AI Video Agent: Describe the Video in Chat, Get It Finished

What an AI video agent does differently

A video form asks you to fill in every field before anything happens. An agent works the other way around: you say what you want, the way you would brief a person, and it works out the rest. It keeps the whole conversation in mind, so a later message such as "make it one minute and landscape" changes only what you changed.

FacelessCast's agent turns the conversation into a render brief: the topic, the format (a narrated explainer, a promo with a presenter, or a story with characters), the length, the frame, the language, the voice, captions and music. When the brief is complete it shows a plan card with the price in credits and waits for your answer.

What you can ask for in plain words

Everything the studio can make is reachable from the chat. These are requests it understands and turns into settings:

• A length from 15 seconds up to the longest your plan renders: 3 minutes on Starter, 5 on Creator, 10 on Pro, 20 on Studio. "20 seconds" is made as 20 seconds, not rounded up.
• A frame: vertical 9:16 for Shorts, TikTok and Reels (the default), landscape 16:9 for YouTube, or square.
• A language: it replies in the language you write in and narrates in any of 37 languages.
• A woman's or a man's narrator, captions on or off, no background music or a mood such as calm, epic or upbeat.
• A resolution up to what your plan renders: 4K on Pro and Studio, 1440p on Creator, 1080p on Starter.
• Your own script, read word for word, or a link to a product the video should sell.
• Versions of the finished video in another interface language (English, Turkish, German, Spanish or French) or with other opening hooks to test.

What it asks, and what it decides for you

The agent asks one question at a time, with a few answers you can tap, and only for what changes the video and is truly missing: what the video is about when you have not said, the kind of scenes when you have not said it, and the length or platform when nothing hints at it.

Everything else gets a sensible default that you can see on the plan card: the format, the tone, the visual style, the presenter of a promo. It does not ask again what you already wrote, so "a 30 second AI video about honey bees" goes straight to a plan.

The price before anything starts

When the brief is complete, the plan card shows what the video costs in credits and how many you have. It starts only when you tap Accept or reply yes to that exact price, so a video never begins behind your back.

An AI video, where every scene is a generated clip, is priced by the clips its length needs: about 140 credits for 30 seconds at the standard motion quality. Stock footage uses no credits. If your balance does not cover an AI video, the agent offers the ways on: a lighter motion quality, a shorter length, stock footage instead, or a credit pack.

If an AI video cannot be finished (a scene still fails after several tries), the render stops and every credit is refunded, rather than handing you a video with a still picture in it.

Two kinds of scenes, and how long each takes

Renders run in the background, and the page tells you how long yours will take when it starts, so you can plan the next video meanwhile.

• AI video: every scene is a generated clip that moves; presenters and story characters talk with their lips in sync with the voice. About ten minutes of rendering per minute of video, so a 30-second one takes about five.
• Stock footage: real, royalty-free clips matched to each sentence. A Short is usually ready in one to three minutes.

What it tells you it cannot do

An agent that agrees to everything and then does something else costs you a render. This one says plainly when a request is outside what it makes, and what it will make instead:

• It does not post or schedule videos. You download the video; its page has a title, a description and tags ready to paste.
• It makes one video at a time. Ask for five and it starts with the first, then the next.
• It does not use or clone your own voice. It uses its own natural voices, a woman's or a man's on request.
• Every video has a voice-over. It does not make music-only or ambient-sound videos.

Paste your own script

When you paste a script, or say "read this exactly", the agent keeps your words. The server cuts the script out of your message itself, the video runs as long as your words take to say, and the script stage splits it into captions and scenes without adding a hook, a greeting or a call to action you did not write.

From a product link to a promo

Paste the link of what you sell and the agent reads the page. The promo has a presenter who pitches it and shows your real screens for the steps, never an AI drawing of your product, and it ends with the call to action you asked for, word for word. It never invents a feature, a price or a result the page does not state.

An example conversation

A typical exchange, shortened:

• You: "A 30 second AI video about why octopuses have three hearts, for Shorts."
• Agent: a 30-second vertical AI video, narrated, with captions. The plan card shows 140 credits and your balance.
• You: "Make it a man's voice and turn the music off."
• Agent: the plan updates, the price stays the same, and it waits for you.
• You tap Accept. The page says the video takes about five minutes and adds it to your list.

How to brief it for a better video

• Say who it is for and what they should feel or do at the end.
• Give the hook if you have one; otherwise it writes one for the first line.
• Name the length and the platform in the first message, so there is nothing to ask.
• Write the call to action in full ("comment LINK and I will send it"), and it is kept word for word.
• Write in the language the video should be in, or name another one.

Keep reading