Last week, Benchmark Company, the investment banking firm, released a new research report, with a strong focus on how agentic AI, in particular, is set to disrupt ad and marketing technology. The report is worth a read and essentially says that AI agents are likely to transform every aspect of the ecosystem, from creative development to publishing to programmatic buying.
Per the report: “Agentic represents an epic evolution in AI, a shift from reactive systems to autonomous, capable of self-learning and independent decision-making.” It goes on to suggest “extinction of digital agencies” and “legacy DSPs” (!) that fail to adapt. DSPs without “multi-layer agent bid dynamics,” it warns, will fall behind because “unlike traditional rule-based bidding algorithms, these systems analyze thousands of variables, from cross-channel engagement patterns, inventory scarcity indicators, and competitor spend trajectories, to optimize bids at millisecond intervals.”
AI, man. It’s cool and scary at the same time.
Why This Matters:
It’s an agentic world—we’re just living in it.
But, let’s be honest, the adtech industry has been slow to deliver true agentic capabilities. Viant was among the first to recognize the value of this framework for DSPs. A few other companies are building their own agentic capabilities or have released early-stage products that show promise. Still, there’s a long way to go before the industry catches up to the future Benchmark describes.
In December, EMARKETER wrote something interesting. They said: “Agentic AI has even progressed into multiagent systems, where a grouping of AI agents work together to tackle complex challenges. This approach has enabled AI systems to independently manage ongoing tasks, delegate responsibilities among themselves, and even troubleshoot issues as they arise.”
But has it really?
That feels a bit premature. Benchmark’s report makes a compelling case for what’s at stake if this evolution doesn’t happen—especially given the potential of agentic AI to reshape the market entirely.
Experts React:
Before you can reach the top floor of a building, you have to lay the foundation. The same goes for agentic AI: it needs a robust infrastructure and a reliable data spine.
Benchmark’s report highlights Model Context Protocol (MCP) as a critical part of that. MCP is an open standard that allows AI models and LLMs to connect with external data sources, tools, and services in a standardized way. Think of it as an AI plumbing system, linking inputs and outputs behind the scenes so everything flows where it needs to go.
As Benchmark puts it, MCP “will be essential in ad tech and ecommerce vectors (among many others) as it standardizes how AI models connect with diverse data sources, allowing AI agents to access real-time campaign, customer, and performance data, and make faster, smarter decisions.”
Our Take:
Right now, very few adtech companies have true agentic capabilities—and that’s understandable. But building them won’t be easy. Most companies won’t get there without fundamentally rethinking their operating models, hiring talent to ensure agents remain safe and reliable, and investing heavily in compute infrastructure. That’s why larger independent adtech players and Big Tech are better positioned to take advantage of the shift. While innovation may come from startups, enterprise-grade agentic AI will likely belong to the big players. (Is this what MSFT will do, by the way?) A sobering take but a correct one, we think.
