Why Magnite Built a Seller Agent — and What It Signals for AdCP

The Advertising Context Protocol (AdCP) is an early but promising open standard that aims to unify how advertising platforms, publishers, data providers, and AI agents communicate. Built on top of the Model Context Protocol (MCP), AdCP gives buyer- and seller-side AI agents a shared language to interpret campaign goals, inventory, and audience signals without relying on today’s traditional oRTB request/response model.

In this emerging framework, a buyer agent uses an LLM to translate a campaign brief or advertiser site into structured intent. A seller agent then uses an LLM to match that intent with appropriate audiences and inventory. Once both sides agree, a campaign is placed. A perfect use of AI to streamline manual workflows that previously missed important audiences or supply.

While the vision is compelling—simpler buying workflows, broader access to premium ad products, lower operational costs, and reduced tech tax—it’s extremely early. Much of the workflow is conceptual, several critical pipes are still being defined and tests are first coming online. With these early experiments underway, the promise is enough to justify thoughtful exploration.

Magnite’s Perspective

Magnite sees AdCP as a potential catalyst for more open, flexible and efficient buying and selling. AI lowers the barrier for agencies and brands to build or license their own buyer agents, which could decentralize how demand enters the ecosystem. 

In a world where both buyers and sellers are represented by agents, the role of the sell side becomes even more important for enabling supply intelligence, curation and scaled execution. 

Why We Built a Seller Agent Into SpringServe

Magnite’s role is to bring our publisher clients demand across any rails buyers choose to transact upon—whether that’s programmatic, IO workflows, or emerging agent-driven channels. Our first step was to embed a seller agent directly within SpringServe where we launched our first agentic tests in December. Our seller agent allows our publisher partners to benefit from:

• Engaging in AdCP tests 
• Modernizing today’s 1:1 buying methods (IO, PMP) with automation
• Collaborating with the emerging Buyer Agents
• Learning and defining this promising category

Early proofs of concept simply push these campaigns into ad servers, but the industry will ultimately need scaled systems that can support many hundreds of thousands of campaigns and sophisticated optimization.

What’s Next

Magnite ran our first test in December with Scope3 serving as the buyer agent, with additional pilots lined up to test in early 2026. While it’s far too early to predict outcomes, the industry momentum and the potential efficiency gains make this a space worth engaging in thoughtfully.

Several leading partners are already testing Magnite’s seller agent and exploring its potential, offering early perspectives on how agent-driven workflows like AdCP could reshape how premium inventory is discovered, valued, and transacted.

“As a global CTV publisher, LG Ad Solutions is focused on technologies that help buyers better understand and activate against our premium inventory. Magnite’s early work with seller agents marks an important step toward a more automated, intelligence-driven marketplace, one where publishers can more clearly communicate the value of their audiences. While still early, we’re optimistic about AdCP’s potential to create more efficient, transparent paths to demand and encouraged by partners like Magnite helping shape what’s next.” 

— Mike Evans, SVP, Head of US Sales, LG Ad Solutions

“In December we set live one of the first AdCP buys working with Magnite’s seller agent. We’re hopeful this can bring even more supply into the programmatic sphere, and present new ways to deliver client outcomes. In 2026 we’re excited to start scaling agentic ad buying through MiQ Sigma’s trading agent, and in the process deliver better results for advertisers.” 

— John Goulding, Global Chief Strategy Officer, MiQ

“At Warner Bros. Discovery, we’re always looking for ways to deepen the alignment we have with advertisers. Testing Magnite’s seller agent has given us a look into how agentic decisioning can elevate the precision and clarity with which our supply is understood and activated. We see meaningful potential for these emerging workflows to strengthen publisher value, expand access to premium supply, and help shape a more open, intelligence-driven marketplace across linear and digital supply.”

— Todd Overstreet, SVP, Ad Technology and Operations, Warner Bros. Discovery

Building a seller agent in SpringServe is a measured but meaningful first step. It allows us to learn quickly, collaborate with innovators, and ensure that—whatever transactional rails buyers adopt—Magnite is positioned to deliver demand to our publishers and help shape an open, interoperable future.