March 12, 2026
Inside the IAB agentic AI roadmap and new standards
by Damian Naglak
With the official rollout of the IAB Tech Lab agentic AI roadmap, the programmatic advertising ecosystem is undergoing a massive shift. Having followed the initial release and demonstrations closely, what stands out immediately is how pragmatic their approach is.
Instead of forcing the industry to adopt entirely new architectures from scratch, the IAB agentic framework wraps existing standards in MCP and Agent2Agent protocols. The industry keeps the infrastructure that already works.
What does this mean for ad tech platforms? It means:
OpenDirect → Agentic Direct
Direct deals get an agent-to-agent negotiation layer
RTB → Agentic Bid
Real-time bidding enhanced with intelligent agent decision-making
Deals API → Agentic Deals
Deal sync and management automated through AI agents
Audiences → Agentic Audiences
Embedding-based audience matching replaces taxonomy translation
The IAB Tech Lab is shipping these updates extremely fast. We are already seeing six standards getting agentic extensions, reference implementations that are demo-ready, and a dedicated agent registry to support the ecosystem.
How does the IAB agentic AI roadmap streamline programmatic workflows?
To understand the practical impact of AI agents in programmatic advertising, let's take a closer look at a buyer and seller agent use case based on the IAB's early models.
Imagine an EV launch campaign brief goes into the system. It includes budgets, flight dates, and vague audience descriptions like "eco-conscious tech-forward consumers age 25 – 54."
From there, the buyer agent takes over the heavy lifting. It automatically:
- Structures the brief and maps the vague audience descriptions directly to IAB taxonomy segments
- Queries seller agents for pricing and availability
- Gathers and processes responses from various publishers and DSPs
- Matches audience coverage across the available inventory
- Builds a comprehensive execution plan, including PG lines, PMP deals, and performance campaigns
- Asks a human for final approval, and then executes the plan
The execution phase directly hits live systems. Lines appear in GAM seconds after approval. PMP deal IDs are seamlessly issued and returned to the buyer agent, and DSP campaigns are instantly created with attached deals.
Why standardization matters
— Damian Naglak, Head of Engineering at AppliscaleThe core question for ad tech was never whether AI agents would enter the programmatic space. The real question was whether they would speak the same language when they got there.
At Bedrock Platform, my team and I have been actively building agents for supply discovery, campaign optimization, and deal troubleshooting. Because of this hands-on engineering experience, seeing the industry align on these standards matters more than any single platform's implementation.
Thanks to this new framework, it is clear that AI agents in ad tech will speak the same language.
Damian Naglak
Head of Engineering, Appliscale
Damian leads the AdTech engineering practice at Appliscale and heads engineering at Bedrock Platform. He's published a 10-part deep-dive series on agentic advertising standards.