Marketing has always been part art, part science and part glorious chaos.
For decades, marketers have tried to understand customers, tell better stories, build stronger brands and prove the return on every dollar spent. The tools have changed dramatically, from print ads and trade shows to websites, search, social media, marketing automation and data platforms. Yet the underlying ambition has stayed surprisingly consistent: reach the right people with the right message at the right time, and do it in a way that feels human.
Now AI is forcing marketing into its biggest reinvention yet.
In my conversation with Julia White, Chief Marketing Officer at AWS, she put it perfectly when she said that what has not changed is that “great storytelling and connecting at a human level is still the most important thing.” Then she added the other half of the story: “Other than that, I think everything has changed.”
That captures the strange and exciting moment marketing finds itself in. The human heart of the discipline remains the same, but almost every process around it is being rebuilt.
Marketing Is Becoming A Technology Job
Marketing has been moving toward technology for years. Digital advertising, marketing automation, analytics, customer data platforms and SEO all pushed marketers to become more comfortable with systems and data. AI takes that shift to a different level.
White described marketing today as “a technology job too,” especially with AI. She made the point that marketers spent years learning how to master search engine optimization, then “we woke up one day and the world had flipped over to agentic and AI optimization.”
That is a huge shift. For years, marketers created content with human search behavior in mind. We asked how people searched, what keywords they used and how Google would rank the content. Increasingly, the first audience may be an AI assistant. Customers may ask ChatGPT, Claude or another AI tool which product to choose, which supplier to trust or which solution best fits their needs. The AI becomes part researcher, part adviser and part gatekeeper.
This changes the role of marketing. Brands must still speak to people, but they also need to be understood by machines that summarize, compare and recommend. Clear positioning, credible content, structured information and trusted signals become even more important.
White said AWS is already seeing a difference in customer behavior. When people arrive through AI-driven discovery, “they engage more deeply, they’re more qualified.” In other words, AI may reduce some casual browsing, but it can also bring in people who are further along in their thinking.
From Generative AI To Agentic Marketing
The first wave of generative AI in marketing focused on content. Marketers used AI to write emails, draft social posts, summarize research, generate campaign ideas and adapt copy for different audiences. That was useful, but it was also only the beginning.
The next wave is agentic AI, where AI systems can take on whole tasks and move through workflows with a degree of autonomy. White explained that agents take “the capability of those large language models” and put it “into something that is autonomous, can run on its own.”
This is where marketing starts to change at the operating model level.
One of White’s examples was content localization. Like many global companies, AWS has to localize content across multiple languages. In the past, machine translation would provide a first draft, then linguists would spend significant time correcting grammar, adding nuance and making the content locally relevant. White said localization into 16 languages could take “more than two or three weeks.”
AWS now uses agents to improve the initial translation, address grammar, add local relevance and prepare the content before a human linguist reviews it. The human role becomes more valuable because linguists can spend their time making the content more compelling rather than fixing basic mechanics.
This is a very practical example of where AI creates real value. It does not require a science fiction vision of marketing. It starts with a common bottleneck and makes it faster, better and more scalable.
AI Can Collapse Slow Workflows
The real power of agentic AI appears when it changes the shape of work.
White described the traditional content process as linear. A campaign team needs content. A content team works on it, often with agencies. The material then goes through review, web publishing and quality checks. That process works, but it is built around human capacity and sequential handoffs.
AWS has taken what White called an “agent first approach” to parts of this workflow. She gave the example of webpage creation, something AWS does around 10,000 times a year. Now, a marketer can use a natural language prompt to ask for a webpage about a product, campaign or topic. The agent can go into the content management system, pull together the content, apply the right template, include the correct links and images, check SEO and AI optimization, verify links and assess backend rendering.
White said the agent can “collapse what was a pretty linear process into just a single, a single agentic work stream.”
The business lesson here is that the largest gains come when organizations rethink workflows around what AI can now do, not from sprinkling AI across old processes and hoping for magic.
The marketer’s role also shifts. Instead of assembling pieces, chasing approvals and checking links, marketers can ask better questions. What story should we tell? Why should customers care? How do we make this distinctive? Where is the emotional connection?
That is where marketing becomes more human, rather than less.
The Return Of The Art Form
There is a lot of anxiety about AI in creative professions, and marketing is no exception. If AI can write copy, create images, generate campaigns and automate analytics, where does that leave marketers?
