Connect an AI assistant (MCP)

MCP is the standard way AI assistants connect to outside data. Once connected, you can ask in plain words, for example "Which of my products did ChatGPT leave out this month, and what should I fix first?", and the assistant reads the answer from your workspace.

You need an API key first.

Claude Code

Paste the command shown under your new key in Settings, Integrations. It looks like this:

claude mcp add --transport http aeosome https://api.aeosome.com/mcp \
  --header "Authorization: Bearer aeo_your_key"

Cursor and other clients that take a URL

Add a server with this address and header:

{
  "mcpServers": {
    "aeosome": {
      "url": "https://api.aeosome.com/mcp",
      "headers": { "Authorization": "Bearer aeo_your_key" }
    }
  }
}

Claude Desktop and clients that start a program

Sign in once with the command line, then add:

{
  "mcpServers": {
    "aeosome": { "command": "npx", "args": ["-y", "@aeosome/cli", "mcp"] }
  }
}

What the assistant can read

The same data as the dashboard, as tools the assistant picks on its own: a weekly brief, the overview numbers, share of voice, products, the questions shoppers ask and each AI answer, citation sources and opportunities, competitors, fixes, progress, alerts and plan usage.

Every number comes with its unit and period, so the assistant reports how many answers named a product and over which days, never a bare score.