Ask the world a question, not a search engine a keyword
Semantic retrieval across global news and events, ranked by freshness and relevance and returned with citations, so research agents synthesise from current, attributable fact instead of whatever the first page of results happened to contain.
GET YOUR API KEYWhat you get from the layer
Meaning, not keyword matching
Pre-embedded content means a query about central bank policy tightening surfaces the relevant coverage whether or not those exact words appear in it.
Freshness as a parameter
Ask for live, recent or historical explicitly, instead of hoping the ranking algorithm shares your definition of recent.
Citations by default
Every returned fact carries provenance, which is the difference between a research output someone can act on and one they have to re-verify from scratch.
Where teams put it to work
Automated research reports
Agents pull current, cited evidence across sources instead of one search at a time.
Competitive and market intelligence
Track entities, launches and announcements continuously rather than in periodic manual sweeps.
Trend and narrative analysis
Temporal detection shows how a topic developed, not just its current state.
Due diligence support
Surface news, filings and event history for an entity in a single query.
Media and content monitoring
Watch coverage across regions and languages without a scraper fleet.
What the endpoint gives you
- Semantic search
Across global news and events
- Freshness-ranked retrieval
Live, recent or historical
- Six-dimension filtering
Source, time, entity, domain and more
- Temporal trend detection
How a story developed over time
- Citations and provenance
On every returned fact
- Quorum validation
Consensus before serving
- Cross-agent memory
A long research task retains what it already found
Questions teams ask first
- How is this different from Perplexity's API?
- Perplexity answers questions. Incord returns ranked, cited context for your model to reason over, so you keep control of the synthesis, the prompt and the model.
- Do results include sources?
- Yes, provenance is attached to returned facts.
- Can we filter by date range or source?
- Yes, across six filtering dimensions including time and source.
- How far back does the index go?
- Depth varies by domain. Check with the team if your use case depends on a specific historical window.
- Can research agents run autonomously over long periods?
- Yes. Cloud Agent Sandboxes run agents in isolated environments, and Global Memory keeps context across the whole run.
- Is multilingual coverage included?
- Global ingestion covers multiple languages; confirm specific language coverage for your target regions with the team.