White is optimistic because she believes AI can help marketers return to the real craft of marketing: telling stories that connect with people at a human level.
She also admitted that much of a marketing team’s time is often spent away from that art form. Her estimate was that perhaps only 20% of team capacity goes into storytelling, breakthrough ideas and human connection. The rest is absorbed by the mechanics of modern marketing.
This is the most useful way to think about AI in marketing. The goal should be to automate the friction, not the judgment. AI can take on the repetitive tasks, the assembling, checking, formatting, reporting and basic adaptation. People should move closer to the work that requires taste, empathy, judgment and originality.
White shared a lovely example. When AWS demonstrated a new agentic web assembly process to the marketing team, there was “spontaneous applause in the room.” Why? Because marketers were delighted that they would no longer have to spend time on one of the least enjoyable parts of the process.
That reaction tells us something important. AI adoption improves when people see AI removing the work they dislike, rather than threatening the work they care about.
The Dashboard Era Is Giving Way To AI Thought Partners
Another powerful AWS example involved analytics. White described how AWS previously had a team dedicated to building BI dashboards for marketers. They had around 1,500 dashboards and a backlog of 2,000. That is a familiar story in many large organizations. Everyone wants data, but the data team becomes the bottleneck.
AWS responded by building an agentic BI system called AIRO, short for Agentic Insights and Research Organization. Instead of requesting a dashboard, marketers can ask the system what they need to know. The system brings together data warehouse insights, analytics and causality studies from data science and economics teams.
The shift is significant. Traditional dashboards tell you what happened. AI-powered insight systems can help explain why it happened and suggest what to do next.
White described the difference clearly. In the old model, she said, “I used to just know what happened.” Now marketers can ask why a campaign did not perform and receive an explanation based on causal studies, related campaigns and contextual data.
This is where AI moves from a reporting tool to a thought partner. It can help marketers test assumptions, explore patterns and make better decisions. Of course, humans still need to challenge the output, check the reasoning and bring commercial judgment. But the work becomes richer.
How To Start Without Creating Chaos
One of the most useful parts of my conversation with White was her honesty about what did not work.
A little over a year ago, she challenged her team to become the most AI-forward marketing team. People responded with energy. They started building tools and agents. Then she checked the internal wiki and found 56 different content agents.
Her reaction was: “We do not need 56 different ways to create content.”
This is a warning for every organization. Enthusiasm without coordination can quickly create fragmentation, duplication and what White called “AI slop.” Giving everyone freedom to experiment is useful at the start, but it cannot be the operating model for serious transformation.
AWS pivoted by choosing five major work streams where teams could work together and move quickly. Content was one. Lead management and event management were others. That created focus and made it possible to redesign work properly.
White also shared a simple tactic that any organization can use. AWS asked employees to submit their “paper cuts,” the annoying parts of their jobs they would love to remove. As White put it, “Send us your paper cuts. Send us the things that if you never had to do that again, you’d be delighted.”
This is a brilliant starting point for AI adoption. Rather than beginning with a grand strategy document, begin with the work people already know is broken, slow or frustrating. Solve those problems first and you build trust, momentum and excitement.
The Future Of Marketing Is Agent First
White believes we are still early in the agentic wave. She compared today’s agents to the early days of personal computers, when people had to build their own machines by assembling components. Over time, agents will become easier to use, more packaged and more accessible to marketers who are less technical.
That will open the door to a much broader transformation.
I believe the future marketing function will look very different. AI agents will help plan campaigns, create and localize assets, test messages, analyze performance, manage content operations and surface customer insights. AI assistants will also act on behalf of customers, comparing products, filtering claims and recommending decisions.
This creates a new premium on trust. If AI systems are going to summarize your brand, compare your proposition and recommend your products, vague messaging will not survive. Brands will need to be clearer, more credible and more useful. Content created only to satisfy algorithms will become less effective. Content that genuinely helps customers make better decisions will become more valuable.
The irony is that AI may push marketing back toward its roots. The mechanics will become faster and more automated, but the differentiator will be the quality of the idea, the clarity of the story and the strength of the human connection.
White put it beautifully when she said, “We get to get back to our art form.”
That, to me, is the real promise of AI in marketing. It is not about replacing marketers with machines. It is about giving marketers the time, insight and intelligent support to do the work that made them want to become marketers in the first place. In an AI-shaped world, the best marketing may become more human than ever.
