News API for RAG & AI Agents
Inject real-time, structured news into your retrieval pipelines and agent tools. Clean JSON, 26M+ articles, and precise filtering so your system gets the right context every time.
What makes Currents ideal for RAG
Clean Text & Metadata
Each article returns title, description, and full text when available — plus source, language, category, and publish date. Perfect for chunking and embedding.
Precise Filtering
Search by keyword, date range, language, country, or source domain. Retrieve only relevant context and cut noise from your vector store.
Real-Time + Archive
Build retrieval over 26M+ historical articles and keep it current with 90K+ new articles daily. One API for both backfill and live ingestion.
Agent & RAG patterns that work
Retrieval-Augmented Generation
Query Currents for articles matching the user question, chunk the text, embed it, and inject top-k chunks into the LLM prompt for grounded answers.
Agent Tool Calls
Expose a search_news(keywords, start_date, end_date) tool to your agent. JSON responses are trivial to parse into structured observation strings.
Auto-Update Knowledge Bases
Run a scheduled job against /latest-news with category or keyword filters. Append new articles to your vector DB so answers stay current.
Source-Aware Reasoning
Use the domain and published_at fields to cite sources in agent outputs. Ground claims with specific publications and timestamps.
Build smarter agents with news data
Free tier, no credit card, instant API key.