Claude and MCP for marketing: integration with AMP

Artificial intelligence is already part of the daily work of many marketing teams. We use it to research, analyze information, generate content, or speed up tasks. However, much of this work still happens outside the platforms where our data actually lives.

We export a report, provide it to the AI, add context, get an answer, and return to the original platform to continue working.

That is why we believe the next evolution of AI in marketing will not depend solely on having smarter models, but on connecting them to companies' real data, tools, and processes.

This is where Model Context Protocol (MCP) becomes particularly relevant. At Adsmurai, we are already applying this approach by connecting Claude to Adsmurai Marketing Platform (AMP) through our MCP.

MCP and artificial intelligence applied to marketing

Model Context Protocol allows artificial intelligence applications to connect with external tools and systems in a structured way.

The difference matters. A model like Claude can understand how a paid media strategy works, explain an attribution model, or help us interpret ROAS. But by default, it does not know what is happening in our campaigns, which products are in our feed, or how our brand is evolving.

It needs context.

Until now, that context has mainly been provided manually through files, exports, documents, or increasingly detailed prompts. With MCP, we can move towards a different model, where AI can directly access the authorized tools it needs to handle a request.

Instead of asking what data we need to give AI, we can start asking what tools AI needs to help us.

How Claude works with Adsmurai MCP

Adsmurai MCP allows you to connect Claude Desktop and Claude Code to Adsmurai Marketing Platform.

The logic is simple: Claude acts as the conversational interface, AMP provides the marketing data and capabilities, and MCP connects the two environments.

User → Claude → Adsmurai MCP → AMP → tools and data

The connection is established using an individual personal access token. In addition, the capabilities Claude can use depend on each user's permissions in AMP and the MCP servers they choose to connect.

This allows Claude to work with different areas of Adsmurai Marketing Platform, including Attribution, Benchmarks, Brand Tracker, Campaign Tools, Dashboards, Feeds, and Media Plans.

We are not moving AMP into Claude. We are creating a new way to access its capabilities.

What Claude can do when connected to Adsmurai Marketing Platform

The most immediate application is the ability to access marketing information using natural language.

A user could start by asking:

“Which campaigns are performing best right now?”

Claude can use the tools available in AMP to retrieve the information it needs. From that first answer, the conversation can evolve:

“Which ones have lost efficiency over the past few weeks?”

“Which channels are mainly driving this change?”

“Where should we investigate before making an investment decision?”

What makes this interesting is not simply being able to query information using natural language. It is the ability to build a progressive analysis without having to reconstruct the context at every step.

The same applies to other areas of marketing. We can ask:

“Show me my attribution results by channel.”

Or:

“How is my brand visibility evolving compared to my main competitors?”

Or even:

“Which active campaigns are not linked to a media plan?”

Each question can activate different AMP capabilities depending on what the user needs to solve.

From checking dashboards to answering business questions

This evolution also challenges the way we have traditionally interacted with marketing software.

Until now, finding an answer first required knowing the tool: which platform to open, which dashboard to check, which filters to apply and, in many cases, which information to export and combine with another source.

But business problems are not organized by platform.

An ecommerce business does not ask what it can discover inside its feed management tool. It asks:

“Which products should I prioritize?”

A paid media team does not necessarily want to check five dashboards. It wants to know:

“Why are we losing efficiency this month?”

Answering these questions may require combining information from campaigns, attribution, products, or planning.

With MCP, the starting point can be the business question, while technology takes care of accessing the capabilities needed to help answer it.

We believe this shift is particularly relevant: moving from an experience centered on navigating tools to one centered on expressing intentions.

Claude and AMP: from analysis to action

MCP is not limited to accessing information. Depending on the available tools and the user's permissions, Claude can also help perform certain actions within AMP.

This enables workflows where different stages of the process are connected:

Question → analysis → recommendation → action.

For example, we could detect a campaign experiencing a drop in performance, investigate the causes, and continue working with that information without leaving the conversation.

But the closer AI gets to execution, the more important governance becomes.

The connection to AMP uses individual credentials, respects user permissions, and allows users to select which capabilities are connected to Claude. In addition, certain actions require confirmation before they are executed, and access can be revoked from AMP.

The move towards more autonomous systems only makes sense if automation and control evolve at the same pace.

MCP and the future of artificial intelligence in marketing

The first stage of generative AI was dominated by prompts. The next is beginning to be shaped by the connection between models, data, and tools.

And we believe this is where one of the biggest opportunities for marketing lies.

Almost every company will have access to highly capable artificial intelligence models. The difference will not simply be which model they use, but what context that model can use, which tools it can access, and which processes it can help execute.

At Adsmurai, we see MCP as part of a broader evolution towards connected growth systems. AMP acts as the infrastructure connecting marketing data and capabilities; MCP enables those capabilities to also be used from new artificial intelligence environments such as Claude.

Perhaps that is why the most interesting question is no longer what an AI can answer.

The question is becoming what it can do when it understands our context and can work with our tools.