Site index for AI agents: /llms.txt. Documentation pages under /docs are also served as markdown at the same URL plus .md (e.g. /docs/geo-audit.md) or via an Accept: text/markdown request.

Introducing the AirPulse MCP.

Your AirPulse data now works inside your AI assistant. Connect AirPulse to Claude, ChatGPT, Cursor, VS Code or Claude Code, then ask about your AI visibility in your own words.

Release notes ·

Why we built it.

Teams told us the same thing on calls. They want all their data in one place, and they want to ask about it from the assistant they already use. So we built it. The AirPulse MCP is live.

It gives your assistant direct access to your visibility by engine, your competitors, the stored AI answers and citations behind each number, and your GEO audit findings and tasks.

The launch film, 58 seconds.

What an MCP is, and what this one does.

MCP stands for Model Context Protocol. It is the open standard that lets an AI assistant use another product's data and tools. The AirPulse MCP is a server that speaks that standard, so your assistant can read your AirPulse account when you ask it a question.

  1. You ask

    Type the question the way you would ask a colleague. You do not need to know which report holds the answer.

  2. The assistant calls AirPulse

    Your assistant picks the right AirPulse tool, for example the one that returns visibility by engine, and sends your question's brand and date range.

  3. AirPulse returns the dashboard number

    The server answers with the same figure the dashboard shows, with its brand, engines and dates attached. The assistant does not recalculate it.

  4. You follow the evidence

    Ask what is behind a number and the assistant opens the stored AI answer and the pages that the engine cited.

An example question in Claude, 'Which buyer questions stopped naming our brand this week?', and the AirPulse answer: three questions, each with its engine, last week's position and this week's result, plus a link to open the stored answers.
An example in Claude, with example data. The answer states its scope: the brand, the engines and the two weeks it compares.

What you get from it.

Your numbers in seconds

Ask a question in plain language. You do not open a dashboard, set a filter or export a file.

One source of truth

Each answer uses the same data as your dashboard and states its dates, so the number in the chat is the number in the app.

The proof behind each number

Open the stored AI answer and the pages that each engine cited.

Fixes that reach your developers

Hand GEO audit findings to the codebase that must change. Your developer asks which fixes are open and which are done.

Questions to ask on day one

  • Chart our visibility for the last two weeks, engine by engine.
  • Which competitors are named ahead of us on our top buyer questions?
  • Show me the exact ChatGPT answer for this prompt, and the pages it cited.
  • List the open GEO audit findings for our pricing page and turn them into tasks.
  • Which buyer questions stopped naming our brand this week?

Connect in three steps.

You need an AirPulse login and an assistant that supports MCP. There is no token to copy.

A drawing of the AirPulse app, not a live workspace. MCP is in the sidebar under Foundation, below Brand Settings.

  1. Open the MCP page and copy the server URL

    In the AirPulse sidebar, look under Foundation and select MCP. It is below Brand Settings.

  2. Add it to your assistant as a connector

    Copy the server URL and paste it into the assistant that your team uses. The URL is the same for every workspace.

  3. Sign in with your AirPulse login

    Select your assistant's tab for its steps, then sign in with your normal login and select your workspace.

The server URL

Paste this into your assistant. In Claude it takes two screens: add the custom connector, then sign in and select your workspace.

https://prod-api.airpulse.ai/mcp
Two screens in Claude: the Add custom connector form with the AirPulse server URL, then the AirPulse sign-in screen with a workspace selector and an Allow access button.
Claude: add AirPulse as a custom connector, then sign in and allow access.

Steps for each assistant

Claude

  1. Open Settings, then Connectors.
  2. Add a custom connector and paste the server URL.
  3. Click Connect and sign in with AirPulse.

ChatGPT

  1. Open Settings, then Connectors.
  2. Add a connector and paste the server URL. Your ChatGPT plan must support custom connectors.
  3. Sign in with AirPulse.

Claude Code

  1. Run the command below in your terminal.
  2. Run /mcp in Claude Code and choose Authenticate.
claude mcp add --transport http airpulse https://prod-api.airpulse.ai/mcp

Cursor and VS Code

  1. Add the block below to .cursor/mcp.json, or to the mcp.json file in VS Code. No headers are needed.
  2. Sign in with AirPulse when the editor asks.
{
  "mcpServers": {
    "airpulse": {
      "url": "https://prod-api.airpulse.ai/mcp"
    }
  }
}

Other assistants and agents that support remote MCP servers connect with the same URL. For scripts, CI and clients that cannot sign in, create an access token on the MCP page.

What the assistant can read.

The server has 31 tools at launch. Twenty of them read data, and every member can use them. The other eleven take actions, and they stay off until an admin allows them.

Visibility and competitors

Visibility score, share of voice and position by engine, how you compare with each tracked competitor, and the definition of every metric.

  • get_brand_visibility
  • get_competitor_comparison
  • get_brand_health
  • get_metric_definitions

Answers and citations

The stored answer behind every number, the pages each engine cited, and which of those pages are yours.

  • list_prompt_responses
  • get_prompt_response
  • get_citations
  • list_citation_pages

GEO audits and tasks

Audit runs, each finding with its evidence, the tasks the findings became, and the status of an audit in progress.

  • list_audits
  • get_audit_findings
  • get_audit_finding
  • list_audit_tasks
  • get_job_status

Brand context

Your brands, brand profile, tracked prompts and topics, and search across the knowledge base that AirPulse writes from.

  • list_brands
  • get_brand_profile
  • get_brand_topics
  • list_brand_prompts
  • kb_search
  • kb_list_sources
  • kb_read_page

Actions an admin can allow

With actions on, the assistant can do the work as well as report on it.

Start a GEO audit or a brand auditrun_geo_auditrun_brand_audit
Mark an audit finding or task as doneupdate_finding_statusupdate_audit_task_status
Add or change tracked prompts and topicscreate_brand_promptupdate_brand_promptset_brand_topics
Edit the brand and its profileupdate_brandupdate_brand_profile
Manage knowledge base sourcesupdate_content_sourcedelete_content_source

Your data stays under your control.

Access is limited to your workspace. You select the workspace when you sign in.
Members get read-only access. An admin decides if the assistant can take actions.
You sign in with your AirPulse login, so there is no token to store, share or rotate.
Every answer states its brand, engines and dates. A day without a measurement is reported as missing, not as zero.

MCP access is rolling out workspace by workspace. If your assistant does not connect, contact us or reply to the launch email and we switch it on for your workspace.

Questions teams ask first

It is a server that lets an AI assistant read your AirPulse data. MCP stands for Model Context Protocol, the open standard that assistants such as Claude and ChatGPT use to connect to other products. After you connect it, you ask about your AI visibility in plain language and the assistant gets the answer from AirPulse.

More about the product: AirPulse MCP. Every release: release notes.

Ask your assistant about your AI visibility.

Open the MCP page in AirPulse, copy the server URL and sign in. New to AirPulse? Book a free strategy call and we show it on your brand.