Six AI assistants can now make images and video in the studio

You can now connect six AI assistants to the studio: Claude, Claude Code, ChatGPT, Codex, Cursor and Visual Studio Code. Ask for an image or a clip in the chat you already work in, and the assistant picks a model, starts the run and puts the result in your library.

Every assistant on the list was connected with a real account and made a real image before it went on the site. An assistant we have not tested does not get a page.

The connection gives the assistant the same tools you have in the studio, within the permissions you approve when you connect:

• Make images and video with any model in the catalogue, and see the price before it runs.
• Work from your own uploads: a product photo, a portrait, a clip you filmed.
• Build a character and use it again in every image and video.
• Run the flows and templates you saved in the studio.

Every run costs credits from your own balance, and the assistant sees the price before it starts. You can set a daily limit per connection under Connected apps, and revoke a connection at any time.

That matters for an assistant that works on its own. A limit you set once is easier than checking every request.

An assistant does what you ask, including asking twice. Read the price it reports before you confirm a long video.

Each assistant has its own connect page with the steps for that app, from a one-click link in Cursor and VS Code to a command in Claude Code and Codex. The connection uses OAuth, so you sign in to the studio and approve what the assistant may do. No key to copy, no password shared.