Prism AI
Two different things are called AI here, and only one of them is ours.
| Who provides it | What you pay | |
|---|---|---|
| The model that drives Prism | You — your ChatGPT, Claude or Gemini plan | Prism takes nothing for it |
| Prism AI | Us | Per use, from a prepaid balance |
That split is why Pro is a flat $6 rather than a share of what you already pay.
The six capabilities
Six, and never a seventh. Each one is metered, has a panel button, and can be called by your AI.
| Capability | You give it | You get back |
|---|---|---|
| Transcribe | An audio layer | Word-level timestamps |
| BPM | An audio layer | Tempo and beat markers |
| SFX | A description | A sound effect on your timeline |
| Music | A description | A music bed |
| Image | A description | A still, placed in the comp |
| Video | A description | A clip, imported and placed |
The balance
| Plan | Pro · Free |
| How it works | Load it when you want it. It never renews on its own |
| Expiry | Never |
| Charged | Only when you ask for something. Nothing generates in the background |
| Minimum top-up | $5 |
Video is the slow one
Video generation takes minutes, not seconds. Do not cancel and retry — a second request is a second charge for the same clip.
How you trigger them
| Ask your AI | "Transcribe this audio layer" — it calls the capability for you |
| Click in the panel | The Prism AI tab has a button per capability |