All features

One button

Auto Edit & AI

Talking-head recordings become finished edits — short-form and long-form — with the pieces landing as ordinary editable clips. AI is bring-your-own-key and never proxied through Streamix.

3 features · 3 shipped

Auto Edit → YouTube — B-roll picks per product with licence badges
Creator+

Auto Reels

Whisper word timings → jump cuts and filler removal → hook-first cold open → subject-tracked 9:16 reframe with alternating punch-ins → beat-synced FX → word-by-word captions → loudness-normalised export.

  • Six styles (Talking Head Pro, Hype / Beat, Clean Minimal, Vintage / VHS, Cinematic, Podcast Clip) and seven karaoke caption looks
  • 9:16 / 4:5 / 1:1 with TikTok / Reels / Shorts safe zones; 15 / 30 / 60 / 90 s or full
  • Everything is one undo step; Regenerate reproduces the same edit from the same seed

Read the doc: docs/AUTO_REELS.md

Auto Reels wizard — styles, caption looks and toggles
Timeline after Auto Reels — cuts, captions and FX as editable clips
Creator+

Auto Edit → YouTube with verified product B-roll

The whole talk tightened top to bottom with chapters, an intro lower third, section cards, phrase-level captions and −14 LUFS — and web B-roll of the product you discuss, laid over the talk exactly while it is mentioned.

  • Openverse / Wikimedia keyless, Pexels / Pixabay with your key, and the manufacturer’s own page (Shopify JSON, JSON-LD, srcset) verified and tiered EXACT
  • Licence badges on every pick, a credits block and robots.txt respected (Google-style wildcards)
  • Three titles + a description + tags; swap any pick from the timeline, search again
  • Polish and Deliverables in the same flow: pick a family in the style step, tick formats in the report, Export all

Read the doc: docs/AUTO_EDIT_YOUTUBE.md

Auto Edit → YouTube — B-roll picks per product with licence badges
Auto Edit → YouTube report — exact-product B-roll, Polish family, Deliverables and description
Creator+

AI providers — bring your own key

Turn one photo of your product into B-roll stills and short cinematic clips, and write chapters / titles / description / tags from the transcript with your own OpenAI and Higgsfield accounts.

  • Keys in safeStorage, real test connection, model pickers, spend cap + usage meter, offline toggle
  • Higgsfield DoP image → video, OpenAI image edits + structured outputs; Sora not integrated (deprecated)
  • Every paid call is gated in the main process, not just the UI

Read the doc: docs/AI_PROVIDERS.md

Settings → AI Providers with OpenAI and Higgsfield keys