Now loading...
Amazon has unveiled a new tool to make its advertising platform more accessible to artificial intelligence systems. During the Interactive Advertising Bureau’s Annual Leadership Meeting, the company launched an open beta for its Amazon Ads Model Context Protocol Server, designed to streamline connections between AI agents and Amazon’s ad infrastructure.
This server enables advertisers and technology partners to integrate their AI tools with Amazon Ads using just one setup, eliminating the need for tailored links for each process. Paula Despins, vice president of advertising measurement at Amazon Ads, explained that users can link their agents to the system in mere minutes, transforming complex interactions into something simpler.
Drawing from the open standard Model Context Protocol created by Anthropic, the server functions as an intermediary that converts everyday language instructions into precise application programming interface requests. Positioned between AI agents and Amazon’s ad APIs, it empowers these agents to perform essential functions such as launching campaigns, modifying spending limits, overseeing product listings, and retrieving performance data.
Despins highlighted how the protocol offers a unified communication method, allowing AI systems to interact with Amazon’s platforms without bespoke coding for every connection. A key feature involves bundling routine advertising tasks into simplified tools, which let agents complete multi-part operations via a single directive.
Advertising processes frequently demand several coordinated steps across various tools, but these MCP tools consolidate them, easing the burden on AI agents. Despins noted that current agents often grapple with grasping API functionalities, their mechanics, and appropriate versions, which can cause delays or errors.
Internal tests illustrated this challenge: In one trial, an AI agent devised custom code to compile a conversion path report spanning over three years of data through Amazon Marketing Cloud, bypassing a ready-made API. In other instances, agents selected obsolete but valid API versions, errors unlikely from experienced developers. The MCP tools counter these issues by providing clear guidance for standard procedures, freeing agents to focus on higher-level decisions.
Despins emphasized that teaching agents the intricacies of ad systems is resource-intensive, but MCP shifts the approach by starting with natural language and supplying step-by-step manuals for frequent tasks, ensuring accuracy on fundamentals.
This public beta builds on a limited pilot Amazon ran last year and aligns with broader industry initiatives, like the Ad Context Protocol introduced in 2023 to standardize AI interactions in advertising. AdCP represents a similar push as companies move from hands-on routines to automated, agent-driven operations.
The timing reflects Amazon’s growing emphasis on AI in commerce. Last year’s third-quarter earnings showed advertising revenue climbing 24 percent to $17.7 billion, with CEO Andy Jassy suggesting AI agents could reshape online shopping experiences. Those results underscore the company’s momentum in this space.
Amazon’s advertising collaborators had specifically requested such a standardized interface, though Despins did not disclose participation numbers from the prior pilot.
