For agents

What this is

Every story on this site is a source headline that an automated pipeline judged to be genuinely about AI’s impact on a real-world sector, placed in exactly one of 18 sectors, with the outlet, the time it was first reported and a direct link to the publisher. The API and the MCP server give agents that record as data: search it by sector, country, outlet, development type, headline keyword and date, fetch one story with every outlet that covers it, or poll for what is new since a timestamp.

What it is not: we hold no article text, no summaries and no bylines, and the responses carry none. There is no company, ticker, agency or amount field. Keyword search is headline full-text, not semantic. The record starts on 24 June 2026. Country is set on about three in five stories and development type on fewer than one in ten, so treat both as hints. Two sectors, AI Industry and AI Think Pieces, are about AI itself rather than AI inside a traditional sector; filter by sector to leave them out.

The API

Plain JSON over HTTPS, no library needed. The self-description at https://everythingmeetsai.com/api/v1 lists every endpoint, parameter and limit.

curl "https://everythingmeetsai.com/api/v1/stories?sector=healthcare_life_sciences&country=US&q=hospital&from=2026-09-01&limit=10"

• /api/v1/stories: search. sector, country, outlet, type (all repeatable), q, from, to, paywalled, sort, limit, offset.
• /api/v1/stories/<id>: one story with every outlet.
• /api/v1/changes?since=<ISO time>: what we published since then, with next_since for the next poll.
• /api/v1/sectors, /api/v1/outlets, /api/v1/types: the vocabularies, with counts.

Each story comes back as headline, publisher link, outlet, sector, time first reported, our publication time, countries, development type and a paywall flag. The link is the same one the page shows: the strongest free source where one exists.

MCP

The same data as seven tools for any MCP client, over Streamable HTTP at https://everythingmeetsai.com/mcp. With a key, use https://everythingmeetsai.com/mcp/k/<your key> instead; that is how Claude and ChatGPT connectors, which cannot send a header, carry it.

• Claude (web, desktop, Cowork): Settings, Connectors, “Add custom connector”, paste the URL. No OAuth fields needed.
• ChatGPT: Settings, Connectors, Advanced, turn on developer mode, then Create a connector with the URL and authentication “none”.
• Cursor: in ~/.cursor/mcp.json, {"mcpServers": {"everythingmeetsai": {"url": "https://everythingmeetsai.com/mcp"}}}.
• Claude Code: claude mcp add --transport http everythingmeetsai https://everythingmeetsai.com/mcp.
• stdio-only clients: npx -y mcp-remote https://everythingmeetsai.com/mcp.

The tools are search_stories, get_story, changes_since, list_sectors, list_outlets, list_types and about. Start with list_sectors for the slugs.

Limits

The service is free. Anonymous calls are allowed; a free key raises the allowance and lets us see which questions people actually ask. Limits are fixed windows per caller.

A 429 carries Retry-After. Send a key as Authorization: Bearer or ?key=. Responses are cached at the edge for a minute, so a poller asking every thirty seconds costs little.

Get a free key

Terms, in plain words

• Free, read-only, offered as is. We may change the limits or the shape of the data; the self-description at /api/v1 is the current truth.
• Please attribute “EverythingMeetsAI” and link the original publisher when you reuse a story. The headline and link belong to the publisher.
• Do not use a self-declared bot name or several keys to get around the limits. We will revoke keys that do.
• Tell us what you build, and what you asked for and did not get: that is how we decide which sectors and sources to add. Contact.

Also machine-readable: /llms.txt and the RSS and JSON feeds.