In recent months, a misleading narrative has taken hold around the future of agentic advertising: Ad Context Protocol (AdCP) or the IAB Tech Lab’s Agentic Real-Time Bidding Framework (ARTF)? Which framework wins? Which one will matter?
As agentic roadmaps and execution frameworks move from theory to transactions, this framing risks obscuring what is actually happening in practice. The industry is not choosing between two competing protocols. It is navigating the emergence of multiple agentic approaches that reflect different starting points, priorities, and paths toward scale.
I’ve watched this industry move through multiple eras — the rise of programmatic, platform consolidation, debates around privacy and identity, regulatory pressure, and now the rapid emergence of autonomous systems. In each era, success was never determined by a single technology. It was determined by whether the ecosystem aligned around shared foundations while allowing innovation to thrive.
The same is true now. Agentic advertising is not converging on a single standard or framework. It is emerging through parallel initiatives that will need to coexist, even where there is overlap. PubMatic’s role is to help the industry understand how these efforts fit together in real production environments.
Why AI Demands a Shared Architecture
AI will only reshape advertising if it can operate across the entire market, not just inside isolated platforms. Autonomous agents will make decisions, optimize outcomes, negotiate value, and verify results in real time. But that intelligence cannot scale unless it is grounded in shared language, interoperable systems, and execution environments built for machine-to-machine interaction at market speed.
This is where the current debate often misses the point. The rise of agentic advertising is not a contest between protocols. It is an architectural shift that depends on layered standards, clear roles, and practical paths to execution. When discussions collapse into “either/or” choices, they obscure the fact that different agentic efforts are aiming to rapidly address the relevant problems for achieving autonomy.
Two Agentic Paths, One Industry Reality
At a high level, the industry is seeing two major agentic paths emerge.
The first is agent-native frameworks, such as AdCP, which focus on how autonomous agents express intent, constraints, permissions, and reasoning in a shared, structured way. AdCP is not a single protocol, but a family of concepts and specifications designed to ensure agents can communicate meaning consistently across systems.
The second is the IAB Tech Lab’s Agentic Roadmap, which takes a different starting point. Rather than defining new agent-native semantics from the ground up, the roadmap focuses on extending the standards the industry already relies on — OpenRTB, AdCOM, OpenDirect, GPP, TCF, and others — to support agentic execution, governance, trust, and measurement.
The Agentic RTB Framework (ARTF) is one execution-focused initiative within that broader roadmap. It addresses how autonomous agents can operate safely and predictably inside real-time transaction environments, under platform-defined controls and performance constraints. It does not replace existing transaction standards; it extends them to support agentic behavior. This list of protocols is a snapshot in time, and the IAB Tech Lab and its member community will continue to make needed enhancements where necessary.
These paths are not identical, and they are not perfectly aligned. Some overlap is inevitable — and expected — as agentic systems mature. The real question is not whether one approach replaces the other, but how platforms and implementers navigate coexistence without fragmenting the market.
Layers in the AI Stack
One useful way to make sense of this landscape is to think in layers.
The application layer is where intelligence is expressed and orchestrated — where humans and systems define objectives, constraints, and strategies. Agent-native frameworks like AdCP are deeply relevant here, providing shared semantic structure for intent and reasoning.
The transaction layer is where autonomous decisions are executed — where economic outcomes are realized inside governed environments. This is where the IAB Tech Lab’s execution initiatives, including ARTF, focus their attention, ensuring that agentic behavior can operate safely, transparently, and at scale within existing market infrastructure.
These layers intersect at execution time, even when they originate from different standard bodies. That intersection is not a flaw; it is a reflection of how complex systems evolve when a technology as transformational as AI emerges.
Lessons from Programmatic’s First Decade
Anyone who remembers the early days of programmatic will recognize the pattern. Before shared standards took hold, integrations were bespoke, fragile, and slow to evolve. Innovation stalled because fragmentation consumed engineering effort that should have gone towards improving performance, transparency, and measurement.
When the industry aligned on foundational standards, programmatic scaled. Importantly, that alignment did not eliminate competition or diversity of approaches. It created a stable base on which innovation could flourish.
Agentic advertising is at a similar inflection point.
Where We Go From Here
The practical challenge facing the industry is not selecting a single agentic framework. It is ensuring that multiple agentic initiatives can coexist and evolve in ways that support real-world execution at scale.
That requires clarity about roles, layers, and points of integration — and a willingness to engage across standard bodies rather than treating them as opposing camps. At PubMatic, our leaders sit on the boards of the IAB, IAB Tech Lab, and Prebid.org, and we were founding members of AdCP and are committed to building transparency and interoperability into the standards that define the future of agentic advertising. Our focus is on helping make that coexistence work in production, where performance, governance, and trust ultimately matter most.
Understanding the broader agentic landscape — rather than debating individual protocols in isolation — is a necessary starting point for the next phase of agentic adoption.
