Brand brief
Attach a brand kit as the brief and the agent works from your colours, font and logo. Kits live under Brand kits in the app.
Haze Agent
One agent. Seven ways to make a video.
Give the agent what you have: a script, a short voiceover, a track section, or a video you like. It generates each scene, cuts the result together, and drops the finished video in your projects, ready to clip, caption and publish. It can also turn a script into an explainer, bring a photo to life, or make a video of you from one photo.
AI video is priced in credits. The composer shows what a video will cost before anything is made.
The agent lives in the signed-in composer. Not a member yet? Start free first.
Made with the agent
A real generated video, not a mock-up
This clip started as a few sentences of script. The agent generated a scene for each one, read them in an AI voice, and cut the result together. It then went through the ordinary clipper for captions, the same as any upload. The other 4 are single-scene previews of the visual styles the agent generates in.
- The Real Work in Video Making Isn't EditingHype Burst · 7s
- The Two-Minute RuleClaymation · 5s
- Every Expert Was Once a BeginnerWatercolour · 5s
- I Walked Out of That Office for GoodAnime · 5s
- Why a Tiny Habit Beats a Giant Goal3D cartoon · 5s
Four ways in
Start from whatever you already have. Each mode ends the same way: a finished video in your projects. Next to these four, the composer has Explainer, Image to video and AI clone.
Script to video
Write the narration. Every sentence becomes a scene.
Paste or write a script. The agent splits it into scenes, generates a shot for each one, reads the whole thing in an AI voice, and cuts it together in the order you wrote it.
You provide
- A script, as long or short as you like
- An aspect ratio: 9:16, 1:1 or 16:9
You get
- One video, one scene per sentence, narrated
- The per-scene cost before you commit
Illustration
- Most people think editing is the hard part.
- The hard part is deciding what the video is about.
- So start with the question, not the footage.
AI voice · one read of the whole scriptnarrated
Audio to video
Upload a short voiceover. Get scenes cut to it.
Upload a short voiceover, hook or verse, up to about 90 seconds. The agent listens, plans a scene for each beat, generates the visuals, and cuts them to your audio. Your recording stays exactly as it is. Generation is metered per scene, which is why uploads are kept short.
You provide
- A short voiceover, hook or verse (up to about 90 seconds)
- An aspect ratio
You get
- A video timed to your recording
- Your original audio as the soundtrack
Illustration
Your recording · untouchedup to ~1:30
Song to video
Upload a short track section. Get a visual cut to it.
Drop in a short section of a song or an instrumental: a hook, a chorus, a verse. The agent builds a sequence of generated scenes cut to it, so you have a music video or a visual bed for a lyric clip without shooting anything.
You provide
- A short track section
- An aspect ratio
You get
- A video cut to the music
- Your track as the soundtrack
Illustration
Your track section · scenes cut on the beata hook or a verse
Reference to videoBeta
Point at a video you like. Get its structure, on your topic.
Point at a video you like, a YouTube link or an upload, and the agent writes a new script in that structure and pacing about your topic. You edit the script, see the price, then generate.
You provide
- A YouTube link or an upload to use as the reference
- Your topic
You get
- A new script in that structure and pacing, yours to edit
- The price, then the video, when you say so
Illustration
A video you like
reference · read for structure
- 1Hook
- 2Setup
- 3Turn
- 4Payoff
Your topic
A script in that shape, ready for script to video.Same controls, every mode
Three formats
9:16 for Shorts, Reels and TikTok. 1:1 for feeds. 16:9 for YouTube and the web. Pick it before you generate.
Cost before commit
The composer prices the whole video before anything is made — scene by scene in the scene modes — and nothing runs until you accept it.
After generation
The video lands in your projects like any upload.
Nothing about a generated video is special once it exists. It is a source in your projects, so everything Haze Clips does to an upload, it does to this.
- 01
It lands in your projects
A generated video is stored as an ordinary source, next to your uploads.
- 02
Clip it, caption it
Run the clipper on it, restyle the captions, pick a template, the same as any upload.
- 03
Publish it
Post or schedule to your connected channels from the editor. No watermark on paid plans.
Haze Agent FAQ
What is the Haze Agent?
The generation side of Haze Clips. Instead of starting from a video you recorded, you start from a script, an audio file, a song, or a reference video, and the agent generates the scenes and cuts the result together. It lives in the composer at /create once you are signed in.
What does it cost?
AI video is priced in credits: 1 credit is 1 second of generated video, and an explainer is 1 credit per 5 seconds. Before anything is made, the composer shows what it will cost, and nothing runs until you accept that figure. Planning and pricing a video is free, and a job that fails is refunded in full.
Which modes are there?
Script to video, audio to video and song to video; reference to video (beta), which writes a new script in the structure of a video you like, for you to edit; explainer, which draws a narrated explainer with maps, photos, counters and charts from your script; image to video, which turns one still into a 5 or 10 second clip; and AI clone (beta), which makes a video of you saying your script from one photo. Every mode lives in the composer at /create, and a mode that is not switched on for your account yet says so there.
Whose voice reads the script?
The voice you pick. The voice picker has accent and language filters, with native Latin American Spanish, Brazilian Portuguese, Arabic and Korean narrators, and on paid plans you can clone your own voice. In audio to video your own recording is the soundtrack and stays exactly as you uploaded it; the agent only generates the picture.
How long can the audio be?
Short. Audio to video takes a voiceover, hook or verse of up to about 90 seconds, and song to video takes a short section of a track. Generation is metered per scene, so the uploads are kept short by design.
Can I clip a generated video?
Yes. The finished video lands in your projects as an ordinary upload, so the clipper, captions, templates and publishing all work on it.
Ready to Create Viral Clips?
Turn your long videos into short, viral content.
- No Credit Card
- Free Forever
- 10 Free Clips
- Cancel Anytime