We are on the brink of an exciting top-to-bottom structural transformation of our industry, with agentic AI reshaping everything from how media is discovered to how it’s planned, bought and measured.
But this transformation won’t happen instantly and without effort.
Agentic AI needs schemas and standards to work. They provide invaluable, referenceable context, so the agents get trained to execute exactly what you asked them in natural language with repeatable accuracy.
Which means ripping out battle-tested infrastructure and starting from scratch – as some emerging agentic protocols propose – is the slowest and most painful path you can possibly imagine.
Why ditch existing schemas?
The open standards that have emerged for allowing AI agents to communicate with each other – Model Context Protocol (MCP) and Agent-to-Agent (A2A) – are fundamentally schema-driven. The schemas they rely on are the shared protocol that enables automation. Without them, agent-to-agent programmatic negotiation is impossible.
The industry can take two different paths regarding the protocols that underpin agent-to-agent communication:
- Invent entirely new schemas in the blind hope that every stakeholder instantly agrees to adopt all of it without debate. This approach has never worked in the entire history of the universe, but, hey, maybe we’ll get lucky!
- Enable instant, industrywide interoperability by using existing, fully embraced schemas, standards and related taxonomies like the IAB Tech Lab’s OpenDirect, AdCOM, OpenRTB and related schemas.
I may be biased, but which path do you think is faster, safer and more predictable?
No standards, no value
In order for agentic AI to avoid the pitfalls that have plagued programmatic’s past, we need to empower it to clean up the shortcomings of our existing ecosystem.
Every step in a typical advertising process sheds information. The brief in a planner’s head rarely maps cleanly to the targeting options inside a DSP; the nuance of a publisher’s content gets flattened in standardized inventory feeds; performance insights often trickle back into planning too slowly to matter.
This is where AI’s ability to reason across complex systems creates real value. It transforms fragmented, lossy workflows into ones that are clear, connected and explainable.
An agent with a deep understanding of publisher inventory, audience taxonomies and content context can match advertiser intent with opportunity far more precisely, because natural language interfaces can capture nuances that drop-down menus can’t. AI can surface connections between a brand’s customer segments and a publisher’s audience composition that would take humans weeks to discover.
But this value creation depends entirely on precise, deterministic standards.
Garbage in, garbage out
What happens when AI systems operate without deterministic grounding? They hallucinate.
What happens when agents orchestrate complex workflows across multiple systems? The hallucinations compound. Ambiguity turns catastrophic.
Audiences make no sense. Placements are misrepresented. Content is misclassified. Impression goals become budgets. Budgets become impression goals. If you want to create openings for fraud at scale, this is the perfect way to do it.
Agentic systems cannot be trusted unless they can use the shared definitions, transparent interfaces and enforceable governance that enable trust and accountability.
Standards, in short, are everything. When an agent says “video impression with autoplay sound-off on a news site reaching adults 25-54 interested in cooking,” every term in that phrase needs to resolve to a specific, industry-agreed definition.
Which is fortunate, because every term in that phrase is an already-defined industry standard.
Think starting over is really a faster way to get value from agentic? Maybe you’re the one who’s hallucinating.
The speed of innovation
It’s not about defending the past; it’s about speed-to-opportunity.
Using existing industry standards means tapping into compressed industry knowledge refined through billions of transactions.
AdCOM provides canonical domain objects: What is a placement? What is a video impression? What are the attributes of a device or user?
OpenRTB handles real-time bidding with battle-tested semantics.
OpenDirect manages programmatic guaranteed workflows for direct media buying.
The Ad Management API standardizes creative submission and approval workflows between buyers and sellers.
The Deals API standardizes the synchronization of deal ID metadata.
Critically, all of these share largely the same underlying object model. A video impression means the same thing whether you’re executing a real-time bid or setting up a programmatic guaranteed deal. This semantic consistency is precisely what agents need.
The IAB Tech Lab’s Agentic road map is phased, starting with foundational capabilities and expanding as the industry builds trust in agentic workflows.
We’re starting where it will generate the most economic value across the ecosystem: helping agencies and advertisers discover publisher inventory more efficiently.
As we build trust, we’ll expand to more semi-autonomous workflows. Each one will be built on the deterministic standards that are vital to agentic systems we can trust.
At IAB Tech Lab, our goal is the same as yours: We want agentic AI to happen fast. We’ll help the industry build interoperable, standards-compliant agents that work together.
But a fragmented ecosystem serves no one.
“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Follow IAB Tech Lab and AdExchanger on LinkedIn.
For more articles featuring Anthony Katsur, click here.
