⚡ Founder Intelligence Engine

A 6-agent structured AI reasoning system that validates startup ideas from concept to investor pitch — powered by Claude.

---

🏗️ Architecture

founder-intelligence-engine/
│
├── app.py                      ← Flask entry point + route orchestration
│
├── agents/
│   └── agents.py               ← All 6 reasoning agents
│
├── scoring/
│   └── scoring_engine.py       ← Weighted viability score formula
│
├── utils/
│   ├── claude_client.py        ← Claude API wrapper (with retry + JSON parsing)
│   └── prompt_templates.py     ← Structured prompts for each agent
│
├── templates/
│   ├── index.html              ← Founder input form
│   └── result.html             ← Full analysis results
│
├── Dockerfile                  ← Docker support
├── requirements.txt
└── README.md

---

🤖 The 6 AI Agents

#	Agent	Role
1	Idea Generation Agent	Creates a tailored startup idea from founder profile
2	Market Validation Agent	Analyzes market size, segments, demand signals
3	Risk Analysis Agent	Identifies risks by category with mitigations
4	Monetization Strategist Agent	Designs pricing tiers, revenue streams, unit economics
5	MVP Architect Agent	Plans tech stack, features, build phases & costs
6	Investor Attack Agent	Simulates VC scrutiny: killer questions, red/green flags

---

📊 Viability Score Formula

Viability Score = (
    market_score        * 0.30  +
    (100 - risk_score)  * 0.25  +
    monetization_score  * 0.25  +
    mvp_score           * 0.10  +
    investor_score      * 0.10
)

Range	Grade	Label
80-100	A	Exceptional — Pursue aggressively
60-79	B	Strong — Validate with customers ASAP
40-59	C	Moderate — Needs significant refinement
20-39	D	Weak — Pivot or rethink core assumptions
0-19	F	Critical — Major issues, reconsider entirely

---

🚀 Local Setup

1. Clone / unzip the project

cd founder-intelligence-engine

2. Create a virtual environment (recommended)

python -m venv venv
source venv/bin/activate        # macOS/Linux
venv\Scripts\activate           # Windows

3. Install dependencies

pip install -r requirements.txt

4. Set your Claude API key

macOS/Linux:

export ANTHROPIC_API_KEY="sk-ant-..."

Windows (PowerShell):

$env:ANTHROPIC_API_KEY="sk-ant-..."

Or create a .env file:

ANTHROPIC_API_KEY=sk-ant-...

5. Run Flask

python app.py

6. Open in browser

http://localhost:5000

---

🐳 Docker

docker build -t founder-engine .
docker run -p 5000:5000 -e ANTHROPIC_API_KEY=sk-ant-... founder-engine

---

🔌 JSON API

The app also exposes a JSON API at POST /api/analyze:

curl -X POST http://localhost:5000/api/analyze \
  -H "Content-Type: application/json" \
  -d '{
    "skills": "Python, ML, backend development",
    "interests": "productivity, B2B SaaS",
    "budget": "$5,000 - $25,000",
    "time_commitment": "40 hours/week (full-time)",
    "risk_tolerance": "medium — comfortable with calculated risks"
  }'

---

📝 Notes

• Each analysis runs 6 sequential Claude API calls — expect ~30-60 seconds total
• All agents enforce strict JSON output with automatic retry on parse failure
• The model used is claude-sonnet-4-20250514