War Room Multi-Agent System

📋 Overview

A production-ready multi-agent system that simulates a cross-functional "war room" during a product launch. The system analyzes metrics and user feedback to produce a structured launch decision: Proceed / Pause / Roll Back. Built with an Orchestrator pattern and exposed via both a Command Line Interface (CLI) and a FastAPI REST API.

🏗️ Architecture

Input Data → Data Analyst → Marketing Agent → PM Agent → Risk Agent → Final Decision

Agents

Agent	Responsibility
Data Analyst	Analyzes metrics, detects anomalies, identifies trends
Marketing Agent	Assesses sentiment, extracts top user issues
PM Agent	Defines success criteria, evaluates user impact
Risk Agent	Challenges assumptions, identifies blind spots
Orchestrator	Coordinates workflow, produces final decision

Tools

Tool	Purpose
detect_anomalies()	Statistical anomaly detection in time-series metrics
compare_trends()	Multi-metric trend comparison
analyze_sentiment()	Sentiment distribution analysis
extract_top_issues()	Keyword-based issue extraction from feedback

🚀 Quick Start

1. Clone Repository

git clone https://github.com/muhammedafsalpm/ai-launch-decision-system.git
cd ai-launch-decision-system

2. Setup Environment

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

3. Configure LLM Provider

Option A: OpenAI (Cloud)

cp .env.example .env
# Edit .env and set:
# LLM_PROVIDER=openai
# OPENAI_API_KEY=your-key-here

Option B: Ollama (Local)

# Install Ollama first: https://ollama.ai
ollama pull llama3.2

cp .env.example .env
# Edit .env and set:
# LLM_PROVIDER=ollama
# OLLAMA_MODEL=llama3.2

4. Run the System

The system supports two execution environments: CLI and REST API.

Mode A: CLI Execution Run the orchestration flow on local files directly from the terminal.

python main.py

Mode B: REST API Execution Launch the FastAPI server to expose endpoints for the war room multi-agent simulation.

python api.py

Interactive API Swagger documentation is available at http://127.0.0.1:8000/docs

5. View Output

For CLI: Results are saved locally in the output/ directory as JSON files:

cat output/AI_ML_Engineer_[YourName]_April2026.json

For API: You can interact with the API endpoints dynamically. For example, upload existing mock datasets via curl:

curl -X POST "http://127.0.0.1:8000/analyze/upload" \
  -H "accept: application/json" \
  -H "Content-Type: multipart/form-data" \
  -F "metrics_file=@data/metrics.json" \
  -F "feedback_file=@data/feedback.json" \
  -F "release_notes_file=@data/release_notes.txt"

📊 Input Data Structure

data/metrics.json

14 days of time-series metrics including:

• DAU (Daily Active Users)
• Error Rate
• Latency (p95)
• Adoption Rate
• Crash Rate
• Support Tickets

data/feedback.json

25-30 user feedback entries with sentiment classification.

data/release_notes.txt

Feature description and known issues.

🌐 REST API Endpoints

• GET /: Retrieve base API information and available routes.
• GET /health: Health check endpoint reflecting current LLM provider and API status.
• POST /analyze: Submit structured evaluation targets (metrics, feedback lists) directly via JSON body payload.
• POST /analyze/upload: Pass metrics, feedback, and release notes via multipart form file uploads to trigger agent evaluations.

📤 Output Format

{
  "decision": "PAUSE",
  "rationale": "Key drivers with metric references...",
  "risk_register": [
    {"risk": "...", "mitigation": "..."}
  ],
  "action_plan": [
    {"action": "...", "owner": "...", "timeline": "..."}
  ],
  "communication_plan": {
    "internal": "...",
    "external": "..."
  },
  "confidence_score": 0.85,
  "confidence_increase_condition": "..."
}

🔧 Environment Variables

Variable	Description	Required For
LLM_PROVIDER	openai or ollama	Always
OPENAI_API_KEY	OpenAI API key	OpenAI
OPENAI_MODEL	Model name (default: gpt-4o-mini)	OpenAI
OLLAMA_BASE_URL	Ollama server URL	Ollama
OLLAMA_MODEL	Ollama model name	Ollama

📝 Traceability

All agent actions and tool calls are logged to console with [TRACE] prefix.