People Inc. Builds Custom APIs to Serve Journalism Directly to AI Agents
The Development
People Inc. is shifting toward custom APIs that give AI agents article text and key details in a simplified format, rather than forcing them to parse full webpages. Jon Roberts, the publisher's chief innovation officer, laid out the rationale at Digiday's Media Strategies London event. His argument: when an agent reads an HTML page, roughly 90% of the content is built for browsers, not machines. That clutter distracts the agent and makes processing expensive, because the agent burns tokens on material it does not need. The API route strips the page down to what an agent can use. The move positions People Inc. to control how its journalism is delivered to agentic systems instead of leaving extraction to crawlers.
Our Take
This is a distribution play disguised as a technical fix. Roberts frames it as an efficiency problem, but the strategic prize is control. An API is a gate. Crawled HTML gives agents free, uncontrolled access; a custom feed lets the publisher decide what agents see, in what form, and on what commercial terms. The token-cost argument is also the sales pitch: agents that pay per token will prefer clean sources, which makes publisher-supplied structure a ranking advantage in agent-mediated discovery. Expect this to become the default posture for premium publishers within a year. The ones still relying on robots.txt and litigation will be negotiating from weakness.
What Changed
Publishers can now offer agent-native content delivery: structured, token-efficient feeds served directly to AI systems through a controlled interface. This replaces passive page crawling with an owned access channel where the publisher defines format, scope and, eventually, price.
Marketing Impact
Content strategy, SEO/GEO teams and publisher-side partnerships feel this first. Brand content and sponsored placements now need to survive an agent-readable format, and teams must plan for audiences that consume articles as structured data with no ad slots, no page layout and no pixel.
Competitive Implication
Large publishers with engineering depth and licensing leverage gain a controlled revenue and visibility channel with agents. Publishers dependent on crawl-based referral and ad-supported page views are vulnerable: agents will favour cheap-to-parse sources, and sites that serve only cluttered HTML get read less, cited less and paid nothing.
Strategic Outlook
Agent-facing APIs will spread across premium publishers through 2027, followed by pricing tiers, usage-based licensing and standardised formats. Agent developers will gravitate to sources that cut token cost, creating a de facto quality filter. Publishers without an agent channel will lose citation share in AI-mediated answers.
The Exploit
Action Item
Publishers' heads of audience and product should scope a minimal agent-readable endpoint for their top 50 evergreen articles during Q1 2027, and test it with one agent platform partner to establish licensing terms before competitors set the benchmark.