Expanto

Experimenter Analytics Tools — A/B testing platform for analysts and data scientists who need full control over their experimentation workflow.

Design Principles

One person, full control — Own the entire workflow from setup to insights using familiar tools (Python, SQL, YAML) without requiring a separate engineering team.

Flexibility over perfection — Real experiments rarely fit templates. Expanto prioritizes adaptability, letting you modify metrics, logic, and queries as needed.

Features

• Experiment management — Create, track, and close experiments
• Flexible calculations — Define custom metrics and pipelines.
• Statistical analysis — Built-in statistical tests and confidence intervals
• AI assistant — Multi-agent system for experiment design and analysis (it is an experiment for now).
• Web interface — Streamlit-based UI for all operations

Demo

Quick Start

# Clone and setup
git clone https://github.com/dmitrkozlovsk/expanto.git
cd expanto
make setup

# Configure your data warehouse
# Edit .streamlit/secrets.toml with your credentials

# Start application
make start

This will:

• Install dependencies via uv
• Copy configuration templates into .streamlit
• Create example experiment in SQLite database
• Start Streamlit UI and AI assistant

Documentation

Core Documentation

• Configuration Guide — Post-installation setup: data warehouse connections, AI assistant, secrets management
• Architecture Overview — System design, components, data flow, and technology stack

Customization Guides

• Query Templates — Create SQL templates for your data warehouse and experiment calculations
• Metrics Configuration — Define experiment metrics in YAML: averages, ratios, proportions

Additional Resources

• Metrics Examples — Example YAML metric definitions
• Query Examples — Example SQL templates for BigQuery and Snowflake