PubMatic and Havas debut first agentic CTV campaign for Spain’s Movistar

This activation signals the transition of agentic AI from experimental pilots to operational infrastructure for major holding companies. By automating the planning and optimization of complex CTV workflows, PubMatic is addressing the industry's critical demand for lower operational overhead and improved transparency. As programmatic markets face increasing fragmentation and privacy-driven data signal loss, agentic systems offer a scalable method for maintaining performance without expanding manual labor pools. Watch for other global holding companies to integrate similar agentic layers into their internal operating systems, such as Havas’s Converged.AI, in the second half of 2026.

The Spanish campaign follows the industry-wide debut of PubMatic’s AgenticOS at CES in January 2026. Per PubMatic, the system was built on NVIDIA-accelerated computing to enable low-latency coordination across millions of advertising transactions per second. At its launch, early partners included WPP Media and MiQ, highlighting a shift among global agencies toward agent-to-agent advertising. These systems utilize the Model Context Protocol (MCP) to bridge Large Language Models like Anthropic’s Claude with live performance data, effectively turning natural language intent into real-time media buying.

This movement is categorized by a broader push toward 'Agency-as-an-Operating-System.' In June 2026, Havas CEO Yannick Bolloré stated that agencies successfully integrating AI will inevitably replace those that do not, noting that Havas has trained 23,000 employees on agentic tools. Similarly, per VideoWeek reporting in June 2026, other publishers like Hearst Magazines have launched internal agentic layers such as 'AURA IQ' to automate the RFP process and build bespoke audience segments from first-party data.

However, some analysts remain cautious about the rapid pace of adoption. Per MediaPost (June 2026), while agentic AI can streamline the programmatic waterfall, it introduces new transparency challenges regarding the 'reasoning tokens' used by AI to make buying decisions. Procurement specialists have warned that without human-in-the-loop guardrails, the cost of the compute required to run these autonomous agents could eventually offset the efficiency gains realized through lower CPMs.