PubMatic introduces AgenticOS to enable autonomous agent-to-agent programmatic buying

PubMatic has launched AgenticOS, a new operating system designed to support autonomous, agent-to-agent execution of programmatic advertising across premium digital inventory. The company said the platform is intended to address growing complexity in digital advertising by enabling software agents to plan, transact, and optimize campaigns continuously within defined constraints.

The launch comes as advertising expands across new formats, devices, and channels, increasing pressure on traditional programmatic workflows that rely heavily on manual setup and fragmented decision-making. AgenticOS is positioned as a system-level layer that translates advertiser intent—such as objectives, budgets, brand-safety rules, and creative parameters—into coordinated execution handled by multiple intelligent agents.

According to PubMatic, the platform consolidates several years of development in agentic AI into a single operating environment. Capabilities that were previously deployed as individual tools, including yield optimization, inventory discovery, forecasting, pacing, and troubleshooting, are now orchestrated within a unified framework designed to operate at scale in live markets.

The architecture of AgenticOS is structured across three layers. Its infrastructure layer uses NVIDIA-accelerated computing to support low-latency inference and real-time decisioning across millions of ad transactions per second. The application layer embeds agentic functionality throughout PubMatic’s platform, allowing agents to interpret intent via emerging standards such as the Ad Context Protocol and Model Context Protocol. The transaction layer connects agent-driven decisions directly to PubMatic’s Activate buying platform, enabling execution across Programmatic Guaranteed and private marketplace deals while remaining compatible with existing buying models.

PubMatic said early tests of agent-led campaigns have shown reductions in operational overhead, including shorter campaign setup times and faster issue resolution. One pilot campaign, run in December with independent agency Butler/Till and Geloso Beverage Group’s Clubtails brand, used natural-language input through a large language model to guide automated planning, buying, and optimization within predefined limits.

The company is launching AgenticOS in partnership with agencies, advertisers, and technology firms including WPP Media, Butler/Till, MiQ, Wpromote, and others. These partners are participating in live testing and deployment as agentic workflows move from experimental use cases into active market execution.

To support broader adoption, PubMatic has also introduced an Agentic AI Acceleration Program aimed at helping ecosystem participants move from pilots to in-market campaigns within weeks. The initiative focuses on governance, execution readiness, and measurable outcomes, with deployments expected to continue through the first quarter of 2026.