As CES approaches, we’re seeing a surge of “agentic AI” announcements across adtech. But does the messaging match the reality? Frans Vermeulen, President of Swivel, separates what’s real from what’s aspirational.
Why are DSPs emerging as the earliest movers in agentic AI?
DSPs sit at the heart of fragmentation, complexity, and control. Campaign execution still depends on human operators managing budgets, creative rotations, flighting, and performance– often across multiple platforms and with high performance expectations. That manual overhead makes DSPs a natural proving ground for agentic AI. It’s high-volume, high-stakes work that’s ripe for intelligent automation.
The first wave is focused on productivity: reducing friction, speeding up decision making, and streamlining execution inside a single environment. Agentic AI enters as a force multiplier, operating within defined guardrails to streamline decisions and eliminate repetitive tasks. But the next phase of agentic deployments will shift toward outcome-based orchestration, where the system interprets operator intent and drives toward business goals.
The early moves we’re seeing from Big Tech like Google, Amazon, and Yahoo aren’t a reflection of unique readiness; it’s about operational urgency. With the demand behind DSPs, they’re most exposed. In that sense, DSPs are early movers because agentic AI is being deployed where it can deliver obvious, immediate ROI.
What is technically real today versus what will remain early-stage through CES?
We’re in the first or second inning of agentic AI in advertising. What’s real today are agentic workflows inside single platforms– with measurable reliability and speed. These systems are not theoretical; they’re already driving workflow acceleration where structure and scale exist.
What’s still emerging is interoperability. Agent-to-agent coordination across platforms– especially trading between buyers and sellers– remains in active testing, including through initiatives like AdCP. But those cross-platform scenarios remain early-stage and will stay in limited rollout through the beginning of next year.
But this isn’t vaporware; it’s just early. Phase one is about intra-platform automation. The next frontier is multi-agent orchestration across the ecosystem. That’s where the technical and trust barriers to adoption are highest and will still be in progress before CES.
How will standards like AdCP shape the next phase of interoperability?
AdCP is designed to do for agentic AI what ORTB did for programmatic: establish a shared language that enables systems to transact without custom integrations. Instead of brittle, bespoke APIs, AdCP defines how agents communicate so developers can build once and connect everywhere.
That is a foundational shift for the ecosystem, removing a key bottleneck to scaling agentic systems. It also creates a clean separation between how agents talk to each other and the strategies they execute, allowing for innovation without breaking compatibility.
As more participants build and deploy agents, AdCP enables interoperability at scale. Without it, every interaction becomes a custom job– but with it, the ecosystem can evolve faster and much more efficiently.
How is agentic logic already reshaping sell-side operations across major TV and streaming publishers?
Agentic logic is changing how ad ops teams work and what they work on. It’s shifting the role of operators from task execution to strategy activation. Publishers no longer spend hours triaging and troubleshooting to keep campaigns on track. That work is increasingly handled by agents, which deliver speed, consistency, and executional coverage at scale.
This has opened the door for publishers to prioritize yield and revenue strategy, experiment more frequently, and respond to market signals faster than ever before. As operational complexity increases, agentic systems are becoming the only scalable way to keep pace without overextending teams.
What publishers and platforms are actually deploying (vs. what’s just being pitched)?
The most forward-thinking sell-side teams are already using agentic systems in production to handle mission-critical work. Publishers are applying agentic logic to live campaigns, automating optimization workflows, and reducing the manual lift of managing complex workloads across multiple platforms.
This activity is live and measurable. Agents are handling tasks at a volume that exceeds what human operators can manage. In some environments, these systems are executing tens of thousands of actions per day with clear gains in efficiency, revenue, speed, and scale.
The shift is underway. While others continue to explore what’s possible, many publishers are already applying what works and scaling it in real-world conditions.
