Pipeline
1
Upload or import
Upload via multipart (large files) or paste a public URL. Autoposting downloads, normalizes, and extracts duration metadata.
2
AI detection
Audio is transcribed with word-level timestamps. AI scores segments for virality, hook strength, and retention. Face detection builds per-speaker crop keyframes.
3
Review candidates
AI-ranked clip candidates appear with scores for
hook, retention, clarity, and visual. Select a candidate and choose a layout.4
Render
Dispatch a render job. Track progress via
GET /clips/:id. Credits are deducted only on successful completion.5
Publish
Download the rendered clip or create a post draft pre-filled with the video for direct platform publishing.
Import via URL
GET /clips/:id for processing status — the status field moves from processing to ready when candidates are generated.
Layouts
Split layout requires at least 5 qualifying two-face frames spanning more than 3 seconds. If
splitEligible: false, Split options are disabled and render falls back to Focus.Submit a Render
Output Formats
Credit Costs
Status Reference
Posts
Publish rendered clips as scheduled posts across platforms.
Brands
Set default clip templates and aspect ratios per brand.
Carousels
Turn topics into multi-slide carousels instead of video clips.
API Reference
Upload, render, and download endpoint reference.

