The latest Prebid webinar on May 14th explored a shift that feels similar to the early days of header bidding, but with bigger implications: the rise of agentic systems in programmatic advertising, and how Prebid is positioning itself to connect both sides of the ecosystem.
The webinar focused on one idea: enabling better coordination between seller-side and buyer-side agents, without sacrificing transparency, control, or interoperability. The session was joined by not only publishers (sell-side) but also buyers for the most holistic approach.
If you missed the webinar, you can watch the recording here.
Prebid Agentic PMC Update & Roadmap
Speakers: John Rosendahl (Director of Product Management @ Optable)
Olha Paramonova (VP Ad Tech @ Sigma Software)
Prebid is leaning hard into agentic systems, AI-driven agents acting on behalf of buyers and sellers, as the next major shift in programmatic.
To get ahead of this, Prebid launched an Agentic Working Group and took ownership of the seller agent within the AdCP protocol. The idea is to make sure publishers have representation in a world where machines increasingly make decisions.
Become a Prebid member to join the PMCs & contribute to the marketplace!
From RTB to Agentic Interactions
Unlike OpenRTB’s fragmented, bid-by-bid transactions, agentic protocols enable:
- Direct interaction between buyer and seller agents
- More holistic, outcome-driven negotiation
What the Seller Agent Means for Publishers
- Better communicate the value of their inventory
- Move away from commoditised impressions
- Participate in more direct, intelligent transactions
Where Prebid Agentic Development Stands
Building the Core Protocol Layer
Most of the work right now is happening at the protocol level, specifically around AdCP.
The priority is to:
- Define how buyer and seller agents communicate
- Standardize requests, responses, and negotiation logic
- Create a shared “language” for agent interactions
Seller Agent Development
Prebid is actively developing the seller agent, which will:
- Represent publisher inventory in agentic environments
- Translate publisher data into something machines can act on
- Act as the interface between supply and buyer agents
Early Testing & Limited Scale
Adoption is still in early testing phases, with:
- Initial spend in the low six-figure range
- Gradual ramp expected through the year
Focus on Interoperability
A big part of the roadmap is ensuring:
- Different agents can work together
- No single platform controls the ecosystem
What We Don’t Have Yet
- Mature publisher tooling
- Standardized way to package inventory for agents
- Widely adopted workflows
Buyer Perspective
Speaker: Ingmar Zach (President, Product Data & Intelligence @ WPP)
From the buy-side, the conversation focused on efficiency and outcomes.
Buyers are increasingly relying on automation to:
- Optimize campaigns toward CPA or ROAS goals
- Navigate fragmented supply paths
- React to performance signals faster than humans can
However, many current solutions, especially those from large platforms, operate as closed systems. While they deliver results, they limit visibility into how decisions are made.
Ingmar highlighted the need for:
- Better access to structured, high-quality supply signals
- More transparency in auction dynamics
- Open frameworks that allow flexibility in how buying strategies are executed
Case Study / Marketplace Insights
Speaker: Brian O’Kelley (Founder & CEO @ Scope3)
The key takeaway was that misalignment between buyers and sellers is still a major inefficiency in programmatic.
Examples discussed showed:
- Sellers optimizing for yield without visibility into buyer goals
- Buyers optimizing for performance without understanding supply constraints
- Missed opportunities due to a lack of shared signals
Agentic systems offer a way to fix this by enabling continuous feedback loops.
For instance:
- Seller-side logic can adapt floors or packaging based on demand behaviour
- Buyer-side systems can respond dynamically to changes in supply quality or pricing
But again, this only works if both sides can interact through a shared framework like Prebid.
Q&As
- How are you thinking about how publishers represent their audiences agentically? Buyers today already have a high level of skepticism about the scale & composition of audiences during real-world sales pitches?
Buyers are right to be skeptical. However, we’re building the feedback loops into the protocol so they can measure performance in real-time. In the current protocol, there’s already governance around how a publisher represents its audiences, and how that gets validated. - In addition to WPP, are there other agencies currently involved? Do you have any information on what the current budget spending is by agencies?
Yes, multiple agencies and brands are already testing agentic setups, though not all are publicly disclosed yet. Spend is still early but ramping quickly, moving from low six figures toward potentially massive scale. The expectation is exponential growth rather than gradual adoption. - What happens when GAM builds its agent?
It’s not necessarily a threat. If major platforms like Google build agents, they simply become part of a broader ecosystem. The key difference with agentic is that it’s not as tightly constrained as today’s programmatic pipes, so multiple agents can coexist rather than locking everything into one system. - Is the intent to make Prebid Seller Agent compatible with other standards, such as IAB Tech Lab?
Maybe. Prebid aims to:
- Stay interoperable with existing standards like those from IAB Tech Lab
- While also developing new frameworks (like AdCP) that better support agentic interactions
The idea is to work alongside current standards where possible, but not be limited by frameworks that weren’t built for agent-to-agent communication.
- The examples speak of agents as campaign coordinators, bridges between buyers & sellers, but I’m curious how we think the agentic framework impact will materialize at the transaction level. In other words, what are the anticipated impacts on real-time bidding? Will we still be in a world where ad opportunities are auctioned off in real time, or will buyer/seller agents constitute the new basis for direct selling?
RTB isn’t going away. Agentic will sit alongside it as another transaction model. Think of it like waterfall still existing today. Agentic opens up more direct, outcome-driven interactions, but auctions will continue to play a role. - Where do you see the biggest operational or infrastructure gaps for publishers right now, especially as agentic buying and AI-driven decisioning start entering the ecosystem?
Publishers are not structurally ready. Their systems are built for human sales, not machine-to-machine communication. The biggest gap is translating internal knowledge, like audience, context, and value, into structured data that agents can understand and act on. - Is the business model of seller agents providers mainly a SaaS model, rev share or tech fee?
Nothing is fully defined yet, but expectations lean toward a SaaS-style model, similar to existing Prebid ecosystems. Core infrastructure will likely stay open, with monetization happening through services, tooling, and added layers on top. - In regards to provenance, do you guys see this becoming a standard signal in the bid request and what benefits do you think it can present to the ecosystem?
Yes, provenance is expected to become more important. It helps establish trust and transparency around where inventory comes from, which is critical in an agentic world where decisions are automated and need reliable inputs. - There’s a number of non-technical aspects of Agentic that need to be figured out like trust models, reputation, anti-fraud, privacy, etc – all the things that AdTech has had on its plate since the beginning, which are presumably going to become even more critical with AI. Are these aspects going to be looked at by this group?
Yes. A lot of it will be the output from AAO and Prebid will build against that.