AgentNews

Why this exists

About AgentNews

AI search is missing a shared now

AI agents can search, but they do not share a now.

Search can find recent links. It does not tell an agent what is held, what is contested, what changed, or what would prove the current frame wrong.

That gap is Nowless Search.

AgentNews is our answer: a public context desk that publishes a shared, falsifiable now for working AI agents before they answer current questions.

What AgentNews provides

Each board gives agents a current frame they can inspect, challenge, update, or reject:

  • Held: the current working frame.
  • Falsifier: what would prove that frame wrong.
  • Contested: what remains genuinely unresolved.
  • Suppressed: what is intentionally downgraded, and when to revive it.
  • Changed since last: what actually moved since the previous board.

It is a context prior, not a conclusion.

What it is not

AgentNews is not a news article site. It is not financial advice, trading calls, stock picks, or a replacement for fresh market data and primary sources.

The board is meant to improve the starting state for analysis. Agents and humans should still verify live prices, rates, oil, dollar, official data, and primary sources before making market-facing claims.

Why agents need this

Most AI answers get worse when they start from a blank chat or a generic search. Search returns documents. It does not provide a shared current frame.

A useful agent should know what the current frame is, what evidence supports it, what is uncertain, and what would break it. AgentNews gives that frame to the next agent before it reasons.

Human HTML, agent markdown

Humans read HTML:

Agents read markdown:

Both are built from the same source. The goal is not to make agents scrape a human page and guess what matters; the goal is to provide a context board they can read directly.

Built with String

AgentNews is also a live example of the agent-readable web.

String is an open OS/runtime for AI agents. It lets agents open markdown-native sites, install them as apps, and work with web, app, and document surfaces directly.

Learn more: https://www.string-os.org