Quick Start Guide

Get ABMeter running in your application in just a few minutes

The Easiest Way: Use Your MCP Client

Connect the ABMeter MCP server to your MCP client and let it guide you through the entire process.

Safe to explore: ABMeter includes a demo space with simulated data where you can experiment freely without affecting production.

Step 1: Setup MCP Connection

Add the ABMeter MCP server to your MCP client:

Server URL: https://mcp.abmeter.ai

View detailed MCP setup instructions →

Step 2: Start Your Experiment Journey (Claude Code examples)

After connecting, you can interact with ABMeter using these prompts:

List Available Prompts

/mcp

Shows all available MCP prompts in your AI assistant

Get Started

/abmeter:welcome (MCP)

Introduction to ABMeter and overview of available features

Design & Launch Experiment

/abmeter:design_and_launch_experiment (MCP)

Step-by-step guide to design and launch a new A/B experiment

Analyze Running Experiment

/abmeter:analyze_running_experiment (MCP)

Determine if an experiment can answer its hypothesis

Conclude Experiment

/abmeter:conclude_experiment (MCP)

Analyze results, recommend a decision, and guide through stopping and concluding

Ask Naturally

"Help me test if a green button converts better than blue"

Your AI assistant will help create parameters, set up variants, and launch your experiment

Your MCP client will guide you through SDK installation, help configure your API key, set up parameters in your code, create and launch experiments, and analyze results when ready.

Safety Built In

These MCP prompts are equipped with all the tools needed to create, run, and analyze experiments and feature flags. For critical operations like starting or stopping experiments, we've implemented human-in-the-loop safeguards. The MCP tools are designed to require your explicit confirmation in the browser for any potentially impactful operation, ensuring you maintain full control over experiment decisions.

Step 3: Explore the Demo Space

The demo space contains realistic experiments at various stages-some ready for analysis, others still collecting data. Use your MCP client to explore these experiments, review their configurations, and practice making decisions based on simulated results. Feel free to modify or delete anything-demo data can be reset and regenerated from scratch at any time. We recommend familiarizing yourself with the platform here before working with production data.

Open Demo Space →

ABMeter SDK Integration (Ruby examples)

The MCP prompts above will guide you through integrating the ABMeter SDK into your application. Here's what your Ruby integration will look like:

Configure ABMeter SDK

ABMeter.configure do |config|
  config.api_key = ENV['ABMETER_API_KEY']
end

Configure the SDK with your API credentials and logging preferences

Get Parameter Values

user = ABMeter::Core::User.new(
  user_id: current_user.id.to_s,
  email: current_user.email
)

button_color = ABMeter.resolve_parameter(
  user: user,
  parameter_slug: 'button-color'
)

Resolve experiment parameters for your users

Track Events

ABMeter.track_event('purchase', user.user_id, {
  plan_slug: plan.slug,
  price: price
})

Track conversion events to measure experiment impact

Creating parameters (like 'button-color') and defining events (like 'purchase') is done via the MCP prompts above-your code simply references them by name.

Next Steps

Setup Browser Link

Connect your MCP client to the browser to enable human-in-the-loop confirmations for critical operations

Browse Experiments

See your active experiments and their performance

Read API Documentation

Learn about advanced features and API endpoints