AI’s rapid transformation of the media and marketing sector has reached a new inflection point, as agentic technologies and integrated generative tools move from experimentation to deployment, promising sweeping changes—and raising equally sweeping questions—about creativity, workflow, and even the future of content production itself.
In the past year, the discourse around artificial intelligence in the media and marketing world has shifted from “if” to “how fast.” The latest surges in adoption—driven by both platform giants and established brands—signal a maturation nobody in the industry can afford to ignore. AI is now considered a core operational technology, not a side project. It is no longer about flashy demos or headline-grabbing creative stunts; the proliferation of agent-enabled workflows and data-driven content generation is truly remaking how media gets made, delivered, and personalized.
Research published by Marketing Week, Kantar, and Google revealed a striking milestone: over half of marketers now report using AI to deliver campaign creative. This includes deploying machine learning to generate asset variations, personalize messages at scale, and pre-test creative effectiveness before launch. Crucially, the adoption rate is even higher in B2C sectors than B2B, underscoring AI’s central role in consumer-facing campaigns. Cross-verification with industry surveys from Deloitte and the IPA corroborates the trend: AI-powered automation and content personalization are at the heart of modern marketing practice.
According to Unilever executives, the goal is nothing less than “tripling the velocity” of creative output for digital channels. Sketch Pro enables rapid iterations and fine-tuned variants, purportedly producing effective campaign content up to three times faster than previous workflows. This comes as global brands confront a fractured content landscape, where traditional broadcast is no longer the default and nimble, data-responsive messaging is key to reaching audiences wherever they congregate online.
While Unilever’s claims are plausible—given the demonstrated efficiency of AI models like Veo 3 in generating video and imagery—it is true that such advances foreshadow disruption not just for marketers, but for the broader production sector. Agencies and vendors accustomed to lengthy shoot schedules or intricate post-production pipelines may now find themselves at existential risk. Industry figures have cautioned that “just-in-time content creation” could compress budgets and job opportunities, a concern reinforced by public statements from creative labor unions.
This development takes Google’s legacy of surfacing information into a new operational frontier. By deputizing AI as a proactive agent, Google aims to help users “get more done,” with early availability restricted to subscribers in the US and enhanced access for AI Pro and Ultra users. While in many ways this resembles prior experiments (like Google Duplex), the direct integration into core search and commercial tasks signals a blurring of boundaries: search engines are no longer just about finding information; they are becoming brokers and agents for user intent.
The risk, of course, is twofold. On one hand, customer service work—already vulnerable to automation—faces renewed pressure, as AI agents outcompete human intermediaries for routine queries. On the other, business owners must contend with AI acting as a gatekeeper for key customer data, potentially altering how customer relationships are managed, especially as more commerce and service discovery flows through these automated channels.
Luke Bradley-Jones, The Economist’s president, frames this as making expert analysis more accessible to a digital-first audience without compromising on editorial independence. Still, the move suggests a calculated gamble: that premium, trusted IP, when coupled with AI-powered content delivery, can command engagement and loyalty even amid a torrent of generative “content” from less reputable sources.
Other major outlets—including The New York Times and Financial Times—are reportedly exploring similar arrangements, often with a strong eye on protecting copyright while simultaneously unlocking new revenue streams via AI-enabled subscriptions. The outcome of these early partnerships could presage how much of the world’s premium knowledge base filters through AI models—and who ultimately profits from that dynamic.
As reported by Bloomberg and confirmed by multiple analytics firms, this signals a persistent challenge: even “free” or enterprise-licensed AI doesn’t guarantee adoption. Microsoft’s bundling with the Office suite provided a powerful distribution channel. However, user perceptions of accuracy, utility, and overall experience appear to lag behind OpenAI’s more widely hyped (and sometimes more “conversationally flexible”) offering.
Furthermore, enterprise buyers—long Microsoft’s core audience—have voiced concerns about Copilot’s integration with legacy systems, limitations in supported languages, and the need for robust privacy controls. While Microsoft has signaled ongoing investment in improving Copilot’s functionality and reach, the company’s relative underperformance is a case study in how speed to market does not always equate to product-market fit, particularly in fast-evolving technology segments where user trust is paramount.
OpenAI claims that combining the capabilities of ChatGPT, “Operator,” and deep research allows its AI to “do work for you”—from summarizing emails to automating business workflows via integration with services such as Hubspot. The implications are formidable: rather than “AI as widget,” users are now experimenting with full-stack, workflow-driven automation. Early tests (accessible to Pro subscribers, with broader rollout pending) show that even technically non-savvy users can create powerful automations—suggesting a significant democratization of what was formerly the domain of IT and data engineering teams.
