Run BrandViz.AI from your own AI agent. Approve the fix where it lands.
BrandViz.AI runs a native Model Context Protocol (MCP) server, so an AI agent such as Claude, Claude Code, Cursor or ChatGPT can read your visibility gaps, audit findings and recommended actions straight from your own reports. The same agent can queue work and start Engine jobs that land as pull requests or CMS drafts in your own repository or CMS. Nothing is built in a BrandViz.AI-hosted editor, and nothing publishes without your approval.
7 days free. Cancel anytime. Nothing publishes without you.
- Cannot publish to your site
- No API key to copy
- Reading your data works on every plan
How it works
Agent accessConnect your agent
Add the server URL, then sign in with your BrandViz.AI account.
The agent reads
Visibility gaps, audit findings and recommended actions from your own reports.
The Engine builds the fix
Tasks and jobs start from the agent, and the output is built against your site.
Your team approves
A pull request on a brandviz-agent/ branch, or an unpublished CMS draft.
What does the BrandViz.AI MCP server expose?
Tools only, in three groups: read, setting and work. There are no MCP resources or prompts. The server sits at https://mcp.brandviz.ai/mcp and signs your agent in as you with OAuth.
- Server URL
- https://mcp.brandviz.ai/mcp, a remote MCP server over HTTP.
- What it exposes
- Tools only, in three groups. There are no MCP resources or prompts.
- Read
- Your visibility scores, citations, audit findings, AI crawler activity, tasks and recommended actions, plus what a job changed in your repository or CMS. An agent can also put a question to the live assistants.
- Change what is measured
- Your questions, markets, aliases, brand profile and writing rules. These change what BrandViz.AI measures and how it writes, and none of them touches your site.
- Queue and start work
- Content and technical tasks on your Tasks board, where approving a task starts its job, and Engine jobs started directly. Every job that builds something ends as a pull request or an unpublished CMS draft, and a planning job only proposes.
- Authentication
- OAuth sign-in with your BrandViz.AI account. Your assistant signs in as you, so there is no API key to copy, and it can reach only the brands your account can open.
- Supported clients
- Claude (web and desktop), Claude Code, Cursor and ChatGPT, plus any other MCP client that supports remote servers with OAuth sign-in.
- Plans
- Reading your visibility data works on every plan. Tasks, audits and jobs need the Engine, which the Starter and Growth plans include.
- Approval flow
- Jobs that build something through the server deliver a pull request on a brandviz-agent/ branch through the GitHub integration, or an unpublished draft in your connected CMS. A person on your team merges or publishes it. Jobs count toward your monthly content allowance.
Last verified: November 2026, against the server's tool registry and the setup docs.
How do agents use it?
An agent reads the gap, files or starts the fix, and reads back what changed. Your team approves the result where it lands.
- 01
Query the gaps
The agent finds your brands, then reads your scores, citations and latest audit to see where AI models skip you and which fixes are still open.
- 02
Generate the fix
It files a task on your Tasks board or starts a job directly. A planning job proposes pieces from your gaps and writes nothing to your site, while content and technical jobs build the page or the fix.
- 03
Open a pull request or CMS draft for approval
The Engine delivers the result as a pull request on a brandviz-agent/ branch or an unpublished CMS draft. The agent can read back what changed, and your team approves it where it lands.
Can an AI agent change my live site?
No. The BrandViz.AI MCP server cannot publish. Jobs started through it deliver a pull request on a brandviz-agent/ branch or an unpublished CMS draft, and a person on your team merges or publishes it. An agent can change what BrandViz.AI measures and approve a pending task, but neither touches your live site.
What BrandViz.AI can do
- Read your visibility, citations, audit findings, AI crawler activity and recommended actions for the brands your account can open.
- Change what BrandViz.AI measures and how it writes: brand profile, questions, regions, aliases, lessons, blog style and image templates.
- Queue content tasks on your Tasks board, and approve a pending task, which starts its job and uses monthly allowance.
- Start an Engine job directly, delivered as a pull request or an unpublished CMS draft.
What BrandViz.AI can never do
- Publish. A person on your team merges the pull request or publishes the draft, where it lands.
- Change your live site through the setting tools. None of them touches your site.
- Reach a brand your BrandViz.AI account cannot open.
- Run tasks, audits or jobs on a plan without the Engine. Reading your visibility data works on every plan.
How do you connect an agent?
Three steps, and the same server URL for every app. One connection reaches every brand your BrandViz.AI account can open.
- 01
Add the server URL
In Claude, Claude Code, Cursor or ChatGPT, add a remote MCP server at https://mcp.brandviz.ai/mcp. The setup docs have the exact steps for each app.
- 02
Sign in as you
The first time it is used, the app sends you to BrandViz.AI to sign in and approve the connection. There is no API key to copy, and the app can reach only the brands your account can open.
- 03
Start with an open question
Ask how visible your brand is in AI answers this month. The agent starts by checking which brands you have and what data each one holds, then drills down from there.
To disconnect, remove the BrandViz connector in the app that holds it, and that app can no longer reach your brands. Jobs started through the server need a destination, so connect GitHub or a CMS from the integrations hub first.
Frequently asked questions
Read the MCP server setup docs or the guide to running GEO fixes from your AI agent.
What is an MCP server for SEO?
An MCP server for SEO is a service that gives AI agents access to search or visibility data and actions through the Model Context Protocol, an open standard for connecting assistants to tools. BrandViz.AI's MCP server covers AI search visibility: how ChatGPT, Claude, Gemini and Perplexity describe your brand, which pages they cite, and the technical audit findings and AI crawler activity behind the gaps.
Which AI agents can connect?
Claude on web and desktop, Claude Code, Cursor and ChatGPT can connect, as can any other MCP client that supports remote servers with OAuth sign-in. Each one is pointed at https://mcp.brandviz.ai/mcp and sends you to BrandViz.AI to sign in the first time, so there is no API key to copy.
Does it change my site without approval?
No. The server cannot publish. Jobs that build something through it deliver a pull request on a brandviz-agent/ branch or an unpublished CMS draft, and a person on your team merges or publishes it. An agent can approve a pending task, which starts a job and uses monthly allowance, and it can change what BrandViz.AI measures, but neither touches your live site.
Which plans include the MCP server?
Reading your visibility data through the MCP server works on every plan. Content tasks, audits and jobs need the Engine, which both the Starter and Growth plans include. Setup takes a connector URL and a sign-in, and the steps for each client are in the MCP setup docs.
Other integrations
How every integration works
Work from your agent, approve where it lands.
Start the trial, connect your agent to the MCP server, and review the first pull request or CMS draft where your team already works.
7 days free. Cancel anytime. Nothing publishes without you.