An AI-powered business idea analysis tool that provides comprehensive feedback through Herman Poppleberry, your friendly AI business plan review assistant.
- Intelligent Analysis: Uses Gemini Pro 2.5 to analyze business ideas with venture capitalist-level insights
- Audio Feedback: Generates personalized audio feedback using Gemini TTS with Herman Poppleberry's voice
- Email Delivery: Automatically sends analysis and audio feedback via email using Resend API
- Multiple Voice Styles: Choose from upbeat, serious, or flippant delivery styles
The agent processed a real business idea about QR code-enabled baby onesies for Home Assistant integration.
Organized Output Folder: agent/feedback/20250828_170126_baby-qr-onesies/
The agent automatically:
- Generated AI-suggested filename:
baby-qr-onesies - Created timestamped folder with descriptive name
- Produced comprehensive venture capitalist-style analysis
- Generated multiple formats for different use cases
# Install dependencies
pip install -r requirements.txt
# Copy and configure environment variables
cp .env.example .env
# Edit .env with your API keysAdd your API keys to .env:
GEMINI_API_KEY="your_gemini_api_key_here"
RESEND_API_KEY="your_resend_api_key_here" # Optional for email delivery
USER_EMAIL="[email protected]"
# Auto-process all pending ideas
python business_idea_agent.py
# Choose voice style and skip email
python business_idea_agent.py --voice-style serious --no-email# Processes all .md files in agent/user-ideas/pending/
python business_idea_agent.pypython business_idea_agent.py --voice-style seriouspython business_idea_agent.py --no-email- upbeat (default): Encouraging and enthusiastic tone
- serious: Authoritative, judge-like delivery
- flippant: Sardonic, defeatist tone
The agent generates:
- Text Analysis: Saved to
agent-feedback/analysis_TIMESTAMP.txt - Audio Feedback: Saved to
agent-feedback/business_analysis_TIMESTAMP.wav - Email: Sent with both text summary and audio attachment (if configured)
- Agent Review: Gemini Pro 2.5 analyzes your business idea using the carefully crafted system prompt
- TTS Generation: Gemini 2.5 Flash Preview TTS creates audio feedback with Herman Poppleberry's voice
- Email Delivery: Resend API sends results with audio attachment and summary
- Text Analysis: Gemini 2.0 Flash Exp
- TTS: Gemini 2.5 Flash Preview TTS (Voice: Charon)
├── business_idea_agent.py # Main agent implementation
├── design/
│ ├── system-prompt.md # Herman Poppleberry's instructions
│ └── gemini-audio-instruction.md # TTS voice guidelines
├── agent/
│ ├── feedback/ # Generated outputs
│ └── user-ideas/
│ ├── drafting/ # Ideas being developed
│ ├── pending/ # Ideas awaiting analysis
│ └── evaluated/ # Processed ideas
└── requirements.txt # Python dependencies
- Draft: Create business ideas in
agent/user-ideas/drafting/ - Queue: Move ready ideas to
agent/user-ideas/pending/ - Process: Run
python business_idea_agent.pyto auto-process all pending ideas - Review: Check results in
agent/feedback/and moved files inagent/user-ideas/evaluated/


