Structured, not raw
Every story tagged by region, category and domain, with the people and companies it mentions.
News Intelligence API
One engine for the world's news. Top sources from every continent, structured into events and entities, de-duplicated, scored, and served to developers and AI agents.
Live
Not a mock. This runs against the production engine, server-side, and returns the same JSON a client gets.
{
"query": "Nvidia",
"results": [
{
"title": "Nvidia: Trump will immer noch Huang als Nvidia-Chef zurück",
"source": "www.handelsblatt.com",
"category": "tech",
"region": "europe",
"published_at": "2026-11-20T21:30:36Z"
},
{
"title": "Saída de Jensen Huang da Nvidia pode frear a onda da IA?",
"source": "g1.globo.com",
"category": "tech",
"region": "south_america",
"published_at": "2026-11-21T06:00:00Z"
},
{
"title": "Ações da Nvidia têm pior sequência de quedas na bolsa desde 2008",
"source": "g1.globo.com",
"category": "ai",
"region": "south_america",
"published_at": "2026-11-20T18:47:51Z"
},
{
"title": "«Βουτιά» στις τεχνολογικές μετοχές με sell-off στη Wall Street: O Nasdaq χάνει όλα τα κέρδη του...",
"source": "www.iefimerida.gr",
"category": "ai",
"region": "europe",
"published_at": "2026-11-20T17:38:14Z"
}
]
}
Pipeline
Deterministic stages, no hallucination. Every article is normalized, de-duplicated, enriched and clustered into events.
Capabilities
Every story tagged by region, category and domain, with the people and companies it mentions.
The same story from many outlets becomes one event, with every source attached.
A knowledge graph of people, organisations and places, with stable identifiers.
Query by keyword, filter by region, category, domain, language or date.
Authority, freshness and cross-source confirmation produce one relevance score.
A REST API today, with an MCP server on the roadmap, so developers and AI agents can query the world in one call.
API
| GET | /v1/stats corpus totals and coverage |
|---|---|
| GET | /v1/articles filter by category, domain, region, lang, country, since, ids |
| GET | /v1/articles/{id} article + sources + entities + events |
| GET | /v1/events cross-source events |
| GET | /v1/events/{id} event + articles + entities + timeline |
| GET | /v1/entities search the entity graph |
| GET | /v1/entities/{id} entity + articles + events + related |
| GET | /v1/sources the source registry |
| GET | /v1/search full-text + fuzzy search |
curl "https://api.kontextnews.online/v1/search?q=Nvidia&limit=5" \ -H "X-API-Key: kn_..."
import requests
r = requests.get(
"https://api.kontextnews.online/v1/search",
params={"q": "Nvidia", "limit": 5},
headers={"X-API-Key": "kn_..."},
)
for a in r.json():
print(a["title"], a["url"])
const res = await fetch(
"https://api.kontextnews.online/v1/search?q=Nvidia&limit=5",
{ headers: { "X-API-Key": "kn_..." } },
);
type Article = { id: string; title: string; url: string };
const articles: Article[] = await res.json();
console.log(articles.map((a) => a.title));
const res = await fetch(
"https://api.kontextnews.online/v1/search?q=Nvidia&limit=5",
{ headers: { "X-API-Key": "kn_..." } },
);
const articles = await res.json();
console.log(articles.map((a) => a.title));
Per-key rate limits. JSON responses. Keys shown once, stored hashed.
Coverage
Top outlets and wires from around the world, normalized into a single structure and updated continuously.
Who it is for
One REST API instead of scraping dozens of feeds.
Saved queries, events and timelines, delivered where you work.
Our publications consume the same engine through an API key.
An agent-native surface: search news, events, entities and timelines.
FAQ
An API that turns the world's news into structured, queryable data: articles, cross-source events, entities and relations, from top outlets on every continent.
265 active feeds and wires across Europe, North America, Latin America, Africa, Asia, the Middle East and Oceania, each scored for authority. Dead and empty feeds are removed automatically.
Feeds are polled continuously. Search is ranked with a 30-day recency half-life, so fresh reporting outranks older archive that merely matches the query.
Aggregation. Every item links back to its source. We add the structure on top: classification, de-duplication, events and entities.
Hybrid search: PostgreSQL full-text plus trigram similarity, blended with a freshness weight. Filter by category, domain, region, language and date. See Search & ranking.
Not yet. The engine is exposed as a REST API today; an MCP server is on the roadmap.
Per key and per plan, from 10 requests per minute on Free up to a custom limit on Enterprise. Exceeding it returns 429 with a Retry-After header. See Rate limits.
Redistribution is covered by the Enterprise plan. Lower tiers are for internal use and products that link back to the source.