PubMatic launches tool for AI-supported advertising campaigns

PubMatic has announced the launch of AgenticOS, an operating system for agent-based advertising delivery. The AI-powered system coordinates autonomous advertising campaigns in premium digital environments and delivers programmatic advertising quickly, consistently and in a controlled manner across different system levels.

At the launch of AgenticOS, PubMatic is working with advertisers, agencies and publishers who are actively engaged in testing, designing and deploying agent-driven workflows. Initial partners include WPP Media, Butler/Till and MiQ.

Intelligent agents instead of manual control

With AgenticOS, advertisers define their goals, parameters, brand safety requirements and creative parameters via their preferred LLM interface. PubMatic's platform then transmits these specifications to a series of coordinated, intelligent agents that continuously plan, execute and optimize campaigns.

Unlike isolated execution layers or single-protocol connectors, AgenticOS runs on NVIDIA-accelerated AI infrastructure and is embedded in PubMatic's global environment. This foundation enables the coordination of millions of advertising transactions per second.

"Agents will change advertising technology in every way this year and AgenticOS is already showing excellent results in terms of planning, forecasting, troubleshooting and measurability," says Nicolas Poppitz, who has been Country Manager for the Central Europe region at PubMatic since January 2026. He adds: "PubMatic's ambitions to bring AgenticOS to market quickly are particularly interesting for advertisers who advertise in complex environments such as CTV."

Three-layer architecture for autonomous decisions

AgenticOS is based on PubMatic's Architecture of Advertising Intelligence, a three-layer framework designed to support autonomous decision-making responsibly and on a global scale.

The infrastructure layer uses NVIDIA-accelerated computing and enables microsecond inference and real-time privacy-compliant integration across millions of auctions per second. This results in up to five times faster decisions, sub-millisecond response times and significantly fewer auction timeouts.

At the application level, agentic functions integrated into the PubMatic platform interpret specifications that are defined via protocols such as the Ad Context Protocol (AdCP) and the Model Context Protocol (MCP). These agents automate and coordinate central advertising functions such as planning, forecasting, pacing, optimization, troubleshooting and measurement. Initial tests with agent-based campaigns show that the setup time for campaigns is reduced by 87 percent and the resolution of problems by 70 percent.

The transaction layer puts agent-based intelligence into practice. An interoperable execution environment connects agent-based decisioning directly to PubMatic's Activate buying platform, bringing together premium inventory, live campaign signals and bidding infrastructure to execute transactions in real-time across Programmatic Guaranteed and Private Marketplace transactions.

Agentic AI Acceleration Program launched

To support the introduction in practice, PubMatic is launching the Agentic AI Acceleration Program. The initiative is designed to help advertisers, agencies, publishers and partners move from initial testing to live Agentic workflows within weeks. The program supports the market launch in the first quarter of 2026 and focuses on execution, governance and measurable results.

Further information on PubMatics AgenticOS and participation in upcoming Agentic pilot projects can be found at https://go.pubmatic.com/agenticOS available. A demo version can be found at https://pubmatic.com/video/pubmatic-agenticos-in-action/.