In programmatic advertising, speed, accuracy, and collaboration define success. Media planners need to know what inventory is available, how much it will cost, and what reach or performance it can deliver. Traditionally, this discovery and forecasting process required multiple tools, spreadsheets, and extensive coordination between buyers and supply partners.
With the rise of Agentic AI, that workflow is changing dramatically. These agents can act as intelligent digital representatives for both buyers and sellers, communicating directly to accomplish complex planning tasks at machine speed.
Building on the First Open Spec: From Deals to Discovery
In September, PubMatic introduced the first in a series of open draft specifications defining how AI agents can collaborate across the programmatic ecosystem, focusing initially on deal management. Today, we’re releasing our second open specification, tackling the next critical challenge: inventory discovery and forecasting. Each step in this journey is about creating a common language for automation and transparency in digital advertising.
Standardizing how agents share availability, pricing, and performance predictions will enable a new level of interoperability across the ad supply chain – and this new specification provides the blueprint for making it happen.
From Manual Queries to Autonomous Collaboration
Consider a buyer agent planning a new connected TV (CTV) campaign targeting North American sports enthusiasts on a $250,000 budget.
On the supply side, a PubMatic agent manages access to publisher inventory, forecasting models, and pricing signals within the PubMatic ecosystem.
Instead of a human planner logging into dashboards or sending manual requests, the buyer simply formulates an intent such as:
“Find available CTV inventory for sports audiences in North America for Q4, with a $250K budget.”
The PubMatic agent authenticates the request, interprets the intent, queries internal forecasting APIs, and returns structured data with expected impressions, pricing, and confidence levels.
The beauty of agent-to-agent communication? It’s not static. The agents can iterate in real-time, testing different parameters until campaign goals are met – all without human intervention.
Why It Matters
- Speed: Weeks of manual coordination collapse into near real‑time insights.
- Precision: Agents draw directly from live inventory and forecasting data, ensuring accuracy and timeliness.
- Transparency: Every exchange is structured and logged, enabling auditability and strengthening partner trust.
- Efficiency: Agent‑to‑agent communication reduces friction between buyers and sellers, streamlining workflows and minimizing repetitive tasks.
- Reduced Engineering Effort: Shared schemas enable connections without custom API builds, significantly lowering integration and maintenance costs.
- Automation‑Ready: Once standardized, these interactions can connect seamlessly to campaign planning, optimization, and automated deal creation.
Design and Governance Considerations
Building such a system requires more than intelligent agents. It requires shared standards, mutual trust, and governance. Common schemas and ontologies ensure every agent interprets terms such as “audience,” “impressions,” and “forecast window” in the same way.
Security and trust are equally critical. Agents must authenticate each exchange, access only approved datasets, and leave transparent audit trails. Embedding human‑in‑the‑loop controls maintains accountability when final plans or budgets are committed.
This work aligns with the Model Context Protocol (MCP) and Ad Context Protocol (AdCP) specifications, which provide a framework for secure, interoperable agent‑to‑agent communication at scale.
The Road Ahead
As these capabilities mature, we can imagine continuous learning systems where buyer and seller agents optimize not only for CPM and reach, but also for sustainability metrics, brand alignment, or audience diversity.
By defining open standards for how agents communicate, PubMatic is laying the groundwork for a more transparent, efficient, and collaborative ecosystem, where discovery, forecasting, and deal management flow seamlessly between intelligent systems. This isn’t a distant future – this is happening now. PubMatic is charting the course for how the industry will get there.
Ready to explore the next frontier of programmatic collaboration? Review our new open draft specification for inventory discovery and forecasting and help shape how intelligent agents will define the future of media buying.