• by Laurie Sullivan , Yesterday

The Interactive Advertising Bureau (IAB) has released a whitepaper exploring autonomous AI systems that have begun to reshape how video is planned, optimized, measured, and created.

"AI-Powered Video Outcomes: Agentic AI" outlines the future of performance and tools such as standardized measurement, clear guardrails for autonomous execution, and greater transparency into how AI-driven decisions are made.

Autonomous decisions, measurement and attribution, creative, and privacy are all aimed at helping move agentic AI from experimental to operational.

Although some forms of AI have dominated projects for years, agentic has just begun to make an impact because it can monitor and change performance in real-time.

"Agentic AI is already beginning to reshape parts of the ecosystem today, particularly in optimization and decisioning, but a fully real time, end-to-end autonomous ads marketplace is likely not going to happen anytime soon in order to scale," said Jamie Finstein, IAB vice president of Media Center. "The timeline isn’t just about the technology, it’s about the infrastructure catching up, including standardized data, interoperability across platforms, and clear governance."

Finstein told MediaPost that until those pieces are in place, the industry "won’t flip overnight," but the IAB sees a rapid, phased shift where more of the AI advertising life cycle will be created, bought and optimized.

The IAB report also emphasized that the infrastructure is still being built, and true autonomous media processes are an estimated to be a year or two away.

One of the biggest changes, according to the report, is with measurement -- which will become active, not passive. Nearly 70% of analytics teams have already begun to scale AI. Without transparency into how decisions are made, trust in outcomes is at risk.

Measurement and attribution continue to shift video from static, retrospective reporting to a dynamic and continuously “learning system” that uses and compares data and methods such as multitouch attribution, marketing mix modeling (MMM), incrementality testing, and clean room analysis across digital video environments.

The whitepaper released today points to the IAB State of Data 2026: The AI Powered Measurement Transformation, where it states 69% of analytics teams are now integrating and scaling AI processes, leading adoption across agencies, and 44% are actively implementing agent-based platforms.

AI can validate signals, normalize disparate data inputs, flag inconsistencies, and reduce reporting discrepancies across connected TV (CTV), online video (OLV), social and free ad-supported streaming television (FAST).

As AI continues to move from a media that recommends executing tasks, it shifts control of planning, buying, and optimization from humans to machines.

Rather than rely on static rules or changes, AI agents reevaluate tasks and performance signals like attention, reach, frequency, and outcomes within milliseconds of making real-time decisions. Changes that optimize for performance are done to meet goals across CTV, streaming, and digital video.

Scaled full automation is still a ways off, and qwcd vthe optimization layers being implemented by agencies and demand-side platforms (DSPs) can automatically optimize toward key performance indicators such as conversions, CPA, and return on investments, according to the whitepaper.

Media buys, performance, and creation in the future will become autonomous across a fragmented digital video supply chain, according to the IAB, which is why media budgets will shift faster and outcomes will become more performance driven.

As the IAB states, these agents will make decisions, analyze options, and actively make choices before ultimately taking actions, with preset parameters and guardrails.

Advertising agents that sell space, time, and media will interpret briefs, determine video strategies, select inventory, and generate full media plans without much human input, making decisions that will influence creative, audience intent, context, and measurement goals -- all while taking privacy and regulations into account.