AI generation workspace

Create generation studio

Generate images

Generate start + end frames

Start/end frames

Virtual try-on

Virtual try-on

Image inspiration library

Upload reference images/videos/text for image prompts.

Generate video

Hook → Meat → CTA builder

Build videos as assembled ads: hook first, meat that fulfills the hook, then a clear call to action.

Video generation

Extend an existing video

Use the /api/generate/extend-video endpoint with a generated video, GCS URI, or uploaded clip.

Extend video

Video inspiration library

Upload images, videos, and text docs. Text files can be inserted into prompts with {{file name}} tags.

Splice clips and transitions

Create an edit decision list for FFmpeg: liveavatar segments, transition videos, generated clips and uploaded videos. Rendering is scaffolded and checks for FFmpeg in the container.

Music inspiration library

Upload reference images, videos, lyrics, briefs, or text docs. Text files can be inserted into Lyria prompts with {{file name}}.

Artist profiles

Create fictional artist profiles. Style can be inspired by real artists, but the system should generate original songs, vocals, names, art, and branding.

Lyria song generation

Generate standalone songs or use an optional artist profile/style. Fast uses one prompt, Mid creates song-section rows, Advanced creates timestamp rows.

Structured prompt builder

Use rows for timestamps or song sections from the Lyria notebook: Intro, Verse, Chorus, Bridge, Outro, BPM, lyrics, instrumental-only.

OrderTime / SectionTypeDescription editable by userAI

Mix builder

Combine songs with uploaded/generated videos and images. Visuals can change at each song start/end.

Channel management

Link artist channels, language channels, and themed mix channels. Metrics are pulled once YouTube OAuth is connected per channel.

Thumbnail/title testing

Create variants from uploaded or generated thumbnails/titles. Automation is scaffolded to rotate variants and compare views over time through YouTube Analytics.

LiveAvatar livestream manager

Plan a HeyGen LiveAvatar stream, RTMP destination, chat source, background, and show rundown for TikTok, Twitch, YouTube, Facebook, or a custom RTMP server.

Stream preview

Embed the public playback/control-room URL for YouTube, Twitch, TikTok, Facebook, or any browser-playable stream preview. RTMP ingest URLs cannot be previewed directly in a browser.

Add a playback URL in setup to preview the stream here.

Backgrounds

Create reusable 16:9 or 9:16 backgrounds for LiveAvatars and generated videos.

Upload source assets

Upload images/videos to use as AI references, avatar backgrounds, transition videos, or source clips.

Schedule a product livestream

Build a schedule from product/model assets, for example a wig image plus model images for try-on. Sections can be live avatar chat, transitions, or prerecorded videos. Use library asset IDs/paths/URLs, one per line.

Add rundown cue

Chat → live avatar response

Paste or receive chat text; Gemini drafts a concise spoken reply. If HeyGen session ID is set and “send to avatar” is enabled, it sends a talk task to the LiveAvatar.

pi-fleet launch

Launch content plan → Create prompts

Pull Launch's content plan, product dashboard, audience stages, blogs, videos, and landing-page plans into Create. Select any item to fill image, video, music, or LiveAvatar prompts.

Library

/ Loading folders…