On the sidelines of the Human Ventures event, I spoke with several industry insiders who outlined how publishers can reconfigure their business strategy and tech stacks to take advantage of the A.I. boom—including by changing their approach to metadata, the back-end gobbledygook that gets fed into Google. Yes, we’ve been down this road before. During the 2010s, an entire generation of digital media companies like BuzzFeed, Business Insider, and HuffPost were built atop that traffic firehose. Later in the decade, they all pivoted to trying to gamify traffic on Facebook and Twitter. In the end, of course, both models proved to be transitory.

Now, as media professor Daniel Trielli recently predicted, we’re likely to see more and more journalism targeting audiences that aren’t human at all. “Agentic journalism would break from our traditional article format,” Trielli wrote for NiemanLab. “A.I. systems do not need ledes, nut-graphs, or narrative flows; they need user-relevant, novel, and machine-readable content.” Even publishers that continue to serve ordinary readers, like The Atlantic, will need a metadata strategy to make their content more appealing to various bots and crawlers.

As ever, the tricky part is finding the correct balance. Traditional publishers waited too long to treat the internet as a dead-serious disruptor, and are still paying the price. At the same time, dozens of ill-fated digital startups leaned too far into optimizing for various S.E.O. gimmicks, like Mark Zuckerberg’s infamous “pivot-to-video,” only to get burned when those algorithms changed. It’s entirely possible that the pivot to A.I. could also be a mirage or head fake, as the industry whipsaws toward another shiny object.

Wise media executives should learn from the recent past that there’s little long-term upside to monetizing third-party platforms. Yes, X and Facebook are still important marketing funnels—and as places where brands need to show up to build awareness. But there’s danger in trying to scale audiences on those platforms, especially if only a tiny sliver of the audience converts to paying subscribers. Media will have the same problem with the A.I. giants: While some outlets have cut licensing deals with OpenAI, nobody is getting significantly paid when their work appears in A.I. summaries—at least not yet. And while publishers need to be thinking about how to optimize content so that it’s competitive in A.I. search and discoverable by A.I. agents, nobody knows exactly how these markets will ultimately function.

For some, the best offense may be a good defense. The New York Times, for one, has sued OpenAI, and when New York Times Company C.E.O. Meredith Kopit Levien was interviewed by Ben Thompson last week, she specifically pushed back on the idea that NYT Cooking, for example, can be replicated by ChatGPT. “The New York Times creates human-led, high-quality news journalism and all this other stuff, including recipes that are better because of the humanity,” she said. “We think in an information ecosystem where it’s harder and harder to find quality stuff—brands are going to matter more, and human-made content is going to matter more.”

The other great source of A.I. anxiety in media pertains to the future of advertising. Industry leaders like the Times will continue to be walled gardens for subscribers where advertising flourishes. It’s foolish to assume that no human will ever visit a news website ever again. At the same time, it’s worth considering how the rise of A.I. platforms and agents could impact the advertising market—especially if more consumers shift to using bots for personalized feeds. After all, even if readers maintain subscriptions to their favorite news sites, advertisers aren’t paying for impressions from bots.

That future is still a ways off: Agents remain difficult for the average non-A.I. enthusiast to set up; trust in the information that A.I. platforms surface is still low; and faith in A.I. in general is actually decreasing. Only around 10 percent of A.I. users feel comfortable with fully A.I.-generated news platforms, per the Reuters Institute’s most recent survey into news habits. Perhaps even more damning, less than 20 percent of Americans trust an A.I. system to take an action—even something as simple as summarizing various articles—and 41 percent say they don’t trust A.I. to make recommendations or provide information, per YouGov.

Alas, the lack of clarity surrounding future user behavior makes it all the more difficult for media publishers to figure out where, and when, to invest in A.I.-proofing their businesses. How imperative is it, really, for newsrooms to reinvent themselves to serve bots in addition to people? And how much of the hype around A.I. is being inflated by the tech companies that stand to benefit? Is the agentic web truly the next version of the internet, or is this just another pivot-to-video-esque distraction that’ll lead an already struggling media industry into another dead end? In her interview with Thompson, Kopit Levien expressed an unshakeable faith that her New York Times readership will continue to prioritize human-made content. But, of course, not every media company is the Times.