• by Laurie Sullivan , Staff Writer, 10 hours ago

There is an ongoing debate about whether advertisers are ready for agentic agents to take complete control of advertising platforms -- and even rewrite them in real time with the ability to create, build and serve personalized ads to consumers.

This is similar to the way Google Gemini or Adobe Firefly draw creative in the moment from prompts.

In April, Adobe announced several agentic ad tools at its Summit conference that are specifically designed to automate and personalize advertising and digital marketing workflows. 

"CX Enterprise" can execute entire campaigns based on one goal provided by the advertiser, and "CX Enterprise Coworker" in the CX Enterprise suite coordinates multiple other agents to carry out complex marketing workflows. 

"Adobe Brand Intelligence" is a new tool that ensures all AI-generated advertising content remains on-brand. It continuously learns from real-world feedback like campaign approvals or rejections to evolve with the brand.

All these tools use AI agents to manage advertising workflows, but Greg Collison, Adobe head of product and design, told MediaPost that the complete vision described above requires many changes and challenges, as well as a lot of trust that advertisers are not ready to give.

“Before AI agents can create and serve ads in real time it will require advertisers to trust technology to approve and serve ads as it creates them,” Collison told MediaPost, adding: “There must be quality control, and the industry is not yet ready for something like that.”

Automated reviews that score ads based on a brand's requirements will need to mature, Collinson said.

Programmatic ad assets are built in milliseconds to fit into the timeline of getting the page loaded and the ad being served, he explained.

A digital twin is also required, he said, adding that Adobe is working on this.

Brands may not need 10 times more pieces of creative that are required now. “There’s a need for the image of pants at Banana Republic, for example, to be an exact representation of its products,” he said.

"Remy" -- a codename for an AI personal agent inside the Gemini app -- is being tested inside Google to take actions on behalf of humans.

The AI assistant can keep track of events, proactively handle complex tasks, and learn the preferences of a human over time.

It should not come as a surprise to anyone that Google employees have been testing the new AI agent, which "runs in a staff-only version of Google's Gemini app and can integrate with a range of Google's other services," according to internal document seen by Business Insider and two people familiar with the situation. After all, that is Google's goal -- to create an AI agent to support all of the company's apps in the hands of humans.

The timeline for a public launch remains unclear, but Google Marketing Live takes place later this month, and hopefully the company will demonstrate the agent. 

Transitioning from being a provider of information to an enabler of action is Google’s ultimate goal for AI agents as the internet transitions to an agent-first web.

AI acts as a collaborative partner capable of performing complex, multi-step tasks on a human’s behalf.

AI agents act as an agentic expert in the advertising ecosystem, capable of handling operational complexity while advertisers focus on brand strategy. Agentic workflows will become the backbone of advertising platforms.

A key part of this strategy is the Agent2Agent (A2A) Protocol, Google's open standard designed to ensure that agents from different platforms can communicate, creates a multiagent ecosystem where agents based on security, coding, or data analysis can collaborate to solve broader challenges.

"Agents is DeepMind's heritage," Demis Hassabis, CEO of Google DeepMind, said during an interview posted to LinkedIn. Projects such as Atari game research and AlphaGo are examples of agent systems.

DeepMind created AlphaGo and trained the computer to play Go on human knowledge, but the ultimate successor in the lineage became MuZero, which can master games without being told the rules beforehand.

Agents will discover the answers without humans telling them what is required or needed. This may not seem possible yet, but that’s the AGI moment Hassabis has spoken about in the past.