🎬 UGC Ad Pipeline

Enterprise AI-powered video ad generation from any product URL — included in the Hermes Agent Enterprise consultancy package.

1Paste your
product URL

2AI researches
winning hooks

3Generates scripts
in 3 formats

4Selects consistent
AI avatar

5Renders video ads
via OpenRouter

6Publish-ready MP4s
in all ratios

🎬 Demo Video

AI-generated UGC ad for Hermes Agent Consultancy — rendered with Veo 3.1 Fast via OpenRouter.
4 scenes: Agent Reveal → Squad Network → Before/After → Brand Close. 32 seconds, 9:16 vertical.
No humans, no embedded text — subtitles can be added separately for each platform.

Quick Start

# 1. Set your OpenRouter API key
export OPENROUTER_API_KEY="sk-or-v1-your-key-here"

# 2. Check cost estimate first
cd skills/ugc-ad-pipeline
python3 pipeline_runner.py \
    --url "https://yourproduct.com" \
    --name "Your Product" \
    --tagline "Your tagline" \
    --features Feature1 Feature2 Feature3 \
    --dry-run

# 3. Generate ads (~$60-85 for a full run)
python3 pipeline_runner.py \
    --url "https://yourproduct.com" \
    --name "Your Product" \
    --tagline "Your tagline" \
    --features Feature1 Feature2 Feature3

# 4. Find your ads in ~/ugc-ads/

Requirements

Python 3.9+
OpenRouter API key — Get one here
pip packages: pip install requests beautifulsoup4
ffmpeg (optional, for avatar frame extraction) — apt install ffmpeg

Video Models

Seedance 2.0

UGC avatar ads

Face consistency across frames

~$0.80per 8s 720p

Wan 2.6

Tutorial / demo ads

Lip-sync, reference-to-video

~$0.96per 8s 720p

Veo 3.1 Fast

Cinematic / wild-card

Best quality, fastest

$4.00per 8s 720p

API Costs

A full ad run (3 formats × 4 scenes × 3 aspect ratios = 36 renders) at 720p costs approximately $60-85 in OpenRouter API credits — paid directly by you, no markup.

Use --dry-run to see exact cost estimates before rendering. Control costs by choosing resolution (720p vs 1080p) and selecting specific formats.

Customization

Change avatar

python3 pipeline_runner.py --url "..." --avatar 2   # Sam (trustworthy, early-40s)
python3 pipeline_runner.py --url "..." --avatar-desc "A trendy Gen Z creator with pink hair"

Available avatars

Alex

Professional, approachable, mid-30s

Jordan

Energetic, relatable, late-20s

Sam

Trustworthy, calm, early-40s

Riley

Trendy, bold, mid-20s

Morgan

Technical, sharp, early-30s

Skip avatars (text-only ads)

python3 pipeline_runner.py --url "..." --skip-avatar

Override models

# Use cheaper models for testing
export UGC_MODEL_UGC="bytedance/seedance-2.0-fast"
export UGC_MODEL_WILDCARD="google/veo-3.1-lite"

# Or change the API endpoint
export OPENROUTER_API_BASE_URL="https://openrouter.ai/api/v1"

Output Structure

~/ugc-ads/
├── scripts.json           # Generated ad scripts
├── pipeline-report.json   # Cost and status report
├── avatars/               # AI avatar reference files
├── scenes/                # Raw rendered scenes
├── 9x16/                  # Vertical (TikTok, Reels, Shorts)
├── 1x1/                   # Square (Instagram, Facebook)
└── 16x9/                  # Landscape (YouTube, website)

Ad Formats

25sec

UGC

Authentic creator-style, first-person, relatable

Seedance 2.0

28sec

Tutorial / Cinematic

Educational, polished, benefit-driven

Wan 2.6 + Veo 3.1

22sec

Wild-Card

Surprising, humorous, attention-grabbing

Veo 3.1 Fast

Client Ownership

✅ You own the code — included in Enterprise package, no subscription
✅ You pay API costs only — OpenRouter billing, ~$60-85 per full run
✅ Modify and extend — full source code, Python stdlib-based
✅ No vendor lock-in — pipeline runs on your infrastructure
⚠️ 20 months of Higgsfield Business ($99/mo) = more than our Enterprise package — and you still don't own anything

← Back to Hermes Agent Deck