Ascended Labs

AI-powered video production platform for health, peptide, and drug explainer content.

Automates ideation, script generation, image generation, and voiceover — then provides a polished web UI for the manual Seedance + CapCut steps.

Architecture

• app/ — FastAPI + HTMX web application (auth, dashboard, project view)
• pipeline/ — Python library + CLI for the automation steps
• projects/ — per-compound project folders with scripts and assets
• config/ — settings and ElevenLabs voice configuration
• data/ — SQLite database

APIs Used

Service	Purpose
Anthropic (Claude)	Ideation + script generation
Google AI	Reference images + key frames (Imagen / Nano Banana Pro 2)
ElevenLabs	Voiceover narration

Setup

# Clone and enter
cd ascendedlabs

# Python environment
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

# API keys
cp .env.example .env
# Edit .env and add: ANTHROPIC_API_KEY, GOOGLE_AI_API_KEY, ELEVENLABS_API_KEY

# Set your ElevenLabs voice ID in config/voices.yaml

Running the Web App

uvicorn app.main:app --reload

Open http://localhost:8000 — register an account (the first user becomes admin).

CLI (standalone)

# Generate ideas
python -m pipeline.produce ideate --compound bpc-157
python -m pipeline.produce ideate --niche peptides

# Generate a script
python -m pipeline.produce script bpc-157 \
  --title "What Happens When You Inject BPC-157" \
  --hook "Your body already makes it — then stops" \
  --angle mechanism-of-action

# Generate all images for a project
python -m pipeline.produce images bpc-157

# Generate all voiceover
python -m pipeline.produce voiceover bpc-157

Workflow

1. Ideate — Claude researches trends and generates video ideas
2. Pick an idea — Starts a new project
3. Generate script — Claude writes a full time-locked production script
4. Generate images — Google AI creates reference images + key frames
5. Generate voiceover — ElevenLabs produces narration
6. Manual: Seedance — Drag images + paste prompts into Seedance 2.0
7. Upload clips — Drop generated clips back into the project page
8. Manual: CapCut — Drag all assets into CapCut, follow caption/sound design notes

Testing

pytest tests/ -v