Amazon, Dstillery, and Plurio Just Proved MCP Advertising Works

#mcp #ai #advertising #monetization

Three things happened in the first week of March 2026 that should matter to anyone building MCP tools:

Amazon Ads opened its MCP server to public beta
Dstillery shipped DS-1, an MCP-native audience builder integrated with The Trade Desk
Plurio raised $3.5M to build agentic AI for performance marketing

These aren't experiments. They're shipping products.

Amazon Ads Goes MCP

Amazon's MCP server lets sellers create campaigns, set budgets, and optimize bids through natural language. Instead of clicking through the Amazon Ads console, you tell your AI agent "create a Sponsored Products campaign with a $5k budget and auto-targeting" and it happens.

Early numbers: 70-80% reduction in campaign setup time.

This is the biggest company to bet on MCP as an advertising protocol. When Amazon builds an MCP server for ads, the question shifts from "will agents handle ad workflows?" to "how fast will everyone else follow?"

Dstillery Builds on MCP

Dstillery's DS-1 uses MCP to let clients build custom audiences in minutes instead of days. It connects directly to The Trade Desk for activation. Their approach: use MCP as the interface layer between AI agents and ad targeting data.

This matters because Dstillery didn't build a chatbot. They built an MCP server that speaks the same protocol every AI agent already understands. Any agent that supports MCP can now build audiences programmatically.

Plurio Gets Funded

Plurio raised $3.5M seed to apply agentic AI to performance marketing. They already manage over $100M in annual ad spend and processed $20M through AI agent pilots in their first four months.

Venture money follows conviction. When investors put $3.5M into "agentic advertising" as a category, it signals that the market is real enough to bet on.

What This Means for MCP Developers

If you build MCP tools, you're sitting on distribution infrastructure that the ad industry is now actively building around. The pieces:

Supply side: MCP servers that serve recommendations, search results, or product comparisons. These are the surfaces where ads have context.
Demand side: Advertisers who want to reach users at the moment an AI agent is helping them decide. Amazon proved this works.
Protocol layer: MCP itself, which is becoming the standard interface between agents and external services. No proprietary SDK needed.

The gap right now: most MCP servers don't monetize. Out of 11,000+ servers indexed across directories, fewer than 5% generate any revenue. The infrastructure exists. The demand exists. The connection between supply and demand is what's missing.

The Ad Network Layer

This is what we're building with agentic-ads. An ad network purpose-built for MCP tools. Developers integrate three tools into their MCP server:

get_sponsored_recommendations - returns contextual sponsored results
record_click - tracks engagement
record_conversion - tracks completed actions

Revenue split: developers keep 70%. Integration takes about 15 minutes.

The live API runs at agentic-ads-production.up.railway.app. The npm package is agentic-ads.

Amazon, Dstillery, and Plurio validated the thesis from the top down. We're building from the bottom up: give individual MCP developers the same monetization tools that enterprise ad tech companies are deploying internally.