Yahoo has integrated a new Agentic AI layer into its Yahoo DSP buying platform, with the first features now live for global advertisers in Australia, wider Asia-Pacific and other global markets.

The launch embeds AI agents directly into the demand-side platform’s workflow. The system now sits across campaign planning, activation, optimisation and measurement, and builds on Yahoo Blueprint, the platform’s existing AI engine.

The company has also introduced a framework it calls “Yours, Mine and Ours”. Advertisers can connect their own AI models. They can also use Yahoo DSP’s native agents. They can combine both through secure connections using the Model Context Protocol or application programming interfaces.

Yahoo positions the framework as a response to how media and marketing teams work across regions such as Australia and Asia-Pacific. Many large advertisers and agencies now mix proprietary tools with platform-based systems while seeking governance and transparency over data and decisioning.

Adam Roodman, GM, Yahoo DSP, said the new layer alters the mechanics of digital media buying for users of the platform.

“Agentic AI changes how media buying actually gets done,” said Adam Roodman, GM, Yahoo DSP. “By building it directly into Yahoo DSP and allowing advertisers to connect their own AI alongside ours, we’re giving teams a faster, more flexible way to plan, optimise, and act, without sacrificing transparency or control.”

Yahoo has made the initial Agentic AI functions available in markets including Canada, the UK, Australia and Singapore. The company plans a further roadmap of products through 2026.

AI connection

The first set of tools includes what Yahoo describes as “Campaign Activation (Yours)”. Advertisers and partners can use the Model Context Protocol to link external AI agents with Yahoo DSP. One early example is a trafficking agent built by Newton and RPA. This agent manages campaign setup and execution and reduces manual steps in programmatic guaranteed deals.

Yahoo says additional MCP-based integrations will allow other data sources to interact with the DSP. These may include audience data, performance information and campaign insights.

RPA is already using agentic tools in its programmatic workflows.

“By applying agentic AI to programmatic media workflows, RPA is removing operational barriers-enabling our teams to focus their human expertise on diversifying and maximising marketplace partnerships in service of our clients’ business outcomes,” said Lisa Herdman SVP, Executive Director Video Investment & Marketplace Intelligence at RPA.

Proactive troubleshooting

A second feature, described as the “Troubleshooting Agent (Mine)”, runs continuously inside the DSP. It scans campaigns for pacing and delivery issues. When a line item underpaces, the agent analyses possible causes across targeting, supply sources and platform settings. It then proposes corrective actions and can carry them out with user approval inside the interface.

Yahoo has also released an “Audience Exploration (Ours)” function. Partners can access Yahoo DSP audience metadata through APIs or the MCP standard. External AI agents can query segment size, demographics and pricing. They can then suggest relevant Yahoo audience segments that align with campaign objectives.

John Goulding, Global Chief Strategy Officer at MiQ, said his organisation expects operational gains from the embedded agents and interoperability with internal systems.

“Built-in agents speed up tasks like diagnosing under-delivery or building audiences, from hours to seconds. Interoperability between the Yahoo DSP agents and our own internal agents via MiQ Sigma will also enable smarter, more tailored decision-making. The Agentic AI strategy within Yahoo DSP will give our teams real-world benefits, elevating the speed, intelligence, and effectiveness of our daily work,” said Goulding.

Global and local focus

Yahoo is positioning Agentic AI as a globally consistent layer that adapts to market-level requirements. The company highlights varying regulations, media ecosystems and operating models across regions.

“As marketers navigate increasing complexity across markets, channels, and regulations, flexibility and control become just as important as intelligence,” said Jon Finnie, VP, International Sales, Yahoo. “Our agentic AI approach is designed to scale globally while respecting local market realities, giving advertisers the choice to deploy AI in ways that align with their business, their teams, and their regions.”

Agentic AI currently supports five broad categories of agent within Yahoo DSP. Insight agents scan campaign and audience data for trends and anomalies. Traffic agents guide campaign setup. Optimise agents adjust strategies as performance data changes. Improve agents monitor quality and flag issues early. Measure agents turn raw reporting into summaries and outcomes.

Users can describe issues in natural language inside the platform. The agents then respond with explanations and proposed next steps. The system records actions and decisions inside the interface for transparency and review.

Yahoo plans additional agentic features over the next year. The roadmap includes optimisation agents, quality assurance tools and advanced measurement agents. The company is also working on hybrid agents, deeper MCP integration and new APIs that support AI-to-AI and agent-to-agent collaboration across different systems.

These developments signal a broader shift in the DSP’s design. Everyday campaign tasks move towards AI-driven workflows while human teams focus on planning, creative and strategy as Yahoo extends its Agentic AI programme.