ToolsThe story, in brief

Making agent-friendly pages with content negotiation

Vercel just made every page on the web faster for AI agents. Here's why that matters for your infrastructure.

Illustration of independent geometric mechanisms passing paper tasks along branching amber tracks.
AI agents and the coordination of work.AI illustration by KeyNews
The KeyNews take

Why it matters

Vercel is shipping content negotiation tooling that lets AI agents fetch pages as markdown instead of HTML, reducing payload sizes by 99% and freeing up context windows. This is infrastructure-level optimization for the agent economy—making it cheaper and faster for thousands of AI products to consume web content at scale.

The key facts

6 to know
  1. HTML payload reduction: 500KB → 3KB (99.37% decrease)

  2. Content negotiation via Accept: text/markdown header standard

  3. Markdown sitemaps enable agent discovery without site-specific knowledge

  4. Implementation via Next.js 16 remote cache and route handlers

  5. Deployed across Vercel blog, changelog, and documentation

  6. Works for any agent sending Accept header—no site-specific configuration needed

Go to the source

Vercel Blogvercel.com

Publisher excerpt: Agents fetch web pages to answer questions, write code, and complete tasks. When an agent requests a page, it gets everything your browser gets, including navigation menus, stylesheets, JavaScript bundles, tracking scripts, and footer links, when all it needs is the structured text on the page.…
Read original report
Back to today's editionMore tools news

The wider picture

View all
Paper-cut illustration of a coral software window opening into a three-dimensional drafting space.
AI illustration by KeyNews
Tools01

Meta’s Muse is creepy, but maybe not for the reasons you think

A widely-deployed AI assistant shipped with a privacy and transparency gap that raises questions about how these tools actually work and what access they grant themselves. Practitioners need to understand what they're actually allowing when they install agent-adjacent products.

The Verge AI
Illustration of a transparent lens revealing connected networks across layers of paper.
AI illustration by KeyNews
Tools02

TypeSafe AI Releases Jev: A System One Model That Returns Typed, Calibrated Decisions Instead of Text

A new class of models optimized for typed, probabilistic decisions rather than text generation—reshaping cost structure and reliability for classification, scoring, and decision-automation workflows. Changes how teams think about model selection for deterministic tasks.

MarkTechPost
Illustration of independent geometric mechanisms passing paper tasks along branching amber tracks.
AI illustration by KeyNews
Tools03

Unity launches official plugins for Claude Code and OpenAI Codex to stop AI agents from using outdated tutorials

Unity is shipping official integrations that let frontier models access up-to-date documentation in real time, solving a real problem for developers using AI coding assistants: stale training data and outdated API references.

The Decoder