Say the video. Get the video.
One brief in — a finished MP4 out. OnePress writes the script, produces the footage, voices the narration, and cuts it together. Not a timeline you still have to edit.
Made end-to-end by the agent: generated footage, edit, and music — from one sentence. Watch on YouTube ↗
Three ways to a video
Generated footage
Describe a shot or a mood — "Shanghai at dusk, Hong Kong film style" — and the agent produces real video clips, then edits them into a cut with music. B-roll, product shots, scene footage, intros.
Your slides, narrated
A deck becomes a narrated MP4 — the video that presents itself, for people who were not in the room. Full guide →
A digital presenter
A licensed on-screen presenter reads your script with synced lip movement — founder updates, course lessons, weekly shows. The same host across every episode.
They compose: a typical finished video mixes generated footage, a presenter segment, and a narrated voiceover — the agent picks what the brief needs.
How it works
- 1
Describe the video
What it's about, who it's for, roughly how long, landscape or vertical. That's the whole brief.
- 2
Script and plan first
The agent writes the narration and plans the shots before generating a single frame — you can redirect at this stage for free.
- 3
Production
Footage generates, narration is voiced, subtitles and music are laid down, and everything is cut together — each segment saved as its own file you can inspect.
- 4
Review, then re-cut by asking
"Make it vertical for Instagram." "Swap the middle clip." Iteration is a chat message — a vertical re-cut of the same footage is one request away.
A show that runs itself
Recurring video is where this compounds. One brief becomes a weekly series — the workspace keeps every episode, so episode 8 follows the format of episodes 1–7:
"Every Monday morning, a 3-minute video on last week in AI — same host, same format, unlisted to my channel for review."
That's how AI Weekly is produced — see Automations for schedules.
Post straight to YouTube
Connect your channel once — Settings → Channels → YouTube — and the agent uploads the finished MP4 as unlisted: watch it in YouTube Studio, then flip it public when you're happy. Nothing goes live without your review.
Frequently asked questions
Can it clone my face or voice?
No — deliberately. Face cloning, voice cloning, and lip-syncing real footage are not offered: they're the machinery of deepfakes, and our payment provider's acceptable-use policy prohibits that class of tooling. Presenters come from a licensed avatar catalog; your own voice still works — record the narration and the agent lays your real audio over the visuals.
How good is generated footage?
Clips run 4–8 seconds at 720p and read best as b-roll and scene footage — this is a production tool, not a film crew. The agent QA-checks every clip (frames, audio levels, duration) before it goes into your cut, and flags flaws instead of shipping them silently.
Landscape or vertical?
Both, from the same source material — 16:9 for YouTube and embeds, 9:16 for phone-first feeds. Ask for the second format after the first is done; it reuses the edit instead of regenerating.
What does it cost?
Generated footage is one of the more credit-hungry outputs — video models bill per second. Presenter and narrated-deck videos are far cheaper per minute. Free daily credits are enough to try; Pro ($9/month) covers regular use.
One sentence in, one video out
Describe the video you want — the agent produces it while you watch.
Open Video