a full-lifecycle agentic media OS with 20+ AI agents running planning, buying, and optimization across Google, Meta, TikTok, and The Trade Desk inside a single workflow — the first major holdco to put agentic media buying fully into production.
Why it matters: The playbook for building agentic media systems is now visible — agency-built, platform-native, human-at-decision-gates — the architecture growth operators can replicate without the holdco budget.
with an open MCP integration layer letting ad-tech partners, publishers, and data platforms plug directly into real-time media decisioning, going live at Cannes.
Why it matters: MCP-as-backbone is becoming the agency infrastructure standard — operators now expect MCP-native agentic integration, not point-to-point builds.
performance operators can query media quality and campaign data via natural language today, with an autonomous Activation Agent executing approved campaign changes arriving in Q3.
Why it matters: Brand-safety and measurement data are now queryable mid-campaign by AI agents — it changes how agentic optimization loops get built for paid media.
explicitly naming agencies for overpromising AI in pitches and shifting its positioning to proven business outcomes over feature announcements.
Why it matters: The world's largest holdco is running "show, don't tell" as a competitive play — the window for undifferentiated AI-feature hype in agency pitches is closing fast.