Getting Started

New to MultAI? This guide walks you through everything — from Chrome setup to your first simultaneous prompt run across all 7 AI platforms.

What MultAI Does

MultAI is a Python/Playwright browser automation engine that submits a single prompt to 7 AI platforms simultaneously — Claude.ai, ChatGPT, Gemini, Grok, DeepSeek, Perplexity, and Microsoft Copilot — and collates all responses into a single markdown archive.

It runs directly inside Claude Code as a skill (/multai), so there's no separate app to manage. In the Claude Code tab it runs all platforms in parallel via asyncio. In the Cowork tab it runs them sequentially using Claude-in-Chrome browser tools.

Prerequisites

Before installing MultAI, make sure you have the following:

  • Python 3.11 or later — check with python3 --version
  • Google Chrome — must be the standard Chrome browser, not Chromium or Brave
  • Claude Code — with the MultAI plugin installed (see next section)
Optional — for agent fallback: Set ANTHROPIC_API_KEY or GOOGLE_API_KEY in your .env file if you want Playwright failures to automatically retry via the browser-use agent. These keys are not needed for standard operation.

Installing MultAI

The fastest way to install is directly from Claude Code:

  1. 1
    Install the plugin — type the following in Claude Code chat:
    Claude Code
    /plugin install alo-exp/multai
  2. 2
    Run the setup script — from the project root:
    Terminal
    bash setup.sh

    This installs Playwright, openpyxl, and browser-use into a .venv, installs Playwright browser binaries, and launches Chrome with CDP on port 9222.

  3. 3
    Verify installation — you should see /multai listed when you type / in Claude Code.
Alternatively, clone the repo manually — git clone https://github.com/alo-exp/multai.git — and run bash setup.sh from the cloned directory.

Setting Up Chrome with CDP

MultAI connects to Chrome using the Chrome DevTools Protocol (CDP) — a remote debugging interface that lets Playwright inject prompts and extract responses from real browser tabs without needing platform API keys.

If you ran bash setup.sh, Chrome was launched automatically. If you need to start it manually:

macOS
/Applications/Google\ Chrome.app/Contents/MacOS/Google\ Chrome \ --remote-debugging-host=127.0.0.1 \ --remote-debugging-port=9222
Linux
google-chrome --remote-debugging-host=127.0.0.1 --remote-debugging-port=9222
Keep Chrome open throughout all MultAI runs. If Chrome closes mid-run, the orchestrator will fail and partially written output files may be incomplete.

Your First Run

With Chrome running and the plugin installed, you're ready to submit your first prompt:

  1. 1
    Invoke the skill in Claude Code chat: type /multai
  2. 2
    Enter your prompt — MultAI will ask for the research question or task to send to all platforms.
  3. 3
    Choose mode and platforms — select REGULAR (fast) or DEEP (thorough), and optionally limit to specific platforms.
  4. 4
    Wait for completion — REGULAR mode takes 2–5 minutes for all 7 platforms. DEEP mode can take 10–20 minutes.
  5. 5
    Read the archive — results are saved to archive/ as a single collated markdown file. Each platform response is wrapped in <untrusted_platform_response> tags.

What's Next

Now that you've run your first prompt, here's where to go next:

  • Core Concepts — understand DEEP vs REGULAR mode, rate limiting, agent fallback, and how CDP automation works under the hood
  • Running MultAI — learn platform selection with --platforms, the Cowork sequential path, and how to consolidate results
  • Reference — complete list of CLI flags, platform identifiers, .env config options, and output file paths
Next Core Concepts →