IAB Tech Lab unveils Agentic RTB Framework to boost real-time ad trading efficiency | MarTech
Briefly

IAB Tech Lab unveils Agentic RTB Framework to boost real-time ad trading efficiency | MarTech
IAB Tech Lab released the Agentic RTB Framework (ARTF) v1.0 as an open, technical specification to accelerate real-time bidding. ARTF standardizes agentic interoperability within shared data center environments and enables containerized, low-latency interactions among DSPs, SSPs, and ancillary services. The framework claims to reduce bid request-response round-trip times from roughly 600–800ms to about 100ms, enabling richer bid enrichment, more responsive decision-making, and potential energy savings from lower computational overhead. ARTF also supports protocols such as the Model Context Protocol (MCP) and agent-to-agent communication. The spec is open for contributions on GitHub, with public comments accepted through January 15, 2026.
"The IAB Tech Lab has released the Agentic RTB Framework (ARTF) v1.0 for public comment, introducing a new technical specification designed to accelerate and streamline real-time bidding in digital advertising. The framework standardizes agentic interoperability within shared data center environments, enabling faster, containerized interactions between demand-side platforms (DSPs), supply-side platforms (SSPs), and other services. According to the Tech Lab, ARTF can reduce bid request-response latency by up to 80%, cutting round-trip times from 600-800ms to approximately 100ms."
""This framework lays the foundation for interoperable, high-performance, agentic workflows," said IAB Tech Lab CEO Anthony Katsur, positioning the release as a key milestone toward a modernized, virtualized programmatic ecosystem. Industry stakeholders are already weighing in. Index Exchange Chief Architect Joshua Prismon called the spec "production-tested" and "transformational," while Chalice CEO Adam Heimlich said the standard could redefine AI-driven optimization in programmatic media."
Read at MarTech
Unable to calculate read time
[
|
]