People Inc. Builds Custom APIs So AI Agents Can Read Its Journalism

Digiday reports that People Inc. is building custom APIs intended to make its journalism easier for AI agents to use. The same report notes that Forbes has adopted Markdown for agents, and that the FT and the Daily Mail are considering how agents could reach content on behalf of subscribers. Those are the facts as published; the report does not detail the technical specifications, timelines, commercial terms or traffic volumes involved.

What the shift actually concerns

The move is about machine readability rather than audience reach in the classic sense. Agents that summarise, answer questions or assemble briefings need content in a form they can parse reliably, and clean structured output is easier for them to consume than a fully rendered web page. Markdown adoption at Forbes points in the same direction: a lighter, more predictable format for automated retrieval.

The commercial question sits underneath. When an agent answers on someone's behalf, the publisher wants that access to be attributable, permissioned and, ideally, monetised. The FT and Daily Mail angle reported by Digiday — how agents could reach content for subscribers — frames access as an extension of an existing subscription relationship rather than as open crawling. None of the announced plans specify how that would be enforced or priced.

What changes for acquisition and monetisation teams

For most publishers and content-led businesses, the practical consequence is that a growing share of discovery may happen inside an agent's answer rather than on a search results page. That affects two things teams already track: whether the brand is cited when an agent responds, and whether the click still arrives.

Some concrete checks a team can run this week:

• Audit how much of your content is reachable and parseable without JavaScript rendering, since automated retrieval often fails there first.
• Review robots and terms of service to decide, deliberately, whether agent access is allowed, restricted or licensed.
• Look at server logs for non-browser user agents to see whether automated retrieval is already happening and on which pages.
• Decide which assets — pricing pages, product documentation, evergreen explainers — matter most to be quoted accurately, and make those the cleanest to parse.

The open questions

Digiday's report leaves the operational detail unspecified: no information on how People Inc.'s APIs will be authenticated, whether access will be metered, or how subscriber entitlements would be verified by a third-party agent. The same gap applies to the FT and Daily Mail, which are described as considering options rather than committing to a model.

That uncertainty is the point worth planning around. Formatting content for machine consumption is a low-cost, reversible step. Building a licensing or entitlement layer is not, and it depends on standards that do not yet appear settled. Teams that separate the two — fixing parseability now, holding pricing decisions until agent access patterns are measurable — keep their options open without betting on a model that may change.