Media

Create images, video and audio

Everything is generated from inside a thread. Whatever is produced collects in the media pool, ready to be used as material in a composition. This page covers how to ask, and how to find things afterwards.

These are the services this page relies on. The full breakdown is in which feature uses which model.

Making an image OpenAI gpt-image

  1. Set the thread's mode to Image.

  2. Describe what you want — subject, background, mood, aspect ratio (e.g. “product shot against a blue sky, 16:9”).

  3. If you have a reference photo, attach it in the composer before sending.

  4. The result appears in the conversation and lands in the media pool.

This one is billed

If you're using OpenAI's image API, generation is billed separately from any ChatGPT subscription (the app notes “from a few cents per image”). Setup is under OpenAI image generation.

Making a video Higgsfield

Video is generated through Higgsfield. Confirm the connection first (setup).

  1. Set the mode to Video and leave AI engine on Claude.

  2. Describe the video (e.g. “a 30-second vertical shot walking under cherry blossoms”). Length, orientation, camera movement and mood all help.

  3. An estimated credit cost is shown before it runs. Confirm to proceed.

  4. When it finishes, the video appears in the conversation and in the media pool.

The engine is not optional here

Video generation is wired through Claude. Leaving the engine on GPT or Gemini in video mode means nothing generates.

Keeping spend under control

Under SettingsUsage you can set caps and warnings:

  • Daily generation limit (0 means unlimited)
  • Monthly generation limit (0 means unlimited)
  • A low-credit warning threshold

The same screen shows today's and this month's generation counts, plus usage per engine.

Narration and music ElevenLabs

Audio requires an ElevenLabs connection (setup). You can create it from a thread in Audio mode or from inside the editor.

What you're makingWhat to typeWhere
NarrationThe exact script you want read aloudAudio mode, or 🗣 Generate AI narration in a composition
MusicA description of the track (e.g. bright and upbeat, piano-led)Audio mode, or the music lane in a composition

Generated audio joins the media pool and can be placed on the narration or music lane. See The editor and exporting for placement.

Recording your own voice

Narration can also be recorded. A recording is placed on the narration lane and confirmed with Recording added to the narration lane. The first attempt asks for microphone permission.

Finding material in the media pool

Generated and uploaded material collects in the media pool. While editing a composition, you click items here to add them as scenes.

FilterOptions
KindVideo / image / audio
OriginGenerated / uploaded
FolderMaterial sent from your phone is grouped automatically

If you added or removed files directly in Finder, press Refresh to pick up the changes (missing material is tidied up at the same time). Open folder in Finder reveals the originals.

The file manager

For working with everything at once, open File manager.

  • Upload — bring in files from your Mac
  • Filter by Kind (video / image / audio / documents) and Origin (generated / uploaded / audio stock)
  • Search with Search by name or thread
  • Sort by Date added, Name or Kind
  • Switch between List view and Thumbnail view

Deleting a selection moves it to the trash rather than erasing it.

A thread's own files

The right-hand column of a thread (📎) lists the files made in — or handed to — that thread. Where the file manager covers everything, this list is what the conversation you're in is carrying.

What you seeHow to read it
Type iconPDF, sheet, document, image, audio. Its colour is the origin — purple for AI-made, green for files you added
File nameLong names are shortened, but the extension always survives. Hover for the full name
Day headingToday and Yesterday group the rows, so the newest work is obvious
TimeTells apart files made on the same day

Filter by All / AI / Mine. Sort by date, name, type or origin, and the button beside it flips between ascending and descending (day headings appear only while sorting by date).

Hovering a row reveals Copy path and Reveal in Finder on the right. Double-click opens the file.

Where files are saved

Generated files land inside your storage folder (by default ~/Documents/WorkPilot), sorted by type.

FolderContents
generated/Generated images and video
narration/Narration audio
bgm/Generated music
recordings/Recorded audio
作品/Exported video and Premiere exports
received/Material received from your phone

The full picture is in Settings and where data lives.

Where to go next