The Unseen New Audience For News: AI Agents

The Unseen New Audience For News: AI Agents

Every morning, San Diego-based public relations consultant Michelle Seo gets a customized briefing from her personal Muse AI assistant, Chattiana. The roundup lays out her full daily agenda and to-do list, and pulls casual, shareable talking points from the headlines of her subscribed newsletters, complete with direct source links. “As an introvert, I’ve been challenging myself to initiate more casual conversations lately,” Seo tells WIRED. The AI-curated briefing gives her a quick lay of the day’s news, so she always has a relevant topic on hand to break awkward silences on Zoom calls.

Seo is far from an early adopter of AI-assisted news consumption: the trend has become so widespread that a growing number of news outlets are now building content explicitly for AI agents, rather than human readers, as their core intended audience. This shift takes many forms, from AI-optimized content feeds to modular, structured stories designed so AI tools can rearrange them into fully personalized articles for individual users.

A Fundamental Reset For Media Distribution

“We see the entire media ecosystem shifting right now,” says Axios chief revenue officer Jacquelyn Cameron. “The open web is deteriorating, traditional search traffic is shrinking, hyper-personalized content is surging, and large language models are becoming the primary way people access their daily news.”

Major publishers have responded to this landscape shift in a range of different ways. Many outlets, including WIRED’s parent company Condé Nast, have signed formal licensing deals to allow AI platforms to surface their content in chatbot outputs. Teams across the industry are experimenting with “Generative Engine Optimization” (GEO) strategies to make sure their reporting gets prioritized and cited when users submit prompts to AI tools. Last year, AIOSEO, the popular WordPress SEO plugin, rolled out a new feature that generates LLM.txt files for websites, designed to help AI crawlers easily discover and correctly parse site content.

Other outlets have taken a defensive stance instead: The New York Times sued OpenAI and Microsoft in 2023 for copyright infringement, alleging the two firms illegally trained their generative models on the Times’s original reporting, a case that remains ongoing.

Building News Feeds Exclusively For AI Parsing

What Axios and a small cohort of forward-thinking outlets are rolling out, however, represents a far more foundational overhaul of how media operates, and who their stories are built to serve. “We’ve always held that our job is to meet audiences wherever they are,” Cameron notes.

At the Digiday Publishing Summit last September, Axios officially announced the upcoming launch of Axios Direct: purpose-built content feeds designed exclusively for AI agents. “We are confident that a very large share of news consumption will eventually happen through AI systems, LLMs, and smart devices,” Cameron says. She describes the new feeds as a far more efficient way for AI tools to access Axios content: stripped of all unnecessary narrative fluff, they prioritize structured data, making them far more comparable to a Bloomberg Terminal than a collection of traditional news articles.

The first two Axios Direct feeds are built exclusively for enterprise clients. The first, launched this fall, gives the custom AI agents of Axios Pro Deals subscribers direct access to the outlet’s full library of merger and acquisition reporting. The second feed, launching next year, is designed to be used as a curated, high-quality source material for training enterprise in-house large language models.

Starting with B2B audiences is a logical move: corporate budgets are far more willing to fund experimental, unproven products that could deliver even a small competitive advantage than individual consumer subscribers. But the third Axios Direct feed, slated for development next year, will be consumer-facing, letting individual subscribers send Axios’s full reporting catalog directly to their personal AI agents. All three feeds will operate on a similar framework to classic RSS feeds: they have no visual polish whatsoever, as they are not designed for human eyes, and exist purely as standardized lines of parseable text.

An Alternative Approach: News Built As An Extension Of Personal AI

While Axios is trimming its already famously concise content to suit AI consumption, six-year-old tech media and software firm Every is taking a wildly different tack. The company, which announced a $2 million funding round last year with investors including LinkedIn founder Reid Hoffman, builds its content to treat AI as an extension of the individual reader themselves.

Every founder Dan Shipper says the core difference between his outlet’s stories and standard news articles is that writers embed custom AI prompts directly into every piece wherever possible, letting readers copy and paste the prompts to test new concepts firsthand. A writeup of a newly launched AI model, for example, will include a full suite of prompts users can drop into ChatGPT or Claude to see exactly how the model responds to different queries and evaluates different types of input. Practical content like how-to guides also feature one-click buttons that let readers “Read With Claude” or “Read With ChatGPT” – clicking the button loads a pre-written prompt that instructs the AI to “Start with a tight, single-paragraph bulleted summary, then offer to dive deeper into any specific section you’d like to explore.”

Writers are free to pack stories full of granular, niche details without worrying about overwhelming casual readers, since they know personal AI agents will sift out the exact details that matter most to each individual user. Two years ago, Every launched its first “extendable” longform feature: a deep reporting piece bundled with a full database of source materials, from interview recordings to raw news clips and linked YouTube footage, which subscribers could query via a custom chatbot to pull up extra context. Most recently, the outlet published an 8,000-word deep dive on OpenAI’s Codex platform that subscribers’ personal agents can reformat entirely to fit their preferences: one reader might request a bulleted quick summary, while another can ask for a short highlight reel made from raw interview footage.

“As a writer, you always have to make judgment calls about what to cut and what to keep in a story,” Shipper explains. “You’re always left wondering if those calls are the right ones, or if they work for every single reader. With AI, journalists no longer have to make those compromises. People are going to do all kinds of wildly creative things with this format.”

The Open Question: Monetizing AI-Focused News

As outlets increasingly treat AI agents as a core new audience, one major question hangs over the entire trend: can publishers actually turn this shift into sustainable revenue? Rasmus Kleis Nielsen, a communications professor at the University of Copenhagen, notes that the market for agent-targeted publishing is still emerging, immature, and “extremely uneven” across the industry. Core questions around pricing structures, and how AI tools will interact with publisher content long term, remain largely unresolved.

Axios’ CRO Cameron and COO Allison Murphy are deeply optimistic about AI as a proxy for reaching human readers. They point out that Axios’s signature short-form, highly structured news style was already perfectly optimized for AI crawlers, who prioritize speed and clarity above all else. “Time is money for crawlers,” Murphy says. “They don’t want to have to scroll three quarters of the way down a page just to find the core point of the story. That’s a dynamic that already plays to our strengths.”

Nielsen notes that even if AI users never click through to visit Axios’s homepage, the brand visibility of showing up in AI responses is valuable on its own – a dynamic very similar to publishers building viral content for TikTok and Instagram, even when those platforms drive very little referral traffic back to their original websites. Across the industry, publications are grappling with the new reality that generic, widely available content is almost impossible to monetize now that chatbots can serve up answers for free. Audiences will only pay for information that no generic LLM, and no competing outlet, has access to. For Every, that means hyper-granular, hands-on guidance for using AI tools effectively; for Axios Pro Deals, that means real-time, market-moving M&A intelligence no other outlet can deliver as fast. An all-access Axios Pro Deals subscription costs $2,499 a year, while an annual Every subscription costs $288.

Shipper frames that $288 price point as a subscription that covers both a human reader and their personal AI agent. “I might not have the time to sit down and read every single one of your stories every month,” Shipper says. “But if my subscription includes access for my AI agent, I can have it ingest the entire catalog and just flag, ‘Hey, this piece is directly relevant to your work.’”

Murphy adds that the industry should not treat this shift as an unprecedented existential crisis: this is just the latest stage of media distribution evolution. “People already get their news via social media, short-form video, and podcasts. This isn’t really AI as the core audience at the end of the day – it’s AI as a tool that helps us connect with our human audience far more effectively than we ever could before.”