People Inc. builds custom APIs to feed AI agents clean journalism

People Inc. is building custom APIs that hand AI agents its journalism in Markdown, stripping out what its chief innovation officer calls the 90% of a webpage that exists for browsers, not machines.

"If you are an agent and you're reading an HTML page, 90% of the content is junk meant for a browser to read, not an agent," Jon Roberts told delegates at Digiday's Media Strategies London event. "It will really distract the agent, and it will be incredibly expensive to process all these tokens that it doesn't need."

What do the APIs actually deliver?

The custom interfaces send article text alongside metadata: who wrote it, when it was published or updated, and other "trust signals that matter." For AI systems, that matters because large language models (AI systems that generate text from processed data) pay per token — each chunk of text they ingest carries a processing cost.

The shift also hands People Inc. more control over how its journalism reaches AI partners. The publisher first opened its content to partners through web crawling, then through Amazon S3 buckets (cloud storage that allowed it to push updates and retractions), and now moves to purpose-built APIs because crawling is inefficient for AI systems.

"Every token that needs to be processed is expensive," Roberts said. "If you throw it the entire HTML header, that's abuse. Right, that's not reasonable."

Why the urgency now?

The timing is not accidental. People Inc. laid off about 6% of its workforce last week, underscoring the pressure on publishers to find durable revenue as AI reshapes how audiences discover journalism. HUMAN's 2026 State of AI Traffic report adds another pressure point: traffic from AI agents and agentic browsers grew 7,851% year over year.

Daily Mail's director of SEO and editorial commerce, Carly Steven, framed the challenge as immediate rather than coming. "We need to act quickly on it because it's not going to be something in the future — it's happening now," she said at the London event. "So that is a fairly urgent need that we all need to address and get our heads around."

Roberts said AI companies are beginning to recognize the value of knowing when publishers created or updated content. People Inc. builds those details into its own AI tools first, then applies what it learns to partner access. He was explicit about one boundary: the publisher is not focused on advertising directly to AI agents.

Who else is moving?

Other publishers are running parallel experiments:

• The Economist is making marketing and B2B sales material outside its paywall more readable for agents.
• Time has converted its pages from HTML to Markdown and started selling ads to agents visiting its site.
• Forbes has created Markdown versions of its Vetted product reviews and an "MCP layer" (a standardized protocol connecting AI tools to external content), while exploring separate presentations for bots and human readers.

Forbes chief innovation officer Nina Gould said the approach could help the publisher reach people using AI tools — but she was careful about what remains unfinished. A planned ChatGPT product-recommendation plugin has not yet gone live. "On something like ads for agents, we're still working on that," she said.

Notably, no publisher cited here has yet quantified what structured AI access earns them; the builds precede the business model. The industry is standardizing the plumbing now and deferring the revenue question — a bet that being machine-readable early will matter more than monetizing it immediately.