IAB Tech Lab Tackles Agentic AI in Ad Targeting

The IAB Tech Lab has released guidance on operationalizing agentic audiences and AI-driven targeting, addressing one of the most consequential shifts facing the programmatic ecosystem: the emergence of autonomous AI agents that build, negotiate, and activate audiences with minimal human intervention. For ad-tech engineers and platform leaders, the work signals where standards bodies expect targeting infrastructure to head over the next several years.

What "Agentic" Means in an Ad-Tech Context

Traditional audience targeting relies on predefined segments, deterministic rules, and human-configured campaign parameters. Agentic AI flips that model. Instead of a media buyer manually assembling a segment and setting bidding logic, an AI agent interprets a high-level objective—say, incremental reach against a purchase-intent cohort—and then autonomously constructs audiences, selects supply paths, and adjusts targeting in real time based on performance signals.

The IAB Tech Lab's framing distinguishes between AI that assists human decision-making and AI that acts on behalf of the buyer or seller. The latter introduces a new class of participant in the bidstream: software agents that make discretionary decisions previously reserved for humans. That distinction matters enormously for how transparency, accountability, and interoperability standards need to evolve.

Why This Matters for Programmatic Infrastructure

The current programmatic stack—OpenRTB bid requests, audience segment taxonomies, and supply-chain objects like sellers.json and ads.txt—was designed around relatively static, auditable data flows. Agentic targeting stresses those assumptions in several ways:

• Segment provenance: When an agent dynamically composes an audience, the origin and consent basis of the underlying data must remain traceable. Existing segment IDs and taxonomy references assume human-defined, durable segments.
• Bidstream transparency: If an agent is autonomously reshaping targeting, SSPs and exchanges need signals indicating that AI-driven logic is in play—analogous to how the supply chain object surfaces intermediaries.
• Auditability: Deterministic rules can be logged and reviewed. Probabilistic, model-driven decisions are harder to reconstruct, complicating fraud investigation, brand-safety audits, and regulatory compliance.

The IAB Tech Lab's guidance points toward the need for standardized interfaces and metadata that let agents interoperate across DSPs, SSPs, and data platforms without each vendor inventing incompatible protocols.

Interoperability and Standards Gaps

A recurring theme is that agentic systems only deliver value if they can operate across a fragmented supply chain. An AI agent optimizing a video campaign across CTV, programmatic display, and web video needs consistent audience definitions and measurement signals regardless of which SSP or ad server ultimately fulfills the impression.

This is where standards work becomes critical. Without shared taxonomies, consent frameworks, and machine-readable audience specifications, agentic targeting risks becoming a set of walled-garden implementations—each platform's agents speaking a proprietary dialect. The IAB Tech Lab is positioning itself to define the connective tissue: common data structures, transparency signals, and governance expectations that keep the open programmatic ecosystem viable against closed AI-native platforms.

Implications for CTV and Video

Video and CTV are natural proving grounds for agentic targeting. Ad pods, SSAI-delivered inventory, and the premium CPMs attached to connected TV create strong incentives for buyers to deploy AI that optimizes audience composition and frequency across fragmented viewing environments. But CTV also has thinner deterministic identity signals than web, which pushes agents toward probabilistic modeling—precisely the area where provenance and auditability concerns are most acute.

For publishers and ad-serving teams, the practical question is how their infrastructure will need to expose inventory and audience signals to external AI agents while retaining control over pricing, brand safety, and data usage. Expect this to intersect with ongoing work on sellers.json, supply-chain transparency, and consent frameworks.

The Road Ahead

Agentic audiences remain early, and the IAB Tech Lab's guidance is more a framing exercise than a finished specification. But it establishes the vocabulary and problem statements the industry will build on. For engineers and ad-ops teams, the takeaway is to begin evaluating how existing pipelines—segment ingestion, bid-request enrichment, measurement logging—would accommodate autonomous decision-makers, and to watch for the standards that will govern them.

As AI agents move from experimental buying tools to active participants in the bidstream, the difference between an interoperable open ecosystem and a set of proprietary silos will hinge on exactly the kind of standards work now underway.

Stay on top of video ad serving and programmatic. Follow Adelerate.