Get ready for the next wave of ad tech jargon, as terms like “containerization” and “multi-agent” become the buzzwords du jour.
And the IAB Tech Lab is leading the way. The Tech Lab, which is the technical standards operator for the IAB, introduced its Agentic RTB Framework (ARTF) v1.0 for public comment on Thursday.
The goal of the new framework is to pave the way for more comprehensive and efficient agentic media buying marketplaces in the near future, when demand and sell-side tech will integrate through new types of pipes. That’s the vision, at least.
Companies need to better “enrich, inform and analyze programmatic trading, or the bid stream itself,” Tony Katsur, CEO of the IAB Tech Lab, told AdExchanger.
“We needed to containerize the architecture,” he said, “rather than have buying endpoints” like DSPs and SSPs pinging between their own separate data centers.
Thinking outside the box
One point in favor of containerization is its battle against latency.
The current iteration of OpenRTB bid requests lack efficiency, Katsur said. Typical ad serving response times are between 400 and 600 milliseconds. “To any average person, [that] would seem really fast,” he added, but programmatic buyers and sellers are “up against a time crunch.”
The ARTF should cut down that response time by bringing agents into the marketplace via containerization.
Containerization, simply put, is the process of taking a set of code and packaging it to be easily placed in another company’s physical infrastructure. Rather than a DSP and SSP operating in separate servers that talk via API, as is typical for ad tech, one server can host the tech for both companies.
Joshua Prismon, chief architect at SSP Index Exchange, said Index likes the containerized model because it allows companies to “take a black box that somebody else does well,” like model drift analysis or deal activation, and integrate it into their own infrastructure.
AdExchanger Daily
Get our editors’ roundup delivered to your inbox every weekday.
Daily Roundup
Essentially, one company’s code becomes “an agent of the system,” said Prismon, which can be delegated wherever it’s most needed.
For instance, fraud is often caught post-bid, at the point of payment, said Katsur. That’s when fraud vendors look over the URLs and flag potential bad actors. But within the ARTF, those vendors could function as a containerized agent within, say, a DSP and more easily identify fraudulent impressions before they’re bid on.
And since “containerization is inherently secure,” said Katsur, the DSP wouldn’t have access to the measurement company’s code or software – the communication channel is open, but the black box remains closed.
Agents of change
This level of efficiency “has previously only existed in walled gardens and other single server network entities,” said Adam Heimlich, CEO of Chalice, a custom algorithm startup for ad buyers.
Nowadays, marketers in industries like finance or pharma, which have high volumes of sensitive data, or CTV sellers who might be wary of exposing show-level data in the open bidstream, have a trusted environment where that data could be stored and used for advertising purposes, but without the same privacy or consumer data security risks, Heimlich added.
And Chalice doesn’t just talk the talk; it’s been testing the containerization itself by deploying its full model within Index Exchange.
By hosting Chalice’s model, Index doesn’t have to absorb the costs, latency and mismatched IDs that happen when vendors ping back and forth between their servers. And Chalice can observe the stream of inventory within Index’s system and respond directly to bid requests, rather than sending out instructions to a DSP.
“Everybody is achieving a mutual goal based off of what they do best,” said Prismon, “and we’re tying those agents together.”
Prismon predicts that, as RTB becomes even faster and more autonomous, the ARTF is going to “build value on itself” within the ecosystem. For DSPs, SSPs and data partners, he said, this is a “much more sustainable path” than trying to standardize agentic ad buying based on one AI company.
Katsur agrees. An agentic real-time framework, he said, is the next step toward a fully agentic approach to buying and selling media.
This new efficiency, said Katsur, is creating a “rising tide for the industry.”