Yet, there are unavoidable caveats. Concerns about data security, consistency of agentic performance, and the possibility of runaway tasks (sometimes colloquially termed “AI hallucinations”) remain. OpenAI asserts that rigorous safeguards are being put in place, but independent audits and public transparency about the actual rate of AI errors or misfires are still needed before skeptical organizations will hand over mission-critical operations.
Yet, there is an undercurrent of caution as well. Share price improvements may reflect hopes that AI will drive efficiencies and new profit pools, but also contain anxieties about margin compression and competitive upheaval if legacy models fail to adapt. Business leaders now confront a dual imperative: harness AI to fuel growth or risk obsolescence as new entrants and agile incumbents accelerate their digital reinventions.
AI’s ability to generate, test, and refine vast numbers of content variations in real time is a game-changer for organizations tasked with personalization at scale. In the competitive, always-online world of retail and digital media, this is not just beneficial; it is essential.
To thrive, organizations—whether brand marketers, publishers, or technology providers—will need to invest not only in AI tools, but in the training and creative processes that allow humans to steer, query, and supervise their increasingly agentic partners. Ongoing research and rigorous experimentation, coupled with open dialogue about failures as well as successes, will be required to ensure that AI-driven transformation serves the broader interests of audiences, creators, and the public.
As industry leaders look forward to the next wave of conferences and market milestones—from Unmade’s REmade and Unlock events in retail and nightlife, to end-of-year indicator roadshows—it is apparent that AI will be at the center of every strategic discussion. Whether as accelerant, disruptor, or simply the new digital substrate, artificial intelligence is here to stay—arriving not as a helper, but as a formidable workforce in its own right. The challenge now is making sure we manage the agents before they manage us.
Source: unmade.media This week in AI: The agents arrive
In the past year, the discourse around artificial intelligence in the media and marketing world has shifted from “if” to “how fast.” The latest surges in adoption—driven by both platform giants and established brands—signal a maturation nobody in the industry can afford to ignore. AI is now considered a core operational technology, not a side project. It is no longer about flashy demos or headline-grabbing creative stunts; the proliferation of agent-enabled workflows and data-driven content generation is truly remaking how media gets made, delivered, and personalized.Research published by Marketing Week, Kantar, and Google revealed a striking milestone: over half of marketers now report using AI to deliver campaign creative. This includes deploying machine learning to generate asset variations, personalize messages at scale, and pre-test creative effectiveness before launch. Crucially, the adoption rate is even higher in B2C sectors than B2B, underscoring AI’s central role in consumer-facing campaigns. Cross-verification with industry surveys from Deloitte and the IPA corroborates the trend: AI-powered automation and content personalization are at the heart of modern marketing practice.
Unilever’s Sketch Pro: AI and the New Production Paradigm
One of the highest-profile examples of this transformation is Unilever’s launch of Sketch Pro, an AI-driven design and production unit created with IPG Studios. This tool, powered in part by Google’s Veo 3, signifies a fundamental departure from television-first strategies. Instead, it prioritizes “social-first” content—dynamic, platform-native creative at dramatically increased speed.According to Unilever executives, the goal is nothing less than “tripling the velocity” of creative output for digital channels. Sketch Pro enables rapid iterations and fine-tuned variants, purportedly producing effective campaign content up to three times faster than previous workflows. This comes as global brands confront a fractured content landscape, where traditional broadcast is no longer the default and nimble, data-responsive messaging is key to reaching audiences wherever they congregate online.
While Unilever’s claims are plausible—given the demonstrated efficiency of AI models like Veo 3 in generating video and imagery—it is true that such advances foreshadow disruption not just for marketers, but for the broader production sector. Agencies and vendors accustomed to lengthy shoot schedules or intricate post-production pipelines may now find themselves at existential risk. Industry figures have cautioned that “just-in-time content creation” could compress budgets and job opportunities, a concern reinforced by public statements from creative labor unions.
