The tech world left Las Vegas buzzing about agentic AI. Yet the people who actually place the ads—the media buyers—aren’t sprinting to hand the keys to autonomous systems. They’re cautious, and they’re right. My view is simple: agentic AI will shape media buying, but it should not drive it—at least not yet.
This matters because budgets are at stake. Real money, brand safety, and client trust are on the line. Chasing the new thing without proof can burn cash and reputations.
“CES 2026 was all about agentic AI, but media buyers are approaching the hype around autonomous media buying with pragmatism over urgency.”
The Core Argument: Hype Meets Hard Reality
Agentic AI systems promise to plan, place, and optimize ads without human hands. That sounds efficient. It also sounds risky. Autonomy without accountability is not a plan.
Media buyers I spoke with aren’t anti-AI. They’re anti-fantasy. They want tools that help, not bots that guess. They want speed, but not at the cost of control. Pragmatism beats panic.
The mood on the floor was clear: test, verify, then scale. That is not fear. That is discipline.
What Agentic AI Gets Right—and Wrong
AI can spot patterns faster than humans. It can run endless tests, crunch prices, and adjust bids by the second. Those are wins. But the “agent” idea—AI acting on its own—raises issues that a demo cannot solve.
- Accountability: Who answers when an autonomous buy funds junk sites?
- Bias and brand safety: Models can optimize for cheap clicks and ignore context.
- Measurement drift: AI can chase the metric, not the goal.
- Supply quality: Without human checks, bad inventory slips in.
- Regulatory risk: Privacy rules change faster than models retrain.
These are not abstract worries. They are the same problems we fight in programmatic today—only now with fewer manual brakes.
Pragmatism Is a Strategy, Not a Stall
I left believing the smartest approach is human-led, AI-assisted buying. Let AI do the grunt work: forecasting, creative rotation, and bid optimization within clear rules. Let humans set goals, guardrails, and ethics. Keep the pilot in the cockpit.
Some argue that moving slow means losing ground. I disagree. Moving smart beats moving first. The teams that win will define tight control loops: small budgets, clear success measures, weekly audits, and hard stop-loss rules. That is how you learn without bleeding.
Addressing the Counterpoint
Yes, competitors will claim full autonomy and lower costs. They might even show short-term gains. But if those gains come from cheap reach, shady placements, or fragile models, the bill arrives later. Saving a dollar today to lose a client tomorrow is not a victory.
What To Do Now
If you buy media, act like an adult in a toy store: curious, not careless.
- Run controlled trials with strict guardrails and human approval on major decisions.
- Demand audit logs: every action, every source, every spend path.
- Set brand safety as a hard constraint, not a suggestion.
- Measure business outcomes, not just click metrics.
- Use AI as co-pilot features first—forecasting, QA, anomaly alerts.
- Share learnings across teams and pause anything you cannot explain.
Each step builds muscle. You learn what works without betting the house.
Where This Is Headed
Agentic AI will get better. When these systems can prove transparency, control, and repeatable performance, spend will follow. But trust must be earned, not assumed. The future of media buying is not man versus machine. It is smart teams using sharp tools under clear rules.
My conclusion is blunt: don’t outsource judgment. Keep humans accountable for outcomes, and use AI to make them faster and sharper. That balance will protect budgets now and deliver gains later.
If you care about results, push vendors for proof, not promises. Set your standards, build your guardrails, and invite AI in—on your terms. The goal isn’t autonomous buying. The goal is better buying.
