. Import costs 1 credit and the credit is refunded if the import fails. YouTube links usually have no downloadable file; the editor says so.

    Uploads, generations and music show as strips in the Image, Video and Audio tabs. The Brand Kit panel applies your fonts and colours. This is the same account as the Android app on your phone. The long cut happens here. The agent is not the workflow canvas: that screen is scenes as columns. This one is a timeline and a chat.

    Six groups of work

    The agent maps a sentence onto six groups. You do not have to name the group. It helps to know they exist, so you do not ask captions to write a headline or a trim to generate a clip.

    Group What it covers
    Cut & arrange Trim, split, reorder, speed, transitions
    Text & captions Titles, caption presets, burned-in subtitles
    Audio & music Voiceover, music beds, levels, effects
    Generate media New clips, images and sound from a prompt
    Canvas & brand Aspect ratio, background, your brand kit
    Finish & export Dub, upscale, clean up, render

    Cut & arrange. Trim a video, split at a timestamp, reorder clips, change speed, add a transition. "Trim the first 1.5 seconds of clip 1." "Split the talking-head at 4.2 seconds." These land as soon as the agent can apply them.

    Text & captions. A line you wrote is a title or overlay. Speech on the tape is a caption burn. Add a text overlay to my video burns a line you authored and never listens to the audio. Caption the whole cut in the agent transcribes audible speech and burns styled subtitles across every clip.

    Audio & music. Voiceover, a music bed, levels, effects. Add music to a video is the task door for a bed. "Balance the audio" is levels on what is already there, not a new track.

    Generate media. New clips, stills and sound from a prompt, dropped onto this timeline. "Generate a three-second clip of a matcha pour and put it after the hook." That spends credits. You will see the plan first.

    Canvas & brand. Aspect ratio, background, the Brand Kit panel. "Set this to 9:16." "Apply my brand kit to the titles." Fonts and colors live in the kit; the brand kit as the style guide is the longer version of that rule.

    Finish & export. Dub, upscale, clean up, then render. "Dub this into Spanish." A dub or an upscale spends credits. The render is the file you publish.

    Most first-day use is a sentence, not a job name.

    Five automations, exact labels

    These sit on the agent panel. Use the label as written, or type the same words.

    Auto-edit for TikTok. Vertical 9:16, a hook title over the open, dead air cut. Picture and a title. It is not a caption burn and it is not a music bed. Then stop and look at the timeline.

    Caption the whole cut. Burned-in subtitles across every clip. Costs credits. The cut must contain audible speech. Trim first, caption second. Basic presets bill 1×. Dynamic presets bill 2× for the same transcription; Glass is the Dynamic default, so an unnamed click is a 2× look. Preview a style on the first few seconds of this file before you caption the whole cut. Output is an MP4 with captions burned in. 165 language codes. Name the language that is actually spoken.

    Score it with music. A generated music bed under the cut, ducked for voice. Costs credits. Confirm the plan. This is a new bed, not a level nudge on a track you already dropped.

    Balance the audio. Voice up front, music under it. Use it after a bed exists. If there is no music yet, score first, then balance.

    What's on my timeline? A plain-words rundown of the project. Run this before you spend anything on a file you did not cut.

    Caption the whole cut and Score it with music always come back as a plan. The others run as edits. If a step would spend credits, you still get a plan and an OK. There is no silent spend.

    Ten instructions you can type

    The automations are shortcuts. The same work is a sentence.

    • What's on my timeline?
    • Trim the first 1.5 seconds of clip 1 and split the talking-head at 4.2 seconds.
    • Auto-edit for TikTok.
    • Add a title that says WAIT FOR IT over the first two seconds.
    • Caption the whole cut in a Basic preset, language en-US. Preview the style on the first few seconds first.
    • Score it with music.
    • Balance the audio.
    • Generate a three-second clip of a matcha pour and drop it after the hook.
    • Apply my brand kit and set the frame to 9:16.
    • Dub this into Spanish, then render.

    The rundown, the trim, the authored title, the audio balance, and a brand-kit plus aspect change are typical free edits. Caption the whole cut, Score it with music, a generated clip, and a dub spend credits: you get a plan, you OK it, then it runs. Auto-edit for TikTok is the named first pass; if a step inside it would spend credits, you will see that plan before anything is billed.

    You can stack a sentence. "Make this 9:16, cut the dead air, put a hook title on the open, then show me what's on my timeline." Do not stack Caption the whole cut onto a cut you have not locked. Caption last. Generate B-roll before you caption, not after.

    You do not need the exact label. "Put subtitles on this" maps to Caption the whole cut. "Give it a bed and duck it" maps to Score it with music. "Tell me what I have here" maps to What's on my timeline? The sentence is how you add a language, a Basic preset, a timestamp, or a clip number. A silent product plate has nothing to transcribe. A headline you wrote is not a caption job.

    Free now, credits after you OK

    Two lanes.

    Free edits land instantly. Trim, split, reorder, speed, a title you typed, aspect ratio, a brand-kit apply, a rundown of the timeline, a level balance. You type it. The cut updates.

    Anything that spends credits is a plan. Caption the whole cut. Score it with music. Generate a new clip, image or sound. Dub. Upscale. Social import at 1 credit. The agent shows the plan and waits. You OK, or you do not. Close the plan and nothing is billed.

    Every plan gets the same models and editor. Credits are the volume, not a feature gate. Read pricing for the credit buckets. Do not invent a number in the prompt. Read the quote on the plan.

    Lock picture before a credit job. Get media on the timeline, ask What's on my timeline?, cut until the picture is the one you will publish, then generate missing B-roll, score and balance audio, add authored overlays, and caption last. Dub or upscale only if you still need them. Then render.

    Three moves on the first cut

    1. Open the video editor, drop a clip (or Import a TikTok, Instagram or X link from the Media tab), and type What's on my timeline? Confirm the agent can see the file.
    2. Click Auto-edit for TikTok, or type a trim: the dead air at the start, a split on the first sentence. Watch the timeline move. That is the free lane.
    3. If someone is talking, preview a caption style on the first few seconds, then click Caption the whole cut. Read the plan. OK it only if the cut is locked. If you need a bed, Score it with music is the next plan, then Balance the audio.

    Do not caption a moving cut. Do not paste a YouTube URL and expect Import to fetch a file. Start with the rundown, lock the picture, spend credits last.