Made for the phone, not shrunk to fit it
Most video tools are desktop applications with a mobile companion. Captions is the reverse, and it shows in the details: shooting, editing and publishing are one continuous flow on the device that has the footage, with no export, transfer or sync step.
For a creator posting daily, removing that transfer is worth more than any individual feature. The alternative workflow — shoot on phone, transfer to computer, edit, transfer back or upload — is where most people’s posting schedule quietly dies.
Animated captions, done to the platform’s convention
Word-by-word styled captions, timed to speech, in the format short-form social video has settled on.
Doing this by hand is slow, and doing it badly is worse than not doing it — mistimed or unstyled captions read as amateur in a way no caption at all does not. Most social video is watched muted, so this is not decoration.
The same warning applies as everywhere captions are burned in: errors become permanent pixels. Names, brands and jargon are where transcription fails, and those are the words most likely to matter. Check before export.
Eye contact correction, and when to leave it off
Gaze correction adjusts your eyes so you appear to look at the lens while reading a script off the screen.
It addresses a real problem: reading and looking natural are hard to do simultaneously, and a presenter visibly reading loses an audience. For your own video, reading your own script, this is a cosmetic fix comparable to colour correction.
Where it stops being cosmetic is when eye contact reads as sincerity. A testimonial, an apology, a piece to camera about something serious, a fundraising appeal — in those, apparent eye contact is part of the message, and manufacturing it is a small deception a viewer would care about if they knew.
The same judgement applies to the AI presenter mode, which generates a person and a script outright. An audience that believes it is watching a human has been told something untrue, and whether that matters depends entirely on what the video is for.
Neither feature is wrong. Both are worth deciding about deliberately rather than leaving on by default.
Teleprompter and auto-trim
A teleprompter for reading while filming, and automatic removal of pauses, stumbles and retakes.
Together these remove most of the pain of a piece to camera — the two things that make a one-minute video take forty minutes are forgetting your lines and cutting out the four attempts. For a solo creator that is the difference between posting and not.
What you need
- A phone — iOS or Android — with a desktop version available. Mobile is where it is designed to be used.
- An account and, for real use, a subscription. The free tier watermarks output, so it cannot genuinely be evaluated in public posting.
- Yourself, talking. The tool assumes a person on camera.
- An internet connection; processing is remote.
Where it stops
Short-form only in practice. Long videos are outside what the interface handles well, and a multi-speaker conversation confuses features built around a single presenter.
For long-form, Descript is the tool. For cutting clips out of long-form rather than shooting short, OpusClip. Captions is for footage you shot to be short.
Who it suits
Solo creators producing short-form video daily, where the whole workflow needs to happen between other commitments and on the device that shot it.
Small businesses and marketers making social video without a production budget or an editor.
Anyone whose captions are currently done by hand, which is hours a week for regular posters.
It is a poor fit for long-form, for multi-camera or multi-speaker work, and for teams with an existing desktop editing pipeline.
What it gets right for a phone
- Captions that look like the platform’s own, generated in seconds.
- Genuinely mobile — the entire workflow on the device that shot the video.
- Eye contact correction, which nothing else in this batch offers.
- Teleprompter plus auto-trim, removing most of the pain of a piece to camera.
- Dubbing for reaching other language audiences.
Its limitations
- Short-form only in practice.
- Caption errors are permanent once burned in.
- Subscription pricing that is steep relative to how narrow the tool is.
- Watermarks on free output, so it cannot be evaluated in real use without paying.
- Single speaker assumed; conversations confuse it.
What survives cancellation
Exported videos. Projects, caption styles and any presenter setup stay in the app.
Keep your original footage. It is the asset — captions and edits can be redone in any tool from the source, and creators who keep only the exported, captioned version cannot repurpose or recaption it later. Phones fill up and originals get deleted, which is exactly how that happens.
If you outgrow it
- OpusClip — for cutting clips out of long-form rather than shooting short.
- Descript — far more capable, desktop, better for anything over a minute.
- HeyGen — stronger translation and presenter work.
- InVideo AI — for building video without appearing in it.
Compiled from Captions’ documentation and public sources. We have not hands-on tested this tool. Last reviewed 16 August 2026.