From Search to Service: Google’s Agentic Leap
The growing accessibility of agent-like AI in daily workflows also reflects Google’s recent moves. The search behemoth announced a new feature that allows users to let AI make calls on their behalf: scheduling appointments, canvassing local businesses for information, even gathering competitive pricing—all without the user having to pick up the phone.This development takes Google’s legacy of surfacing information into a new operational frontier. By deputizing AI as a proactive agent, Google aims to help users “get more done,” with early availability restricted to subscribers in the US and enhanced access for AI Pro and Ultra users. While in many ways this resembles prior experiments (like Google Duplex), the direct integration into core search and commercial tasks signals a blurring of boundaries: search engines are no longer just about finding information; they are becoming brokers and agents for user intent.
The risk, of course, is twofold. On one hand, customer service work—already vulnerable to automation—faces renewed pressure, as AI agents outcompete human intermediaries for routine queries. On the other, business owners must contend with AI acting as a gatekeeper for key customer data, potentially altering how customer relationships are managed, especially as more commerce and service discovery flows through these automated channels.
The Economist & NotebookLM: Editorial IP Meets AI Distribution
Not to be outdone, premium publishers are also experimenting with ways to blend their painstakingly crafted editorial content into the AI era. The Economist, long a byword for rigorous analysis, has partnered with Google AI to make its flagship annual report, The World Ahead, available on NotebookLM. This tool, still in its early stages, enables users to interactively explore major trends and predictions—drawing on The Economist’s research to fuel “forward-looking journalism.”Luke Bradley-Jones, The Economist’s president, frames this as making expert analysis more accessible to a digital-first audience without compromising on editorial independence. Still, the move suggests a calculated gamble: that premium, trusted IP, when coupled with AI-powered content delivery, can command engagement and loyalty even amid a torrent of generative “content” from less reputable sources.
Other major outlets—including The New York Times and Financial Times—are reportedly exploring similar arrangements, often with a strong eye on protecting copyright while simultaneously unlocking new revenue streams via AI-enabled subscriptions. The outcome of these early partnerships could presage how much of the world’s premium knowledge base filters through AI models—and who ultimately profits from that dynamic.
Microsoft Copilot: Traction Troubles in the Enterprise AI Arms Race
Microsoft’s Copilot, launched to considerable fanfare as the productivity-boosting AI assistant for Office and Windows, finds itself facing a paradox: mass availability, but limited enthusiasm. Despite being shipped by default to millions of enterprise users—79 million downloads, according to Sensor Tower—Copilot is lagging behind category leader ChatGPT, which now boasts 900 million downloads.As reported by Bloomberg and confirmed by multiple analytics firms, this signals a persistent challenge: even “free” or enterprise-licensed AI doesn’t guarantee adoption. Microsoft’s bundling with the Office suite provided a powerful distribution channel. However, user perceptions of accuracy, utility, and overall experience appear to lag behind OpenAI’s more widely hyped (and sometimes more “conversationally flexible”) offering.
Furthermore, enterprise buyers—long Microsoft’s core audience—have voiced concerns about Copilot’s integration with legacy systems, limitations in supported languages, and the need for robust privacy controls. While Microsoft has signaled ongoing investment in improving Copilot’s functionality and reach, the company’s relative underperformance is a case study in how speed to market does not always equate to product-market fit, particularly in fast-evolving technology segments where user trust is paramount.
ChatGPT as an Agent: From Chatbot to Taskmaster
The most transformative development of the week, by most accounts, is OpenAI’s rollout of a new “agent mode” in ChatGPT. Going far beyond simple conversation, this feature allows ChatGPT to act as an autonomous agent: it can access a user’s personal data, manipulate web browsers, and interact with third-party APIs to complete multi-step tasks.OpenAI claims that combining the capabilities of ChatGPT, “Operator,” and deep research allows its AI to “do work for you”—from summarizing emails to automating business workflows via integration with services such as Hubspot. The implications are formidable: rather than “AI as widget,” users are now experimenting with full-stack, workflow-driven automation. Early tests (accessible to Pro subscribers, with broader rollout pending) show that even technically non-savvy users can create powerful automations—suggesting a significant democratization of what was formerly the domain of IT and data engineering teams.
Yet, there are unavoidable caveats. Concerns about data security, consistency of agentic performance, and the possibility of runaway tasks (sometimes colloquially termed “AI hallucinations”) remain. OpenAI asserts that rigorous safeguards are being put in place, but independent audits and public transparency about the actual rate of AI errors or misfires are still needed before skeptical organizations will hand over mission-critical operations.
