Agentic Ads Are Real: What Operators Must Do Now

TL;DR: Agentic media buying, AI-answer visibility, and open web fragmentation dominated both Advertising Week New York and Jupiter Festival Miami in October 2026. DSPs are already fielding agent-driven buy requests, brand safety tools are being rebuilt for LLM answer engines, and the traditional funnel is functionally dead. Operators in high-CAC verticals need structural responses in place now — not in the next planning cycle.

Two Conferences, One Loud Signal

Advertising Week New York and the inaugural Jupiter Festival Miami ran concurrently this October, and despite the geographic split, the rooms converged on the same handful of anxieties. AI-driven media buying. Open web erosion. Cross-platform measurement that still does not work. Retail media networks metastasizing everywhere. And a growing consensus that the click — the foundational metric of digital advertising for two decades — no longer tells you much about anything that matters.

Jupiter Festival, backed by the IAB, BCG, and Art Basel parent MCH Group, positioned itself as a premium entrant on a conference circuit already crowded by Cannes Lions and Possible. The production values were there. A deep roster of entertainment, media, sports, and advertising executives filled three stages. Sessions covered creators’ influence on LLM visibility, AI-driven creative, and live sports ad innovation. Some rooms were standing-room only. Others were not. But the conversations happening off-stage — over rooftop cocktails and oysters — were where the real intelligence surfaced.

For operators spending $10K or more per month on paid acquisition, the themes that emerged are not conference ephemera. They are structural shifts that will reprice your media, change how your leads get qualified, and alter whether your brand shows up at all when a prospect asks an AI assistant for a recommendation.

The Funnel Is Dead. AI Answers Are the New Discovery Layer

The most repeated phrase across both events: “The funnel is dead, and clicks don’t matter. What matters is getting recommended in AI answers.” That is not hyperbole from a vendor slide — it was the consistent read from executives across publishing, adtech, and brand-side marketing.

BrightLocal data shared at the events quantified part of the problem. Among U.S. consumers, 52% started their most recent local business search on Google Search, and another 9% on Google Maps. But for the next search in the same journey, Google Search dropped to 19% — behind Google Maps at 24% and social at 22%. Over the full discovery journey, 30% of consumers turned to social platforms, 23% to AI tools, and 19% to review sites. Three-quarters used more than one channel before making contact.

Operators running iGaming acquisition campaigns or high-intent legal lead funnels cannot ignore this. Your prospect is not clicking one ad and converting. They are moving across Google, TikTok, Reddit, AI chat interfaces, and review platforms before they fill out a form. If your brand is absent from the AI answer layer, you are invisible for a material slice of that journey.

The challenge for generative engine optimization (GEO) tools is still unsolved at scale. Executives at Jupiter disagreed on where GEO tools should plug into the stack and which signals are actually reliable. Can synthetic prompt data capture real user intent? Probably not consistently. But the directional bet — that AI visibility is a new acquisition surface that needs its own budget line — is not seriously in dispute anymore.

Agentic Media Buying: Real Volume Is Coming, Fast

One of the most concrete disclosures from the conference circuit: at least one major adtech platform confirmed it now accepts media buys initiated by AI agents via MCP plug-ins. The volume coming through those pipes today is low. But the infrastructure is live, and the industry knows what the implication is — if an advertiser’s AI agent can autonomously find inventory, evaluate brand-safety parameters, and execute a buy, the traditional demand-side platform loses its value proposition.

This is not a distant threat to DSPs. It is already a strategy conversation happening inside every major holding company. Meta announced it is testing an AI assistant that plans campaigns, adjusts targeting and budgets, and retains memory of past campaign parameters. TikTok announced an AI shopping agent and extended its off-platform advertising reach in the U.S. These are not pilots. These are product launches with distribution behind them.

For operators running paid media across multiple channels, the practical question is: what does your buying infrastructure look like when agents start executing on your behalf? Attribution logic, bid rules, creative rotation, and audience exclusions all need to be documented in a machine-readable format — not locked inside a media buyer’s institutional knowledge. Operators who treat campaign structure as a human-only domain will be poorly positioned when agent-driven buying becomes table stakes.

What This Means for High-CAC Verticals

Operators in forex, crypto, iGaming, and legal are running some of the highest cost-per-acquisition numbers in digital advertising. A structural shift in how media is discovered, bought, and measured hits these verticals harder than most because margin for error is thin and competition for each converted lead is intense.

Here is the concrete read-through from the conference themes:

AI answer visibility is an acquisition channel now. If a prospective forex trader asks ChatGPT which broker to use, or a personal injury claimant asks an AI assistant which law firm handles trucking accidents, brand presence in that answer is a qualified lead. Forex lead generation and law firm marketing teams need to treat LLM citation as a trackable KPI, not an SEO afterthought.

Cross-platform deduplication is still broken. Every vertical covered by DIGI MIRROR runs multi-channel campaigns — Meta, TikTok, programmatic, search, affiliate. The measurement fragmentation problem discussed at both conferences is not new, but the addition of creator integrations and AI-native surfaces is compounding it. How do you value a 20-minute YouTube integration compared to a 30-second Super Bowl ad, as one industry insider put it? The same logic applies at lower stakes: how do you value an AI-referred lead versus a paid search lead when the AI touch happened three sessions before the paid click? A structured marketing audit of your attribution model is not optional when buying is this fragmented.

Brand safety infrastructure is being rebuilt for LLMs. OpenAI is already testing brand suitability controls with DoubleVerify and Integral Ad Science. If your brand appears in an AI answer next to content that violates your brand safety parameters, the existing toolkit may not flag it. Crypto acquisition teams and iGaming operators — both categories with above-average brand safety sensitivity on major platforms — should be tracking how these tools evolve into 2027.

Retail media is a distraction for most operators, but signals intent. McDonald’s, Delta, and a long list of commerce brands launched or announced media networks this fall. For performance operators, the actionable read is simpler: the businesses controlling first-party data and owned inventory are consolidating power. Build your own data assets — email lists, CRM segments, on-site behavioral data — before you need to rent them from a third-party network at a premium.

Practical Steps for Operators Before Q1 2027

Conference themes are only useful if they produce operational changes before the next conference. Here is what warrants action before Q1 planning locks:

Audit your LLM footprint. Run your brand and key product terms through ChatGPT, Perplexity, and Google’s AI Overviews. Document where you appear, what is said about you, and where competitors are cited instead. This is baseline data you need before building a GEO strategy. A dedicated precision targeting review should include this audit as a standard component.

Document your campaign architecture for machine readability. Agentic buying tools need structured inputs — audience definitions, exclusion lists, bid caps, creative specifications, brand safety rules. If that knowledge lives in a Slack thread or a media buyer’s memory, it is not ready for the agent era. Build the documentation now.

Assign an owner to cross-channel deduplication. Every operator running $10K or more per month across three or more channels has a deduplication problem. Assign someone the explicit role of reconciling cross-channel conversions on a weekly cadence. If your current analytics stack cannot do this, evaluate incrementality testing as a complement to last-touch attribution.

Test AI-qualified lead flows before you scale them. Personal AI agents are in mainstream use — Meta’s Muse reportedly has usage from demographics that would surprise most performance marketers. AI-assisted lead qualification on your inbound side is worth a controlled test before agent-driven traffic volumes make it a necessity.

The operators who leave these conferences (or read about them) and make no structural changes will find themselves repricing media from a weaker position in 12 months. The shift is not coming — it is here, and the volume is building.

Originally reported by Adweek, October 2026.