AI agents and workflow tools can read this whole map as JSON — every mode, layer, field, default, price and example — from GET /api/v1/capabilities (free, no key).
The layers
| Layer | What it controls | Fields |
|---|---|---|
| Visuals | What you see | Mode-specific: stock library, your YouTube video, an AI model, or a presenter photo |
| URL inputs | Content you point us at | youtube_url, channel_url, image_url, reference_image_urls, asset_urls, brand.website_url, brand.logo_url |
| Niche | Which footage library | clip_type (e.g. tech, business, finance) |
| Purpose | The angle of the script or clip selection | content_purpose: education, sales, awareness, product_demo, case_study, news, entertainment |
| Script | What is said | prompt / topic (we write it) or script (read word for word); cta, include_hook |
| Voice | Who says it | voice_id, voice_gender, voice_speed / speed, language |
| Captions | On-screen words | include_captions / captions, caption_style (preset, colors, position, size), caption_language |
| Music | Background track | music (enabled, volume) / include_music; native_audio (AI video) |
| Brand | Logo, colors, name and site | brand object (or your account brand) |
| Format | Shape, length, quality | aspect_ratio, duration_target_sec / duration_sec, resolution, quality_tier / model, clip_count |
| Delivery | Where results go | delivery.visibility (private), webhook_url, agent_run_id |
Technical maturity (0–10)
How proven each mode is end to end on the public API — not a quality rating of a single video. 0 not built · 3 available, not yet verified · 5 works end to end · 7 verified with live renders, known gaps listed · 9 proven in production volume · 10 proven at customer scale.
Recommended modes
| Mode | Maturity | URL inputs | Niche | Purpose | Script | Voice | Captions | Music | Brand | Format | Private / webhook |
|---|---|---|---|---|---|---|---|---|---|---|---|
YouTube clipsPOST /clipper/youtube | 8 | youtube_url or channel_url | — | Yes — steers which moments are picked | The speaker's own words | The speaker's voice | Yes (default on) | Original audio | Not yet | aspect_ratio 9:16 / 16:9 / 1:1, clip_count 1–10 | Yes / yes |
Stock videoPOST /generate-and-post | 7 | brand.logo_url, brand.website_url, asset_urls | Yes — clip_type; inferred from brand.industry and your brief, else business | Yes — shapes the script we write | Yes — written from prompt/topic (with optional cta and include_hook), or your script word for word (never cut) | voice_id, voice_gender, voice_speed, language | Yes — caption_style preset/colors/position/size, caption_language | Yes — music.volume; on by default | Yes — per request brand or your account brand | 9:16 only, 15–120 s | Yes / yes |
AI videoPOST /videos/text-to-video, /image-to-video | 6 | image_url (image-to-video), reference_image_urls | — | Yes — writes the narration when you send content_purpose without a script | prompt = visuals; optional script narrated word for word | voice_id / voice_gender (with narration) | Yes when narrated (highlighted active word) | Yes (default on, licensed); native_audio keeps the model's own sound | Yes — brand logo and caption colors | quality_tier low–max or model, resolution, 30 s or 60 s, aspect_ratio | Yes / yes |
AI influencerPOST /videos/ai-influencer | 6 | image_url (any photo, Avatar IV) or avatar_id (a saved persona from POST /avatars); presenter_video_url (your own recording) is not enabled yet | presenter_broll only — broll.clip_type for the B-roll footage, plus broll.footage_keywords / exclude_keywords; broll.visuals: stock (default: local motion graphics of each line's subject + licensed stock footage + headline cards, no AI) or opt-in generated / mixed AI scene stills (third-party image model, Higgsfield; prompts from your script are sent to it) | presenter_broll with topic — content_purpose shapes the script we write (default education) | Your script, word for word (max 120 s), or with presenter_broll a topic: we write a hook-first, one-idea script (no sales pitch), 15–90 s | voice_id, or matched to the presenter (voice_gender), speed; engine: avatar_iii (default for personas), avatar_iv, avatar_v | Yes — default on, from the presenter's speech; presenter_broll: caption_style seam (creator and split default) or karaoke; inspect output | Yes — licensed bed when available | Yes — logo and caption accent; a supplied script is unchanged | format: talking_head or presenter_broll with layout creator (default with 9:16: presenter only on hook, opinion, payoff and close, full-screen footage and graphics elsewhere, no clip over 4 s), split (visuals top, presenter bottom, caption on the seam) or cutaway (full-screen stock B-roll cutaways, broll.coverage 0.2–0.8); best effort, same price; aspect_ratio (9:16 and 16:9 at 1080p; 1:1 at 720p) | Yes / yes |
"Not yet" means the layer is not available for that mode today — sending it is either rejected or has no effect. We add layers to this table only once they are live.
Catalog update: stock footage_keywords / exclude_keywords and influencer captions, music and brand were advertised by the live catalog on October 7. Catalog support does not prove finished-video quality or that every requested layer was applied. See the stock controls and presenter finishing notes. Influencer finishing may fall back to the provider video; completion alone does not prove overlays were applied.
Other modes (not yet verified for partners)
| Mode | Maturity | What it is | Limits today |
|---|---|---|---|
POST /videos/magic | 5 | The same engine as stock video, using your account brand | No per-request brand or private delivery; signed webhooks supported; read the common receipt at data.job_id and data.poll_url. Prefer /generate-and-post. |
POST /videos/motivational | 5 | Classic quote videos (never auto-branded) | camelCase fields; a voiceId is required; no private delivery or webhooks |
POST /blog-to-video | 4 | Narrated video from article text | Uses article_text (the URL is not fetched); long articles narrate fast |
POST /videos/news | 3 | News-style video with broadcast graphics | Being reworked; not recommended for partner use yet |
Other consumer formats are not available on the public API and return 410 MODE_NOT_AVAILABLE. To submit many jobs in one call, use POST /batch.
Choosing settings
- Every option and its live price is listed in
GET /api/v1/tools— let your AI agent read it and pick per job. - Stock video defaults are chosen for B2B short-form ads: 9:16, 30 seconds, captions and music on, your account brand unless you send
brand. Clips default to captions on and 5 clips. - Fields you do not need can be left out; unknown fields on the clipper and AI influencer endpoints are ignored, invalid values return a free
400.