Agentic Commerce for regional media
The next reader is an AI agent
The real danger for publishers is not that people suddenly stop reading the news. The danger is more subtle: people continue to read the news, but no longer where it originates.
A reader no longer automatically reaches for the newspaper, no longer naturally opens the e-paper, and might not even land on the website. They ask their assistant: What's important in Heilbronn today? Why was the city council meeting interrupted? Which school is being renovated? What does the decision mean for my neighborhood?
The assistant's answer is still based on journalistic work. But visit, ad contact, subscription signal, and customer relationship no longer automatically occur at the medium. That is exactly where the new strategic question begins.
For regional journalism, this question is particularly urgent because local information is not arbitrarily interchangeable.
An opinion on federal politics can be found everywhere. A reliable assessment of Bad Friedrichshall, Neckarsulm, Eppingen, Öhringen, Künzelsau, Bad Rappenau, or a construction project in Heilbronn cannot. This is exactly why regional journalism is so attractive to AI systems: It contains information that cannot be arbitrarily replaced from the web.
City councils, club life, police reports, regional economy, schools, culture, sports, obituaries, local debates: For humans, these are news. For agents, they are bases for decision-making. And for a media house, they are an information layer that will increasingly be used by machines.
What is shifting
Classic usage
Medium is surface and source.
Whoever reads the article comes to the app, the e-paper, or the website. That is where reach, subscription signal, brand loyalty, and monetization are generated.
Agentic usage
The medium remains the source, but not the surface.
The assistant provides the answer. The publisher must remain visible, be able to measure usage, and define what processing is allowed.
The classic paywall only partially solves this problem.
The classic paywall asks: Are you a subscriber? Are you allowed to read further? That remains important. But agents ask a different question: Am I allowed to process this content, summarize it, quote it, include it in a briefing, or use it as a basis for a decision?
This usage is not a normal page view. It can even create more value than a fleeting click because it is often tied to concrete decisions: Should I attend this appointment? Does this construction site affect me? Is this development relevant to my business? However, if value is created based on this information, the media house hardly gets a share of it today. The existing model primarily sees the visit, the subscription, or the ad. It does not see that an agent translates the information into a decision. That is why it is not enough to just put everything behind a higher wall. Publishers need a second layer: a fair interface for machine usage.
What a test could look like
A small test instead of a big project
A regionally rooted media house can answer this question in practice rather than in theory. It has digital subscriptions, a strong local brand and content that agents cannot simply replace. Revoki is already developing Morning Briefing, an agent that not only collects news but classifies it for individual roles, industries, and accounts. A joint pilot could show how licensed agent usage works without immediately building a new large system.
User
pays Morning
for a personal briefing
Morning
reads on behalf of the user
and cites the source in the briefing
Publisher
provides content and rules
and receives its share via x402
Publisher → Morning
Articles, facts, metadata, licensing rules
Morning → Publisher
Source attribution, proof of usage, revenue share
The keyword behind this is Agentic Commerce: digital agents that not only read information but act on behalf of a user. If they can select sources, observe budgets, and pay themselves in the future, a second revenue layer emerges alongside the reader subscription: paid usage by agents.
Agents are beginning not only to read, but also to act and pay. Mastercard is working with Agent Pay for Machines on a payment infrastructure for software that independently executes transactions. x402 already makes the old HTTP status code '402 Payment Required' practically usable: A resource tells the agent what it costs; the agent can pay and repeat the request. Cloudflare is testing Pay per Crawl with a related logic, where website owners can allow, block, or charge for crawlers.
A pilot would not have to start with the entire public website. It would make more sense to have protected access to a small section of the article database: article text, metadata, publication time, department, paywall status, and simple licensing rules. An x402 endpoint could be placed in front of it, allowing agents to access the resource only after accepting the terms and paying. In this way, fair AI usage would not just be discussed, but could be tested end-to-end.
The most important sentence for media houses
People subscribe. Search engines link. AI agents license.
This is not a break with journalism. It is its continuation in a new distribution layer. Whoever researches, classifies, and takes responsibility creates value. This value should remain visible and compensable even when it appears in the answer of an assistant.
We would like to test this practically.
Revoki can set up such a test: a small, clearly defined local information feed connected to Morning. Not a large transformation project but a visible prototype: an agent retrieves protected content from the publisher, names the source and pays for access automatically, for example via x402. Small enough to start quickly, concrete enough to matter strategically.
Discuss PilotSources and triggers
The starting point was a Morning Briefing article on Mastercard Agent Pay for Machines, Coinbase for Agents, and x402. Further references: Mastercard AP4M, x402, Cloudflare Pay per Crawl, Reuters Institute Digital News Report 2025.