Industry Sentiment: Markets and Mindsets
On the capital markets front, investor sentiment around media and tech stocks showed renewed optimism. The Unmade Index climbed back to highs not seen since early 2024, with mainstays like Nine and Ooh Media posting modest gains, while IVE Group surged 4% and Enero leaped ahead more than 6%. News Corp, not formally included in the Index, also saw healthy movement—a sign that investor confidence in media sector resilience (or at least adaptability) remains robust, despite ongoing structural change.Yet, there is an undercurrent of caution as well. Share price improvements may reflect hopes that AI will drive efficiencies and new profit pools, but also contain anxieties about margin compression and competitive upheaval if legacy models fail to adapt. Business leaders now confront a dual imperative: harness AI to fuel growth or risk obsolescence as new entrants and agile incumbents accelerate their digital reinventions.
Critical Analysis: The Strengths—and the Threats—of AI Agents
Strengths
Speed and Scale
The operationalization of AI agents and generative tools offers real, quantifiable improvements in speed and creative iteration. Brands like Unilever are not exaggerating when they cite order-of-magnitude gains in production velocity—something that was barely achievable outside of “template-based” creative just two years ago.AI’s ability to generate, test, and refine vast numbers of content variations in real time is a game-changer for organizations tasked with personalization at scale. In the competitive, always-online world of retail and digital media, this is not just beneficial; it is essential.
Data-Driven Creativity
By pairing machine learning algorithms with massive datasets and feedback loops, marketers can finally move beyond “gut feel” creative—a shift highlighted by Kantar’s research into AI-optimized campaigns. Automated pre-launch testing and dynamic performance optimization mean that weaker creative can be shelved quickly, and winning variants can be deployed widely.Workflow Automation
Agentic AI—like OpenAI’s new ChatGPT mode—promises far more than creative augmentation. It can shoulder entire workflows, from customer service ticketing to cross-platform analytics aggregation. This reduces pressure on overworked teams and opens the door for organizations to scale operations without commensurate increases in headcount.Risks
Displacement of Human Labor
The most immediate—and contentious—risk is the continued disruption of jobs in media and production. AI that can generate functional (if not always inspired) creative, schedule social posts, and even manage customer inquiries, inevitably reduces the need for human intermediaries. There are already reports of agencies rethinking staffing, and unions are lobbying for “AI impact assessments” akin to environmental reviews in major projects.Quality, Trust, and the “Flood”
The risk that AI-generated content will swamp platforms with low-value, algorithmically optimized (but contextually bland) material is real. Both consumers and regulators are voicing concerns about misinformation and a possible drop in editorial standards, especially as AI models remix and repurpose existing material with little transparency. Publisher partnerships, as seen with The Economist, may help set a quality bar, but the effectiveness of these arrangements remains to be seen.Privacy and Security
Agent-mode operations and API integrations pose new security challenges. Every new automation endpoint is a potential vulnerability. Additionally, with AI tools now able to access sensitive commercial or personal information, questions of data sovereignty and compliance—especially under evolving data protection legislation—are growing more urgent. Strong, independent oversight will be needed to ensure users’ data is not exploited or misused.Vendor Lock-In and Platform Dominance
As OpenAI, Google, and Microsoft battle to become the primary AI “operating system,” there is a danger that organizations could find themselves locked into proprietary ecosystems, limiting flexibility and stifling diversity of solutions. Already, some businesses are finding that workflow investments made in one AI stack do not necessarily transfer easily elsewhere—imposing both practical and strategic constraints.The Road Ahead for Media and Marketing
It is clear that AI is not just another technology trendline; it stands to reshape the entire arc of media and marketing strategy. The move from AI as assistant to AI as agent is especially pivotal: it introduces a logic of automation that reduces friction but also raises the stakes for oversight, creativity, and long-term employment.To thrive, organizations—whether brand marketers, publishers, or technology providers—will need to invest not only in AI tools, but in the training and creative processes that allow humans to steer, query, and supervise their increasingly agentic partners. Ongoing research and rigorous experimentation, coupled with open dialogue about failures as well as successes, will be required to ensure that AI-driven transformation serves the broader interests of audiences, creators, and the public.
As industry leaders look forward to the next wave of conferences and market milestones—from Unmade’s REmade and Unlock events in retail and nightlife, to end-of-year indicator roadshows—it is apparent that AI will be at the center of every strategic discussion. Whether as accelerant, disruptor, or simply the new digital substrate, artificial intelligence is here to stay—arriving not as a helper, but as a formidable workforce in its own right. The challenge now is making sure we manage the agents before they manage us.
Source: unmade.media This week in AI: The agents arrive