Altia Multi-Agent Intelligence

Seven AI agents collaborate using Google ADK (Gemini 3) to research markets, map competitors, execute Python code, and generate visual reports using Nano Banana.

This repository is 100% open source.

Features

• Multi-agent team orchestration
• Roles: Market Research, Competitor Mapping, Data Ingestion, Deep Search, Python Code Execution, Visualization, Coordinator
• Gemini 3 via Google ADK
• Safe Python code execution sandbox
• Nano Banana charting for visual reports
• CLI to run workflows end-to-end

Quick Start

1. Create and configure Google ADK credentials (env vars):

   • GOOGLE_ADK_PROJECT_ID
   • GOOGLE_ADK_LOCATION
   • GOOGLE_ADK_API_KEY or application default credentials

2. Install dependencies:

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

3. Run the demo workflow:

python -m src.cli run --topic "electric scooters" --region "US"

Outputs will be saved in outputs/ with visual report PNG/HTML.

Structure

• src/agents/: agent role implementations
• src/orchestrator.py: team orchestration logic
• src/code_executor.py: sandboxed Python execution
• src/nano_banana_reporter.py: charts and report generator
• src/cli.py: command-line entry
• tests/: tests

Configuration

Set environment variables or a .env file for ADK:

GOOGLE_ADK_PROJECT_ID=your-project
GOOGLE_ADK_LOCATION=us-central1
GOOGLE_ADK_API_KEY=...