Publishers Are Building Structured News Feeds for AI Agents

Axios is preparing three feeds for AI agents. The first gives Axios Pro Deals subscribers’ agents merger-and-acquisition coverage and launches this fall. The other two are planned for next year.

At the Digiday Publishing Summit in September, Axios announced the coming Axios Direct service. It will send reporting to software agents through standardized lines of text.

Every morning, Michelle Seo receives a briefing from Chattiana, her Muse artificial intelligence agent. The briefing lists Seo’s agenda and to-do items. It pulls water-cooler topics from headlines in her newsletters and includes links to the sources.

“I'm kind of challenging myself as an introvert to start more conversations,” Seo, a San Diego-based public relations consultant, told WIRED. She says the briefing gives her the day’s news and conversational openings when a Zoom call turns quiet.

According to WIRED, that small routine reflects a larger change. Some outlets now write for AI agents rather than human readers alone. Their work includes AI-optimized feeds and articles designed for agents to personalize.

Who is the audience now?

Publishers are responding on two fronts. WIRED reports many, including WIRED’s parent company Condé Nast, signed deals allowing AI platforms to surface their content. Outlets also use “generative engine optimization,” or GEO, to improve their chances in responses to AI prompts.

Last year, AIOSEO, a SEO plug-in for WordPress sites, added a feature that generates LLM.txt files. The files help AI systems discover and interpret websites’ content.

In 2023, The New York Times sued OpenAI and Microsoft for copyright infringement. It alleged that the companies trained models illegally on its reporting. That case remains ongoing.

Axios and a handful of peers are trying something more foundational. They are changing how stories operate and whom they serve. Axios chief revenue officer Jacquelyn Cameron said: “We have always said that we want to be where the audience is.”

“We believe that the media ecosystem is changing,” Cameron said. “We see this with open web deteriorating. We see that search is shrinking. We see that hyperpersonalization is surging, that LLMs are becoming more important in the way that people are consuming” their news.

That is the market pressure behind Axios Direct. Under Cameron’s diagnosis, publishers can no longer rely only on search or homepage visits. Agents may become the layer that selects, summarizes and routes the reporting.

What does Axios Direct change?

Axios is trimming its already-brief content to satisfy AI. Cameron said: “We do believe that a significant portion of news consumption will move into AI systems, LLMs, and devices.”

She describes the feeds as less narrative and more structured data, closer to a Bloomberg terminal than a set of articles.

The first two Direct feeds target corporate clients:

• Merger-and-acquisition feed: It launches this fall and gives AI agents belonging to Axios Pro Deals subscribers access to Axios’s M&A coverage.
• Training feed: It will act as a source for training in-house large language models and launches next year.
• Consumer feed: Axios plans to develop it next year for individual subscribers to receive Axios reporting directly through their agents.

All three Axios feeds will look like RSS feeds. Their standardized lines of text are not meant for human eyes. Axios says they are not designed to be aesthetically pleasing.

WIRED’s business logic for starting with businesses is straightforward. Corporate budgets have more room for experiments than individual subscriptions, especially when timely information might deliver a small advantage.

How does Every make stories extendable?

Every takes a different approach. The six-year-old tech media and software company announced a $2 million funding round last year from investors including LinkedIn founder Reid Hoffman. It expects subscribers to treat AI as an extension of themselves, rather than a separate tool.

Every founder Dan Shipper says the outlet’s writers include prompts in articles whenever possible. Readers can test ideas themselves. An article about a new model, for example, provides copy-and-paste prompts that demonstrate how the model answers questions and evaluates input.

Guides also include buttons labeled “Read With Claude” or “Read With ChatGPT.” Clicking one opens a prompt instructing the AI to say:

“Start with a tight summary: one paragraph, bulleted. Then offer to go deeper.”

That design lets Every pack stories with details without worrying about losing human attention. Agents pick out the pieces most relevant to each reader.

Two years ago, Every published its first “extendable” article. The long-form feature came with a database containing interview recordings, news articles and YouTube videos. Readers could ask a chatbot questions, and it would fetch more details from that database.

Recently, Every published an 8,000-word article about Codex. Subscribers’ agents can draw on the same database to tailor a new piece to their interests. The output might be an article, podcast or video. One reader could request bulleted points. Another could ask for a short film built from interview clips.

“As a writer, you have to make choices about what goes into the story you're telling,” Shipper says. “And there's always a question about, are those the right choices? Or, are those the right choices for a particular reader?”

With AI, Shipper said, journalists no longer choose. “People are gonna do crazy, crazy stuff with this,” he added.

Can publishers make money from agents?

Rasmus Kleis Nielsen, a professor of communication at the University of Copenhagen, says the market for publishing to agents is “emerging.” He calls it immature and “extremely uneven.” Price points and the way agents interact with content remain open questions, he says.

Despite that uncertainty, Cameron and Axios chief operating officer Allison Murphy remain optimistic about AI as an audience proxy. According to the executives, Axios’s short-form, highly structured style already makes the publisher a favorite source for AI systems answering prompts.

“Time is money for crawlers,” Murphy says. “They don't want to have to go three quarters down the page to find the main point. So, that's something that's already benefiting us.”

Even if readers never visit Axios’s homepage, Nielsen says, brand visibility inside a chatbot can still help. He compares that with content made for TikTok and Instagram, platforms that rarely send viewers back to publishers’ sites.

WIRED notes the harder problem: a chatbot can hand readers an answer on a silver platter. Publications then struggle to monetize reporting when other publications and LLMs already offer the same information. People will pay only for what those sources do not have.

Every’s answer is detailed guidance on AI and how to use it. Axios Pro Deals offers speed, with information that could move the market. An all-access Axios Pro Deals subscription costs $2,499 a year. An annual Every subscription costs $288.

Shipper sees that $288 as payment for a human reader and the reader’s agents. “I may not want to read all of your stuff every month,” he says. But if the subscription includes his agent, “I may want my agent to know it so that it can tell me, ‘Hey, this thing is important.’”

Is AI an audience or a tool?

Murphy’s starting point is acceptance. People already use AI to absorb information, and she expects more will do so soon.

“I also try to make it a little less existentially terrifying by thinking, this isn't the first time this has happened in media,” Murphy says. People get their news via social media, short-form video and podcasts.

“It isn't really AI as the audience,” Murphy adds. “It's AI as a tool that helps us reach our audience.”

Those statements describe the tension running through publishers’ agent strategies. Axios’s corporate feed launches this fall. Its training feed and consumer-feed development are slated for next year. Those releases will test whether agent access can become durable revenue without treating an AI reference as equivalent to a human visit.