Feature
AI News
Monitoring
Real-time collection and classification from global sources. Bloom.ai ingests, normalizes, and correlates data across multiple feeds to surface the events that matter.
Continuously monitoring,
always classifying
Ingest
Specialized processors pull data from RSS feeds, stock markets, crypto exchanges, and geopolitical databases in real time.
Normalize
Every data point is transformed into a common raw_event schema with consistent fields regardless of source type.
Classify
AI agents assign categories, severity scores, regions, and extract keywords and entities from each event.
Correlate
Related events from different sources are clustered into unified events using 768-dim embedding similarity.
Sources
Four sources, one pipeline
Each source has a dedicated processor that handles its unique data format and normalizes it into the shared event schema.
News / RSS
Global news articles from curated RSS feeds — The Guardian, Reuters, Al Jazeera, and more. Each article is fetched, cleaned, and normalized into a structured event.
Stock Markets
Live market data from Yahoo Finance covering major indices, equities, and commodities. Price movements are tracked and correlated with geopolitical events.
Crypto
Cryptocurrency prices and market caps via CoinGecko. Sudden moves in crypto markets are linked to regulatory news and geopolitical developments.
Geopolitical
Conflict and crisis data from ACLED and the Uppsala Conflict Data Program. Armed conflicts, protests, and political violence are tracked with precise geolocation.
AI enrichment
Every event is enriched
before you see it
Once ingested and normalized, each raw event passes through an AI classification pipeline. The LLM agent analyzes content and produces structured metadata that powers filtering, search, and correlation.
- Category assignment from 17 predefined topics
- Severity scoring on a normalized scale
- Geographic region detection
- Keyword and named entity extraction
- 768-dimensional embedding via Google Gemini
768
Dimensions
17
Categories
4
Sources
Cross-source intelligence
Connecting the dots
across sources
A sanctions article, a stock market sell-off, a crypto volume spike, and protest data from the ground — these are four independent data points. Bloom.ai's correlation engine recognizes they describe the same geopolitical event and merges them into a single, comprehensive view.
- Cosine similarity on 768-dim embedding vectors
- Title and description synthesis from all sources
- Combined severity scoring across data points
- Full source traceability back to original articles
Capabilities
Everything that powers the news monitoring engine.
Real-time pipeline
The 3-layer pipeline runs continuously — ingesting, classifying, correlating, and surfacing events as they happen with no manual triggers.
Deduplication
Unique index on source type and source ID prevents duplicate ingestion at the database level, keeping the event corpus clean.
Semantic search
768-dimensional embeddings via pgvector enable semantic search across the entire event corpus — find events by meaning, not just keywords.
Multi-dimensional filtering
Filter events by category, severity, region, time range, and source type. Every piece of AI-generated metadata is queryable.
Source linking
Every unified event maintains links to all contributing raw events and their original source articles for full traceability.
Temporal awareness
Events are timestamped at every stage — ingestion, classification, and correlation — enabling precise temporal queries and trend detection.
Every signal, captured
From breaking news to market shifts — Bloom.ai's monitoring engine ensures nothing slips through.