Why Online Publishers Must Embrace Agentic AI — Before It Truly Arrives

Progress is the only constant for publishers. The newsstand preceded the radio dial, which begat the TV guide, which was unseated by the search engine and, ultimately, the algorithm.

Throughout it all, publishers have been the ultimate destination for consumers who wanted to truly understand what was happening in the world or to keep up with their careers or industries or simply for entertainment.

And, yet a new development finally threatens to upend this long-standing dynamic: agentic artificial intelligence (AI). Agentic AI agents act autonomously, make decisions, and achieve goals in complex environments. It requires little human supervision and, unlike a traditional bot, agentic AI doesn’t just scrape or index pages. It acts. These agents don’t rely on users clicking through links on a search results page. Instead, they directly query websites, synthesize information, and return the answers in a personalized format— without the user ever loading the publisher’s webpage.

Strange Bedfellows

Calling the publishing industry’s relationship with agentic AI complicated is an understatement. Companies like Alex Springer, News Corp, and The Atlantic have struck deals with OpenAI. On the other hand, The New York Times sued the ChatGPT owner.  So did a consortium of publishers including The New York Daily News, Chicago Tribune, and Denver Post. Getty has sued image generator Stability AI.

No matter what a publisher’s strategy is, it should understand what agentic AI is, where it’s likely to go, and how it needs to adapt to survive and thrive.

Here are a few emerging strategies worth exploring:

Exploit The Gap Before Agentic Hits

If you write a prompt for a generative AI chatbot right now, you will likely see it check publications and other sources while formulating the answer. Most even link to the publisher, which has created a burgeoning field called generative engine optimization (GEO), a cousin of search engine optimization (SEO), where brands and publishers vie to be mentioned or linked to in GenAI responses. Learning more about how LLMs source content and tailoring how that content is structured to be featured in more searches is a valuable skill for publishers to learn.

Embrace Invisible Interactions Over The Pageview

Agentic AI can quickly learn customers’ interests and preferences. The select few consumers that are always exploring cutting edge use cases will enable these agents to search and interface with other agents once confidence grows through privacy and security standards. A majority of consumers, however, will wait until more safeguards are in place, and publishers that have already demonstrated customer trust and data privacy compliance can help shape how the majority of consumers will adopt agentic AI.

When agentic AI takes over search, research, and even shopping tasks, digital agents, not humans, will visit a publisher’s site. They will be purpose built for the task at hand and won’t consume ads, but it will be AI agents that will click on affiliate links, or fill out forms. Thus, it’s likely that the long-standing pageview-driven monetization model disappears.

So what are publishers to do in this new world? One thing is certain. They are no longer creating content just for humans, they are now meant to serve as a trusted, structured, and monetizable source of content- for humans and machines.

Almost paradoxically, that means publishers need to be even more buttoned-up and service-oriented for the machines than they ever were for humans.

Creating Structured Content Via API Licensing Or Consumption By LLMs

This is the path for some early moving companies, partner with those who seek to disrupt you before you get disrupted. Cloudflare, a company that helps sites manage traffic, recently introduced a pay per crawl setting that helps publishers block or give permission to AI crawlers for a fee.

Publishers with years of in-depth, high utility content, such as product reviews, news, recipes or other tentpole features, should consider offering it through an API. Smaller publishers could strategically offer content for free but establish themselves as the source for LLMs.

Right now, many LLMs are failing at citations and have less than 40% accuracy in a test by the Tow Center. Instead of agents scraping pages, sometimes without compensation, they can license your data in structured, digestible formats for a fee. Considering shopping is one of the biggest likely use cases of AI agents, being a core source for product reviews and analysis is a huge win.

Publishers, Disrupt Thyselves

AI is not merely an external force that threatens publishers’ way of doing business; it’s a tool they can use themselves to improve their connection to audiences. Publishers should already be planning agentic experiences of their own. The Washington Post created an AI chatbot for readers to interrogate facts about climate change. This is no different than publishers experimenting with social media, Substack, Reddit, or Quora— seeking to harness the next tech trend is part of the process.

Reimagine Subscriptions As Access To Expertise, Not Just Articles

If AI can answer most informational questions with a single sentence, the value of raw information will decline. But simple facts don’t truly explain the world. The importance of interpretation, expertise, and authority will only grow.

Publishers should position themselves as domain experts and offer access to communities and their experts. This may come in the form of member-only discussions, editorial curation, live interactions, or premium AI agents trained exclusively on the publisher’s content and point of view.

Plan For A Volatile Future, Hope For A Profitable One

Of course, the agentic future everyone predicts may not arrive. But Google, Apple, Microsoft, Meta, Amazon, and other large enterprise technology companies are plowing billions into generative AI. OpenAI makes $10 billion a year and is valued at $300 billion.

Publishers would be wise to assume agentic AI’s eventuality versus pursuing the status quo. That doesn’t mean abandoning their most hard-won principles, but rather exploring a path of exploration and open-mindedness about the value they provide in an AI-driven world. Publishers will need to move past the historic pageview model to embrace more holistic ideas of expertise while both considering licensing content to LLMs and consider pursuing their own AI chatbots and agents. That’s the best way to create a moat for their business in the era of agentic AI.

Paul Bell is the President at 33Across, a technology company that applies machine learning and addressability to segment data and activate campaigns across digital channels. In his role, he is responsible for driving the company’s growth and market strategy while overseeing the data and identity business. Paul joined 33Across as Chief Revenue Officer (CRO) in 2015, and has over 20 years of extensive experience in data and digital media.