This week, as the industry packed for Cannes, Mediaocean's H2 report handed everyone the same headline: the fastest-growing thing on the media plan is no longer social, search, or CTV — it's ads placed inside AI assistants and agents. Three out of five marketers say they'll spend more on it this half. The debate is over. Underneath the rosé and the awards, the money is already moving toward the answer box. For marketers, that means the question shifts from "should we test ads in AI?" to "can we actually run them before the brand down the street does?"
What happened. On June 16, days before Cannes Lions, Mediaocean released its 2026 H2 Market Report, built on a May survey of 312 marketers across brands, agencies, and media companies. As PPC Land reported, 60% now plan to increase spending on AI media — ads inside AI agents and assistants — up from 54% just six months ago. That makes it the single fastest-growing investment category in the study, while 75% still rank AI as the top consumer trend to watch, the fourth straight period it's held that spot.
Why it matters. The agentic-shopping shift we've tracked all spring now has a budget line, and it's climbing faster than the channels we've leaned on for a decade. Think back to paid search in the mid-2000s: the brands that bought early keywords owned the results page for years, and the latecomers paid rent. Ads inside AI answers look like the same land grab, replayed at AI speed — early position in the answer box may buy the same durable edge. But Mediaocean names the catch, and it's a big one: an implementation gap. Wanting to spend isn't knowing how. Only about a third of marketers have moved off spreadsheets to the API-driven, direct-to-publisher buying these placements need. So the winners this half won't be the teams with the biggest AI-media budget. They'll be the ones who can actually execute it while everyone else is still writing the deck.
Meta unveiled a commerce overhaul ahead of the festival: global Live Video Ads on Facebook and Instagram, virtual-card checkout, and a creator-affiliate expansion — with 3.5 billion people already seeing AI-driven recommendations across its apps daily.
What to do: pick one upcoming livestream and map how you'd turn it into a shoppable ad before the format becomes table stakes.Heading into Cannes, NVIDIA detailed how partners are rebuilding advertising on its AI — from generative campaign production to agentic media systems. The infrastructure under the media plan is being re-poured, vendor by vendor.
What to do: ask your top ad-tech partner what already runs on an agent stack today — and what that changes for your first-party data.Even as spend plans surge, new Digiday+ research found 54% of marketers say their companies don't use agentic AI in workflows yet. The hype is running well ahead of the habit.
What to do: name one repeatable workflow you'll hand to an agent this quarter, so you're in the 46% — not watching it.Hearst Magazines launched Aura IQ this week, an agentic layer built on top of its Aura audience product, debuting publicly at Cannes. It points AI agents at three jobs publishers used to do by hand: building custom audiences, analyzing campaign performance, and answering RFPs — turning work that took hours into minutes. In internal tests across automotive, retail, CPG, fashion, and beauty, the platform doubled clickthrough rates. The lesson reaches past publishing. In a market where buying is going agentic, the moat isn't owning data — plenty of brands sit on rich first-party data and do nothing with it fast enough to matter. The moat is the agent that can act on that data while the customer is still deciding. Owning the warehouse isn't the edge anymore; owning the thing that turns it into a decision in real time is.
Three concrete moves a marketer could make. If you do nothing else from this issue, consider these.
Even a 5% test budget across ChatGPT, Gemini, and Meta's AI surfaces puts you with the 60% already moving — and buys you a learning curve before the placements get expensive. Don't wait for a perfect plan; start the meter running. By end of month.
Pick a repeatable task — RFP responses, audience builds, campaign QA — and pilot an agentic tool the way Hearst built Aura IQ. The gap between wanting AI and using it is exactly where this half gets won. By next issue.
List the customer data you own, then ask the harder question: can anything turn it into a decision in real time? If the honest answer is "a report next week," you're not ready for agentic buying yet — and now you know where to start. Before Q3.
This is not an assignment. It's for your use as you see fit — read it, skip it, forward it, file it. The Playbook is a set of options, not a checklist.
Michigan sentiment edged up to 48.9 in June but still sits 19% below a year ago, while prices climbed 4.2% — the hottest in nearly three years — so the shopper your AI media is chasing is paying more and trusting less, and the brand that wins the answer box still has to earn the heart behind it.
AI in Marketing Weekly is written by Ged King alongside the Master of Digital Marketing & AI — a program where faculty and students alike are working professionals, applying this in real jobs every week, not studying it from the sidelines.
Explore the program →