The pipeline

How it works

Eight tools, one architecture. Understanding these three moves once means you understand every tool.

The idea behind every tool

Heavy work local, intelligence from your own AI, publishing from the cloud.

  1. 1

    Analyze locally

    The app extracts exactly what the AI will need. For a podcast: a word-timed transcript with sentence IDs. For a film: a shot map — every shot with an ID, a description, its dialogue, face count and optional on-screen text. For an article: the text. Models run on your PC: faster-whisper large-v3-turbo for speech, YOLOv8 for faces, a vision captioner for shots, OCR for text. NVIDIA GPUs accelerate everything; CPU works.

  2. 2

    Ask your AI with a Master Prompt

    A Master Prompt is a complete, self-contained instruction: the data with IDs, your settings (durations, style, language, tone), strict output rules and the exact JSON schema. Two ways to run it: automatic — press Generate with Gemini / ChatGPT and the app opens a hidden browser tab of your connected account, pastes the prompt, waits for the answer and reads the JSON; or manual — copy the prompt into any AI (Claude, Mistral, DeepSeek, Grok, an AI IDE with file access) and paste the JSON back.

  3. 3

    Validate, render, publish

    The JSON is validated strictly: sentence IDs map back to real timestamps, shot IDs must exist in the shot map, word budgets and schema are enforced, anything auto-fixed is explained. FFmpeg renders the result with the best encoder on your machine. The finished file is uploaded to your cloud space (24 h) so the publish modal can post now or schedule — the cloud scheduler fires even when your PC is off.

Your AI, connected once

Integrations → AI Chat.

Connect Gemini and/or ChatGPT: a login window opens inside the app; the session is kept in a dedicated browser profile on your PC. From then on every tool can run its Master Prompt in the background — several projects at once, extra ones queue. The same connection draws pictures: post images, article pictures, thumbnails and documentary B-roll are generated by asking the chat for an image and saving the file it returns.

If the provider asks for a human check, the app pauses that job and shows you the window; complete it and the job continues. If the page layout defeats the automation, the error tells you what was seen — copy-paste always works.

  • No API key, no token bill: you use the web chat of your own account.
  • Fallback: if one provider fails, the app can try the other and tells you.
  • Any AI IDE works for the manual path — the Film and Documentary prompts even list the project files the IDE may open (shot map, transcripts, keyframes) to double-check shots.

Rendering on your hardware

FFmpeg 8 with hardware encoders.

Encoders

NVIDIA NVENC (h264_nvenc) when a compatible GPU is present, x264 on CPU. Detected automatically; a banner shows what is in use.

Shorts

1080×1920, face-tracked crop or blurred-canvas layout, word-by-word captions with the spoken word highlighted, hook title pill, cold-open teaser, filler and pause cuts, bleeps.

Film recaps & documentaries

Shots cut to the voiceover beat by beat, copyright shield (zoom / mirror / grade / vignette / short cuts), captions, hook title, music ducked under speech, fade-to-black and echo-to-black rhythm marks, slow motion, B-roll animation.

Thumbnails & covers

AI picture + your text composed locally (Impact or any font, outline, shadow, gradient, highlight colour) in 1280×720 and 1080×1920.

Voiceover

ElevenLabs (cloud key pool, rotated by credits) or Gradium (local vault, word timings) — beat by beat, regenerate any beat, previews before you commit.

Background jobs

Downloads, analyses, renders and uploads keep running when you leave the page. Progress lives on the project card.

Your cloud space

Small, encrypted, 24 hours.

Your Essence Cut account keeps: linked social accounts (OAuth tokens encrypted at rest), your ElevenLabs key pool, cloud copies of finished videos (24 h), scheduled posts and — for Film To Story and Documentary — a project snapshot (script, voiceover, settings, no film file) so you can open the same story on another PC and continue.

Deleting a project deletes its cloud files immediately. Nothing is used for analytics or advertising.

Publishing & scheduling

One modal for every tool.

  • Left: your accounts grouped by platform — tick one or many.
  • Middle: one editor per selected account with only that platform's fields, prefilled by the AI (YouTube title / description / tags / category / privacy, TikTok caption, Facebook & Instagram caption + hashtags + optional first comment).
  • Right: the exact video that will go out, plus its thumbnail or cover when attached (set on YouTube, sent to Facebook, used as the Instagram Reel cover).
  • Bottom: Publish now or pick a time. Scheduled posts fire from the cloud.
  • Facebook first comments can be posted automatically (with a delay) or sent to your Telegram bot for you to post by hand.
  • Buffer: publish through your Buffer channels instead of, or in addition to, direct links.

Where your files are

Settings → Output location.

Every tool writes its projects under its own folder in your projects base: video-to-shorts/, film-story/, video-to-documentary/, facebook-posts/, blog-articles/. Change the base folder any time; existing projects are never moved.

See it on your own videos

Free download for Windows. Your first analysis runs in minutes.

Free · No watermark · No clip limits · Runs on your PC