Skip to content

kingoftech-v01/cutpilot

CutPilot

"The edit you don't notice is the edit that works."

CutPilot is an AI-powered post-production plugin for DaVinci Resolve and Adobe Premiere Pro, built by Rhematek Solutions. It takes raw multi-camera footage - podcasts, YouTube shoots, live events - and turns it into a production-ready long-form edit plus a set of platform-optimized viral social clips, all in under 3 minutes, directly inside your NLE timeline.

No separate app. No third-party upload. No loss of editorial control.

  • Supported NLEs: DaVinci Resolve 18+ · Adobe Premiere Pro 2024+
  • Target processing time: 2–4 min for a 30-min, 3-camera episode (vs. 4–8 h manually)
  • Status: Active development - monorepo with backend, Resolve plugin, Premiere extension and docs

Monorepo layout

cutpilot/
├── cutpilot-backend/     # FastAPI orchestration service (Python)
│   └── app/
│       ├── modules/      # Feature modules (zoom, color, audio cleanup, …)
│       ├── models/       # SQLAlchemy models (projects, sessions, telemetry)
│       └── main.py       # FastAPI entrypoint
├── cutpilot-resolve/     # DaVinci Resolve plugin (OpenFX + Python scripting)
├── cutpilot-premiere/    # Premiere Pro CEP/UXP extension
└── docs/                 # Full product + architecture documentation
    └── README.md         # Master documentation index - start here

Feature modules

Module What it does
Multicam Switching Speaker diarization + LLM-based camera selection with natural cut rules
Smart Zoom & Animation Keyframe-driven push-ins, pull-outs, easing
Silence & Filler Removal Word-level dead-air and "um / uh" removal
Color Grading LUT-based + AI-driven grade, camera matching
Sound Design Context-aware SFX via ElevenLabs SFX v2
Viral Clip Detector Virality scoring + 30/60/90s exports in every aspect ratio
Audio Cleanup ElevenLabs Voice Isolator + RNNoise offline fallback
Transitions Scene-aware cuts, dissolves, whip pans
Script Context Optional script/screenplay ingestion with Scene Intent Map

All modules are validated by a Naturalness Engine - a safety layer that prevents jarring AI edit decisions from reaching the timeline.

Architecture (three layers)

┌─ USER'S LOCAL MACHINE ────────────────────────────────┐
│  Layer 1 - NLE Plugin   (Resolve / Premiere)          │
│              │  WebSocket ws://localhost:8765         │
│  Layer 2 - Orchestration Service (FastAPI, local)     │
│              │  HTTPS                                 │
└──────────────┼────────────────────────────────────────┘
               ▼
       Layer 3 - Cloud APIs (LLM, ElevenLabs, …)

Layers 1 and 2 run entirely on the user's machine, so raw footage never leaves the workstation. Only small, necessary payloads (transcripts, scoring requests) are sent to cloud APIs.

Documentation

Full documentation lives in docs/. Start with docs/README.md - it's the master index linking to:

  • 01 - Product overview: vision, users, competitive analysis
  • 02 - Architecture: three-layer design, data flows, cloud APIs
  • 03 - Features: per-module deep dives (F1 through F9)
  • 04 - Naturalness engine: the edit-safety validation layer
  • 05 - Modular architecture: how modules are isolated and composed
  • 06 - Setup: dev environment, Resolve plugin, Premiere extension
  • 07 - Backend: API reference + database schema
  • 08 - Security: threat model, local data isolation
  • 09 - Deployment: Docker, fly.io, AWS Lambda proxy
  • 10 - Business: pricing tiers, roadmap

Getting started (dev)

git clone https://github.com/kingoftech-v01/cutpilot.git
cd cutpilot/cutpilot-backend
cp .env.example .env          # fill in API keys
pip install -r requirements.txt
uvicorn app.main:app --reload --port 8766

Then build the Resolve plugin or Premiere extension - see docs/06-setup/ for full instructions.

Contributing

Contributions are welcome - particularly new modules, NLE integrations, and model improvements. Please read the architecture docs first, then open an issue before submitting a large PR.

License

See the repository root for licensing. Individual third-party APIs (ElevenLabs, LLM providers, …) have their own terms.

Releases

No releases published

Packages

 
 
 

Contributors