How new infrastructure, like the Model Context Protocol, is reshaping marketing workflows

A new open standard is emerging to aid the connection of data and AI/LLM tools: MCP, or the Model Context Protocol.

Amazon Ads recently introduced an MCP server in open beta, designed to allow AI agents to easily interact with Amazon Ads.

“The Amazon Ads MCP Server is part of our broader investment in agentic interfaces and capabilities,” said Alex Brockhoff, senior technical product manager of developer experience at Amazon Ads. “It provides the foundation that enables our partners and developers to connect their own AI agents to Amazon Ads capabilities — and it powers Ads Agent, which works alongside advertisers in the console to help plan, launch and optimize campaigns using natural language.”

The Model Context Protocol sets an open standard for agent connectivity

The Model Context Protocol is an open-source standard for connecting AI applications to external systems — crucial for interoperability. Using this standard, AI applications (like Claude) can connect to data sources and tools to access key information and perform necessary tasks.

“APIs remain the foundation of how partners programmatically access Amazon Ads capabilities. They play a role behind MCP, and we’ll continue to invest in them,” Brockhoff said. “What MCP adds is a contextual layer that makes those same capabilities easily usable by AI agents.”

Consider setting up a campaign: creating the campaign, configuring ad groups, building out ads — each a separate API call. The tools on the MCP Server orchestrate these steps so an AI agent doesn’t have to figure it out from scratch. An advertiser can describe the campaign they want to launch in a single prompt, and the tool handles the full sequence, returning a complete campaign ready for review and approval. Partners can layer their own intelligence on top of the common advertising workflows offered on the Amazon Ads MCP Server, adding proprietary insights, best practices and domain expertise to the foundation Amazon Ads provides.

The MCP Server’s capabilities continue to expand. A recent addition allows advertisers to run saved Amazon Marketing Cloud (AMC) queries through the MCP server using their own large language models. 

“We’re expanding our MCP tools to include AMC functionality, so advertisers can run their saved queries using their own LLMs and incorporate those insights directly into AI-driven workflows,” Brockhoff said. The goal is to allow advanced measurement and analytics capabilities to become part of the same AI-driven workflows used to create campaigns, analyze results and recommend optimizations.

As the technology supporting AI continues to shift, agentic adoption will dramatically increase. By 2028, 33% of enterprise software will include agentic AI, compared to less than 1% in 2024, according to Gartner. For advertising, that opens a wide lane for partners to innovate.

How one company layered its own intelligence on top of Amazon Ads MCP Server

Hector Ai, an AI-powered commerce media platform and Amazon Ads technology provider, combined its own intelligence with Amazon Ads MCP Server capabilities, enabling advertisers to access Hector Ai’s optimization suite through a connector for Claude and other AI assistants.

Meher Patel, founder and CEO of Hector Ai, knew that Amazon sellers and agencies were already working within Claude and ChatGPT to reason, experiment and build workflows. Rather than add another interface, Patel brought Amazon advertising intelligence to where customers were already working.

“Our hybrid approach uses Hector Ai’s MCP as the intelligence and orchestration layer, providing clean, contextualized insights across complex Amazon Ads insights, while leveraging Amazon’s MCP for direct, high-speed execution,” Patel said. “This separation lowers friction, preserves accuracy and allows us to move faster without duplicating execution logic.”

Because Claude already connects to other MCP servers — Slack, QuickBooks, Notion and creative tools — adding Amazon Ads into that environment means advertising insights can participate naturally in broader, agent-driven workflows. The Hector Ai team shifted from dashboard-centric optimization to workflow-centric decision systems, where advertising insights are no longer siloed.

“With MCP-based architectures, Amazon Ads insights can flow directly into AI-native environments where users are already building agents and workflows, connecting performance insights with finance, operations, creative and planning tools,” Patel said. “This makes decisions more contextual, explainable and faster. It also enables continuous reasoning loops, where AI can analyze performance, explain drivers, recommend actions, execute changes and learn from outcomes — all within the same environment. That moves ad tech from reactive optimization toward adaptive systems that evolve alongside the advertiser’s business.”

MCP has paved the way for many innovations. However, Patel notes that AI systems are only as effective as the data they can reason over, which means longer lookback windows, better historical access and more complete behavioral signals are needed to build meaningful situational awareness.

“While MCP provides powerful tools and actions, intelligent systems still need stronger embedded context to navigate complex advertising ecosystems reliably and avoid misinterpretation,” said Patel. “This includes clearer resource relationships, domain-aware guardrails and better guidance for AI-driven reasoning. At Hector Ai, we’re investing in building these contextual layers around MCP, so AI systems don’t just know what actions are possible, but when and why to use them. That’s what turns MCP from a powerful interface into a dependable decision-making foundation.”

Partner insights from Amazon Ads