Agentic AI: The New Operating System for Ad Spend

The era of AI as a creative assistant is effectively over. As of January 2026, we have entered the era of AI as an operator. The announcements from WPP, Omnicom, and PubMatic this month signal a fundamental shift in how capital is deployed in the advertising ecosystem. We are moving from tools that help humans work faster to agents that do the work themselves.

For founders and CMOs, this distinction is critical. We are no longer talking about generative tools that draft copy or resize images. We are talking about "Agentic AI"—autonomous software capable of planning campaigns, executing media buys across fragmented channels, and optimizing for ROI without human intervention. The holding companies are betting their futures on this, and the infrastructure players are building the protocols to support it.

This is not a futuristic concept. It is a commercial reality that impacts your unit economics today. If your marketing strategy still relies on siloed teams manually pulling levers in Facebook Ads Manager or Google Ads, you are already operating at a speed disadvantage.

From Optimization to Autonomy

The headline news is that WPP and Omnicom are rolling out agentic AI offerings to automate the entire advertising workflow. Simultaneously, PubMatic has launched an "agentic operating system" designed to solve programmatic pain points. But the real story lies in the "Ad Context Protocol" and Google’s "Universal Commerce Protocol."

These developments mean that advertising is moving toward a standardized, machine-to-machine language. Previously, a media buyer looked at data and made a decision. Now, an AI agent perceives the market context, negotiates the bid, and executes the transaction based on a set of governing parameters. The IAB’s involvement suggests a push for standardization to prevent the walled gardens from becoming completely impenetrable black boxes.

Commercially, this flattens the execution layer of marketing. The technical skill of "media buying"—knowing which buttons to click—is being depreciated to zero. The value shifts entirely to strategy, creative inputs, and data governance. If an agent can execute the buy more efficiently than a human, the human’s role becomes defining the objective and auditing the machine.

The Death of Channel-Based Allocation

Gartner forecasts that by 2028, over 60% of brands will use agentic AI for one-to-one marketing, effectively ending channel-based approaches. This aligns with what we are seeing in the market right now. The concept of a "Facebook Specialist" or a "Search Specialist" is becoming obsolete.

Agentic AI does not care about channels; it cares about liquidity and outcomes. If the Universal Commerce Protocol allows an AI agent to execute a purchase for a consumer across any retailer, the ad agent’s job is simply to be present where that transaction probability is highest. This dissolves the artificial barriers between social, search, and programmatic display.

For growth leads, this requires a change in budgeting. You cannot allocate budget to channels if the agents are dynamically moving capital in real-time. You must allocate budget to outcomes and audiences. The friction of moving money between platforms is disappearing, which means the market efficiency—and competition—will increase drastically.

The Protocol War: Platforms vs. Independence

There is a conflict brewing between the major platforms and the open web. Google and Apple are integrating LLMs directly into their operating systems and shopping interfaces. They want their agents to control the entire journey from intent to purchase. In contrast, the PubMatic and IAB initiatives represent the "open" web's attempt to create a counter-weight—a way for independent brands and publishers to transact without paying a toll to the duopoly.

This creates a dilemma for brands. Relying solely on platform-native agents (like Google’s customized shopping ads in AI Mode) offers ease of use but risks total loss of data sovereignty. Utilizing independent agentic protocols offers more control but requires a more sophisticated technical stack.

The IAB has rightly warned that this transition will involve "years of market experimentation." We are in the messy middle. The technology promises efficiency, but the infrastructure is still being laid. The risk of "false starts" is high, meaning early adopters might waste budget on agents that hallucinate or misinterpret commercial signals.

Aragil POV: Navigating the Shift

If a client approached us today asking how to adapt to this agentic shift, our advice would be pragmatic and defensive.

First, we would audit the data infrastructure. Agentic AI is only as good as the signal it receives. If your conversion data, product feeds, and customer profiles are messy, the agent will scale your inefficiencies. We are moving from "garbage in, garbage out" to "garbage in, rapid automated cash burn." Data governance is no longer an IT issue; it is a media performance issue.

Second, we would focus on "AI-Parseable Signals." We need to ensure that your brand’s content and product details are structured in a way that AI agents can easily read and interpret. As search shifts to answer engines and shopping shifts to automated agents, your SEO strategy must evolve from keyword stuffing to entity optimization. The goal is to be the obvious answer for the machine.

Third, we would maintain a healthy skepticism of "set it and forget it" promises. The shift to ROI measurement discussed at Davos indicates that CFOs are demanding accountability. We would use agentic tools for execution but retain strict human oversight on the strategic parameters. We do not trust a black box with the entire budget until it proves it can beat our baseline benchmarks.

Most teams will make the mistake of either ignoring this (and getting outpaced) or trusting it blindly (and losing control of their unit economics). The winning path is to view agentic AI as a high-performance employee that needs clear KPIs and constant supervision.

The Bottom Line

The launch of agentic AI by the major holding companies and tech platforms marks the industrialization of digital marketing. We are trading manual control for algorithmic speed. The winners in this cycle will not be the best media buyers, but the best data architects. Your job is no longer to buy ads; it is to program the machine that buys them for you.