Your AI Can Now Read Search Console and HubSpot. Here's What It Still Won't Do

MCP for marketing means connecting AI tools like ChatGPT and Claude straight to your marketing data, such as HubSpot, Ahrefs or Semrush, so you can ask about live numbers instead of pasting exports. It makes answers easy. It doesn't make anyone ask the questions.

That second part is the gap most marketing teams haven't noticed yet. In 2026, almost every major marketing tool can be plugged into an AI chat. Yet the weekly checks that actually move traffic and leads still go undone in busy weeks, because a connected chat window does nothing until somebody opens it.

This guide explains what MCP is in plain words, what launched over the past year, what connected AI does well, and the three things it still won't do for you.

What Is MCP, in Plain English for Marketers?

MCP, the Model Context Protocol, is a shared standard for connecting AI tools to other software. Think of it as a universal plug: a tool builds one MCP server, and any AI app that speaks MCP can use it to read data or take actions in that tool.

Before MCP, getting marketing data into an AI meant exporting a CSV, pasting it into a chat, and hoping nothing important got cut off. With an MCP connection, you ask "which landing pages lost the most clicks this month?" and the AI fetches the answer from the source.

The standard has moved from experiment to infrastructure quickly. In December 2025, Anthropic donated MCP to the new Agentic AI Foundation under the Linux Foundation, co-founded with Block and OpenAI. At that point its official SDKs had more than 97 million monthly downloads.

If you want the bigger picture of where this fits, our guide to agentic AI for small business covers the shift from AI you prompt to AI that works on its own.

Which Marketing Tools Launched MCP Servers in 2025 and 2026?

Over the past year, the tools marketers use every day have shipped ways to plug their data into AI chat. A few examples:

• Ahrefs made its remote MCP server live in October 2025, so you can ask Claude or ChatGPT for keyword, backlink and competitor data without opening Ahrefs.
• Semrush offers an MCP connector that gives AI tools a direct connection to live Semrush data.
• HubSpot runs a remote MCP server that connects AI tools to your CRM: contacts, companies, deals and more.
• Salesforce made its MCP server for Marketing Cloud Engagement generally available in June 2026 for content and campaign management.
• SearchAtlas released a marketing MCP that works with Claude, ChatGPT and Gemini, covered by eMarketer.

Google Search Console is the notable gap. As of October 2026 we couldn't find an official Search Console MCP server from Google; community-built servers and connector products fill in. That matters, because Search Console is where most small teams' organic traffic questions start.

What MCP-Connected AI Chat Does Well for Marketing Teams

Connected AI chat is genuinely useful, and it's worth being clear about where it shines before getting to its limits. It does four things well:

• Answers without exports. You ask a question in plain words and get an answer from live data in seconds, instead of building a report.
• Joins tools in one question. "Which deals in HubSpot came from pages that lost search traffic?" used to take two exports and a spreadsheet.
• Explores quickly. When something looks off, you can follow a hunch through five questions in five minutes.
• Drafts from real numbers. A summary for your manager or a client update can start from the actual data, not from memory.

For one-off and exploratory questions, connected chat is often the right tool. If you only need to know something once, you don't need anything more.

Have a check that should happen every week, not just when you remember? Sign Up for Free and give it to a teammate. No credit card required · Free to start · Cancel anytime.

Three Things MCP-Connected AI Still Won't Do

Access isn't ownership. MCP gives AI a way into your data, but it doesn't give anyone the job of looking. Three things still fall through:

1. It won't ask on a busy week. Connected chat answers when someone opens it. The weeks your team is too stretched to open it are exactly when a page starts losing clicks or a campaign starts bringing in the wrong leads.
2. It won't remember what it found last week. Each new chat usually starts cold. To know what changed, you have to paste last week's answer back in or re-ask the same question with the right dates, every time.
3. It won't raise the question you didn't think of. You get answers to what you ask. The page sitting in fourth place with thousands of impressions and almost no clicks stays invisible until someone thinks to look for exactly that.

There's a fourth point worth checking too: what the connection is allowed to change. Some marketing MCP servers can create and update records as well as read them. Before you connect a CRM or ad account to any AI, look at what it can write, not just what it can read.

None of this is a flaw in MCP. It's a plug, not a person. The work of deciding what to check, checking it every week and noticing what changed still needs an owner.

A Worked Example: The Same Search Console Question, Three Weeks Running

Here's the difference in practice. Say your marketing lead wants to know which pages are seen in Google but not clicked: pages in the top five positions with more than 1,000 impressions in the last 28 days and a click-through rate under 1%.

With connected chat: in week one, they ask the question, get a list of six pages, and fix the title on the first two. In week two, a launch eats the week and nobody asks. In week three, they ask again, get a list of five pages, and have to work out by hand which are new and whether the two fixed pages improved.

With a teammate that owns the job: the same sentence is set up once as a scheduled task. Every week it reads Search Console, compares 28 days against the 28 before, and keeps notes on what it reported last time. In week three, what's waiting is short: two pages improved after their title changes, one new page has joined the list, and three are unchanged.

Same data, same question. The difference is that the second version keeps going in the week nobody had time, and it tells you what changed instead of starting from scratch.

Connected AI Chat vs an AI Teammate With a Job

The two aren't rivals. Most teams will use both: connected chat when someone has a question, and a teammate for the checks that should happen whether or not anyone remembers.

How Kuvai Teammates Use Search Console and HubSpot

Kuvai is built for the second column of that table: you give a teammate a role and a recurring job, and it works inside your tools. Here's how that works today.

• You connect each tool through its own login. Kuvai teammates connect to 1,000+ tools as of October 2026, including Google Search Console, Google Analytics, HubSpot, Semrush and Google Ads. You sign in through each tool's own window, so Kuvai never sees your password, and you can disconnect one tool at a time.
• You choose what each teammate can use. Giving a teammate access to a tool always needs your explicit approval, and a teammate only uses tools you've connected and assigned.
• It reads by default. Writing, such as updating a HubSpot record, is one explicit step you've asked for, and it won't post the same thing twice.
• Recurring work is set up by chatting. Tell a teammate "every week, check which pages rank in the top five with over 1,000 impressions and a CTR under 1%," and it becomes a scheduled task that runs with nobody in the app.
• It keeps notes. Each teammate remembers what matters about your business and its past work, so this week's check builds on last week's.
• It shows its work. Every answer carries a "Transparency" trail of the steps taken and the sources used, and results can include charts and tables in the answer itself.

Our Search Console use case shows a weekly traffic-to-leads job in detail. If you're weighing the risks of connecting tools at all, read our guide to the real security risks of connecting AI to your business tools.

When Is Connected Chat Enough, and When Do You Need a Teammate?

Use connected chat when the question is one-off, exploratory, or something you'll only need again in a few months. It's fast, and if you already pay for ChatGPT or Claude, the connection is often free or cheap.

Give the job to a teammate when three things are true:

1. The check should happen on a schedule, weekly or monthly, whether or not anyone is free.
2. What matters is the change, not the snapshot: which pages got worse, which deals went quiet, which campaigns stopped bringing in signups.
3. Someone already says "we should look at that more often" and doesn't. That's a job without an owner.

For most small marketing teams, the first job worth handing over is a weekly Search Console review. It's read-only, easy to check, and the kind of work that slips first when the week gets busy. For more on why AI that forgets is the real limit, see why AI forgets your business.

Give your weekly marketing checks an owner. Sign Up for Free, connect Search Console and describe the job once. No credit card required · Free to start · Cancel anytime.