Troubleshooting

Do I need to know programming?

No. The AI agent does all of it. You just say “what you want to make” and “check that it's right”.

Does it cost anything?

Anything that stays on your machine (proxy generation, transcription, editing, exporting) is free. You're only charged when you use an outside API, such as cloud transcription or generated narration, and you are always asked to set it up and approve it before it's used. Billing never starts silently. Lab's assets come in two tiers, “Free” and “Pro” (see “Lab” below).

Could the AI overwrite my footage?

No. It is a design promise that the AI never writes to the footage you shot (assets/ and inbox/) (→ File map). Every edit is recorded in a separate file (edit.json), and your original footage stays untouched.

If I stop partway, do I lose my work?

No. All progress is kept in the project's files, so you can “pick up where you left off” the next time you open it. You can resume from a different entry point, in the app or the terminal (from the terminal, use akari --continue).

The words on screen differ from the guide

This guide uses the new names. The v1.5.0 app may still show the old ones (the new names arrive in the next release). Motion parts = “Overlay”, My styles = “My presets”, Caption style = “Caption template”, Channel settings = “Brand kit”. In the CLI, akari new --scaffold = akari new --template, and akari template export = akari playbook template. Either name works.

This explanation doesn't match what the app shows…

The guide is probably out of date. Compare each page's “Last updated” with the version of your app. The latest source of truth is always the documentation on GitHub. Reports of mismatches are welcome.

Windows SmartScreen shows a warning when I run the installer

The Windows installer isn't code-signed yet, so SmartScreen may warn you on the first launch. If you got it from the official download page, go ahead with “More info” → “Run anyway”. The macOS build is signed with a Developer ID and notarized by Apple.

How do I switch between stable and beta versions?

Settings › About this app › Release channel. The default is Stable only; choose Include prereleases and beta versions with new features arrive early. To stay on stable, switch back to “Stable only”.

How do I update?

Updates are automatic by default. It checks in the background without holding up startup and applies on the next launch. To check by hand, run akari update. Updates never break your projects or data.

Can I change the app's display language?

Yes. In Settings › Display › Language › Display language, choose Auto (follow the OS), Japanese or English. The window reloads when you switch. The language of a video's contents (captions and transcription) is kept per project, separately from the screen language.

I can't find AKARI Chat

AKARI Chat (preview) is hidden by default. Turn on “Show AKARI Chat (preview)” in Settings › Partner. For everyday work, connect a partner such as Claude Code or Codex.

The “Templates” shelf in the Library is empty

There are no bundled templates. Until you add one, it says “No templates yet.” Export a bundle (group) from one of your own videos and it lines up on the shelf (in v1.5.0 that's akari playbook template; from the next release, akari template export).

What's the difference between a playbook and a template?

A playbook is something you read. It holds one video's steps, checkpoints, settings and placeholders, but no footage. Start one from “+ Start a new video” → “Start from a playbook” on the home screen, or with akari new <dir> --from <playbook>. A template is something you place on the timeline. It's a bundle with its length, text, frames and audio decided, and you use it just by swapping footage into the placeholders.

What's the difference between Free and Pro on Lab?

Lab has two tiers, Free and Pro. Free assets are CC0 and work once you link your account. Pro assets unlock according to your purchase or rights, and come with AKARI's asset license. Link your account with akari store connect.

Which engine does transcription use?

The default is automatic. On macOS it uses the local SpeechAnalyzer first, and where that isn't available (such as Windows) or for languages it doesn't support, it uses whisper.cpp (local and free). If the model is missing, install it with akari tools install whisper (or Settings › Tools › Whisper). Cloud transcription is used only when you've set up the connection and chosen it yourself, and it shows the destination and a cost estimate first.

Which engine does export use?

--engine auto (the default) uses the fast GPU export if your footage and effects qualify, and otherwise OSR (which renders the whole screen). When it switches to OSR, the export screen shows the reason. --engine gpu stops with a reason if the conditions aren't met. --engine osr works for any project.

The preview or saving feels slow

In v1.4.1 we fixed a slowdown where the home screen and project list re-read every project on each save. First make sure you're on the latest stable version. If a project is large, run akari clean in the project folder and it lists the intermediate files that are safe to delete, with sizes.

It stopped working. What do I do first?

① Read the guidance shown on screen or in the terminal (it often spells out the next step). ② Run akari doctor to diagnose the required parts and connections. ③ If that doesn't fix it, ask in one of the places below. A screenshot of the error makes it much quicker to solve.